research article blind source separation model of earth ...research article blind source separation...
TRANSCRIPT
![Page 1: Research Article Blind Source Separation Model of Earth ...Research Article Blind Source Separation Model of Earth-Rock Junctions in Dike Engineering Based on Distributed Optical Fiber](https://reader033.vdocument.in/reader033/viewer/2022060910/60a48674852e25584376221c/html5/thumbnails/1.jpg)
Research ArticleBlind Source Separation Model of Earth-Rock Junctionsin Dike Engineering Based on Distributed Optical FiberSensing Technology
Huaizhi Su12 Meng Yang12 Kunpeng Zhao3 and Zhiping Wen4
1State Key Laboratory of Hydrology-Water Resources and Hydraulic Engineering Hohai University Nanjing 210098 China2College of Water Conservancy and Hydropower Engineering Hohai University Nanjing 210098 China3National Engineering Research Center of Water Resources Efficient Utilization and Engineering Safety Nanjing 210098 China4Department of Computer Engineering Nanjing Institute of Technology Nanjing 211167 China
Correspondence should be addressed to Huaizhi Su su huaizhihhueducn
Received 17 October 2014 Accepted 2 March 2015
Academic Editor Zhenhua Zhu
Copyright copy 2015 Huaizhi Su et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited
Distributed temperature sensing (DTS) provides an important technology support for the earth-rock junctions of dike projects(ERJD) which are binding sites between culvert gates and pipes and dike body and dike foundation In this study a blind sourceseparation model is used for the identification of leakages based on the temperature data of DTS in leakage monitoring of ERJDFirst a denoising method is established based on the temperature monitoring data of distributed optical fiber in ERJD by a waveletpacket signal decomposition technique The temperature monitoring messages of fibers are combined response for leakages andother factors Its character of unclear responding mechanism is very obvious Thus a blind source separation technology is finallyselected Then the rule of temperature measurement data for optical fiber is analyzed and its temporal and spatial change processis also discussed The realization method of the blind source separation model is explored by combining independent componentanalysis (ICA) with principal component analysis (PCA) The practical test result in an example shows that the method couldefficiently locate and identify the leakage location of ERJD This paper is expected to be useful for further scientific research andefficient applications of distributed optical fiber sensing technology
1 Introduction
China is a country affected bymonsoons significantly Floodsand other natural disasters are often caused for rainy seasonA large number of dikes are built in order to reduce the losscaused by flooding and invasion of storm tide However theintegrity and safety of dike projects are objectively under-mined by crossing culverts and other hydraulic structuresespecially earth-rock junctions [1] The operating experienceof hydraulic structures of dikes indicated that strengthen-ing real-time location and identification of leakage hazardsfor the earth-rock junctions of dike projects (ERJD) hadspecial significances to ensure the safety of entire culvertsand dike projects [2] Research and analysis of high-techleakage monitoring in ERJD have gotten more and moreattention in engineering and academia during improvement
and optimization of traditional monitoring techniques andmethods [3] The temperature tracing technology is a geo-physical technique based on geophysics and it has broadapplication prospects in seepage monitoring [4] Comparedto the traditional leakage monitoring technology such aspiezometer tubes and osmometers the early temperaturetracing technology was more sensitive and effective and hadlower unit acquisition cost However the early temperaturetracing method was still a point-monitoring technology Thesingular area of internal temperature in buildings would notbe detected inevitably because of a large layout spacing ofthermistor thermometers and other reasons [5]
DTS as a new approach was produced by combination ofoptical fiber sensing technology and leakage risk monitoringtechnique [6 7] The new leakage monitoring method DTSprovides a possible way to the distributed long-distance and
Hindawi Publishing CorporationJournal of SensorsVolume 2015 Article ID 281538 6 pageshttpdxdoiorg1011552015281538
2 Journal of Sensors
large-scalemonitoring [8 9] In general the use of distributedfiber sensing technology in leakagemonitoring of buildings isstill in its infancy stage for theoretical research experimentalsimulation and engineering application research especiallyfor monitoring the combination between two mediums inERJD The consequent analysis of temperature monitoringdata and processing aspects still needmore studiesThereforean experimental model of ERJD was established and the dis-tributed fiber leakage monitoring technology was developedaccording to the working features of ERJD In-depth studiesof models and methods for leakage identification of ERJDhave important science and engineering applications basedon features of monitoring data
2 Denoising Research of DTS MonitoringData with Wavelet Packet
Wavelet transform method was proposed by Morlet in theearly 1980s and then it changed rapidly to be a very powerfultool for digital signal processing image processing datacompression data denoising and so forth
Wavelet is a breakthrough to Fourier transform analysisIt is a time-frequency analysismethodwith fixedwindow sizeand changeable shape time window and frequency windowLocal features signal can be well characterized in both timedomain and the frequency domain
Its main advantage is the ability to do local refinementand analysis for signal which overcomes the difficulties of notprocessing the nonstationary signals for Fourier transformtechnique and short-time Fourier transform method
During practical applications wavelet transform will dis-perse the stretching factor 119886 and translation factor 120591 in orderto adapt to the computer calculation 119886 takes 1198860
0 1198861
0 1198862
0 119886
119895
0
The sampling interval of displacement amount 120591 isΔ120591 = 11988611989501205910
Thus wavelet function is
120601119886120591(119905) = 119886
minus1198952
0120601 (119886minus119895
0(119905 minus 119896119886
119895
01205910)) (1)
where 119895 and 119896 are positive integersTaking 119886
0= 2 and 120591
0= 1 in practical operation wavelet
basis function 120601119886120591(119905) can be expressed as follows
120601119895119896 (119905) = 2
minus1198952120601 (2minus119895119905 minus 119896) (2)
Corresponding discrete wavelet transform is
WT119909(119895 119896) = ⟨119909 (119905) 120601
119895119896(119905)⟩ = int
119877
119909 (119905) 120601119895119896(119905)119889119905 (3)
The signal can be decomposed into a high frequencyportion named 119863 and a low frequency part marked 119860by wavelet transformation Compared to the wavelet trans-form wavelet packet decomposition analysis is a moresophisticated approach which well disassembles the highand low frequency parts of each layer simultaneously Itgreatly improves the resolution of signal and overcomes theinsufficient of wavelet decomposition in ldquolow-resolution ofhigh-frequencyrdquoThewavelet packet decomposition hasmorebroad application prospects and the 3-layer wavelet packetdecomposition process was expressed in Figure 1
S
A1 D1
AA2 AD2 DA2 DD2
AAA3 AAD3 ADA3 ADD3 DAA3 DAD3 DDA3 DDD3
Figure 1 Transform decomposition process of wavelet packet
As shown in Figure 1 the decomposition relationship of3-layer wavelet packet is as follows
119878 = 1198601198601198603+ 119860119860119863
3+ 119860119863119860
3+ 119860119863119863
3
+ 1198631198601198603+ 119863119860119863
3+ 119863119863119860
3+ 119863119863119863
3
(4)
The denoising procedure of wavelet packet for DTSleakage measurement data is summarized as below
(1) Wavelet packet decomposition Select wavelets anddetermine the decomposition level 119873 then decom-pose 119873-layer wavelet packet for original signal 119878During data processing the wavelet packet decompo-sition level is determined by the wavelet packet noisereduction effect
(2) Thresholds quantization Select the appropriatethreshold quantization method to quantify treatmentfor each wavelet packet coefficients decomposed
(3) Reconstruction of wavelet packet Reconstruct thelow-frequency coefficients of wavelet packet 119873-layerand its corresponding high-frequency coefficients
Among the previous steps the keys are selection of thewavelet packet decomposition level and approach to thethreshold quantization Wavelet packet decomposition levelis generally notmore than 5-layerThe threshold quantizationcan be processed by denoising with the soft threshold andhard threshold method
Coefficient of wavelet packet of hard threshold estimatedis
119908119895119896=
119908119895119896
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816ge 120582
0
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816lt 120582
(5)
Coefficient of wavelet packet built for soft threshold isgiven
119908119895119896=
sign (119908119895119896) (
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816minus 120582)
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816ge 120582
0
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816lt 120582
(6)
For (5) and (6) 119908119895119896
represents wavelet packet coefficientand 120582 is preset threshold or threshold value which iscalculated in the form
120582 =
120590radic2 lg119873lg (119895 + 1)
(7)
Journal of Sensors 3
where 120590 is noise variance 119895 is decomposed scale and119873 is sig-nal length
Signal to noise ratio (SNR) and root mean square error(RMSE) are usually selected as a signal denoising perfor-mance evaluation which are calculated by following equa-tions
SNR = 10 lg(sum119873
119899=11199042(119899)
sum119873
119899=1[119904 (119899) minus 119904 (119899)]
2)
RMSE = radic 1119873
[119904(119899) minus 119904(119899)]2
(8)
where 119904(119899) is original signal 119904(119899) represents signal afterdenoising by wavelet packet and119873 is signal length
3 Blind Source Separation Modelsand Methods of DTS TemperatureMonitoring Data
Seepage will lead to a change in the stability temperature fieldin themedia and a partial volatile region can be generated [1011] However the temperature field distribution of structurebody is affected not only by leakage field but also soil char-acteristics natural phenomena optical fiber laying elevationcross artificial cavities and other factors Further the mea-sured value of the optical fiber temperature sensing system isaffected Therefore the temperature field of structure body isa result caused bymultiple-factor under the combined effectsThe positioning of leakage points inside ERJD under nationalcondition can be attributed to the blind source separationproblem for the influence of multiple-factor [12]
Leakages soil characteristics or other factors causingchanges in temperature monitoring data unit have a cor-responding source component which is a function of thedistance between the sampling point and the starting pointSoil characteristics natural phenomena and other factorsshould be possibly separated out of the temperature dataexcept for the leakage factorThe separated data can be a resultof temperature field change only affected by the leakage factoras much as possible And the leakage factor is just the finalconcern
The effects of leakage factor soil characteristics or otherfactors on the measured temperature data can also be con-sidered to be independent units The temperature data inmost regions along fiber are not affected by leakages guttersand other factors which are small probability events on timeand space Based on non-Gaussian distribution and statisticalindependent of the source signal ICA is selected in blindsource separation technique In addition PCA is determinedas a pretreatment method for ICA technique in order toreduce the computational complexity and effective multiple-variable correlation
The key of PCA is the certain degree of relevance andredundancy of information between the variables At thesame time adjacent sampling points have similar work envi-ronment and strong correlation which provide a possibilityto the use of PCA The mathematical model of PCA can beobtained by the following process First a 119901 dimensional
random vector is consisted by 119901 scalars and it is expressed as119883 = (119883
1 119883
119901)1015840 119901 is defined as an arbitrary number 119883 is
used for orthogonal transformation ordering 119884 = 1198791015840119883 and119879 is orthogonal matrix Each component of 119884 is not relevantand the variance of the first variable of 119884 is the largest thesecond variable coming afterThis process can be representedby the following matrix
1198841= 119905111198831+ 119905121198832+ sdot sdot sdot + 119905
1119901119883119901= 1198791015840
1119883
1198842= 119905211198831+ 119905221198832+ sdot sdot sdot + 119905
2119901119883119901= 1198791015840
2119883
119884119901= 11990511990111198831+ 11990511990121198832+ sdot sdot sdot + 119905
119901119901119883119901= 1198791015840
119901119883
(9)
The definition of singular value decomposition (SVD) ofPCA can be described 119872 represents a matrix with 119898 rowsand 119899 columns whose elements belong to the field of realnumbers or complex field There is the following decom-position
119872 = 119880119878119881119879 (10)
where 119880 represents a unitary matrix with 119898 times 119898 order andis called left singular matrix of119872 119881 is a unitary matrix with119899 times 119899 order and is named right singular matrix of119872 119881119879 istranspose ofmatrix119881 the same bellow 119878 represents a positivesemidefinite diagonal matrix with119898 times 119899 order and elementson the diagonal matrix are singular values of 119872 which areequal to the square root of eigenvalues of 119872119879 lowast 119872 It isorthogonal matrix as the elements of unitary matrix are realnumbers Relationship between the main component of 119884and monitoring data matrix119883 can be obtained by
119884 = 119883119879 (11)
Thus
119883 = 1198841198791015840 (12)
where matrix of 119879 is constituted by eigenvectors of diagonalmatrix and119883119879119883 is orthogonal matrix
From the theory of SVD it can be obtained that
119883 = 1198801198781198811015840 (13)
Comparing formula (12) with (13) it can be known that119879 = 119881 then 119884 = 119880119878 = 119883119881
Therefore the main component of matrix is the productof left singular matrix and its corresponding singular valuesand is also equal to the product of monitoring data matrixand the right singular matrix The number of principalcomponents can be determined through the cumulativevariance contribution rate Variance contribution rate andcumulative variance contribution rate are described follows
120593119896=
120582119896
sum119901
119896=1120582119896
120595119898=
sum119898
119896=1120582119896
sum119901
119896=1120582119896
(14)
4 Journal of Sensors
where 120582119896represents the eigenvalues of covariancematrix and
120593119896and 120595
119898are respectively called variance contribution rate
and cumulative variance contribution rateIn order to avoid inconsistencies between dimensions and
units of data the data is typically normalized
119883lowast
119894=
119883119894minus 119864 (119883
119894)
radic119863 (119883119894)
(15)
The mathematical model of ICA can be expressed by
1199091= 119886111199041+ 119886121199042+ sdot sdot sdot + 119886
1119899119904119899
1199092= 119886211199041+ 119886221199042+ sdot sdot sdot + 119886
2119899119904119899
119909119899= 11988611989911199041+ 11988611989921199042+ sdot sdot sdot + 119886
119899119899119904119899
(16)
where 1199091 1199092 119909
119899represent 119899 random variables named
mixed signals 1199041 1199042 119904
119899are 119899 independent components
called source signals 119886119894119895(119894 119895 = 1 2 119899) is mixing coeffi-
cientParallel algorithm calculation steps of several indepen-
dent components are described as follows
(1) standardize the data so that the mean and variance ofeach column are respectively 0 and 1
(2) data whitened of 119911 is gotten from the data obtainedfrom Step (1)
(3) 119898 is selected as the number of independent compo-nent to be estimated
(4) the mixing coefficient of 119908119894is initialized 119894 = 1
2 119898 and every119908119894is changed to be a unit 2-norm
then matrix 119882 is orthogonalised by the method ofStep (6)
(5) update each 119908119894 119908119894larr 119864119911119892(119908
119879
119894119911) minus 119864119892
1015840(119908119879
119894119911)119908119894
the function of 119892 can be selected from formula (17) toformula (19)
(6) matrix 119882 = (1199081 119908
119898)1015840 is orthogonalised 119882 larr
(1198821198821015840)minus12119882
(7) return to Step (5) if the result is not converged
The function of119892 can be selected from the following func-tions
1198921(119910) = tanh (119886
1119910) (17)
1198922(119910) = 119910 exp(minus
1199102
2
) (18)
1198923(119910) = 119910
3 (19)
Leakage location
Figure 2 Model experiment of ERJD
DTS system
Heating system
Figure 3 DTS monitoring and heating system
The corresponding derivative functions are described asfollows
1198921015840
1(119910) = 119886
1(1 minus tanh2 (119886
1119910))
1198921015840
2(119910) = (1 minus 119910
2) exp(minus
1199102
2
)
1198921015840
3(119910) = 3119910
2
(20)
where 1198861is a constant valuing in [1 2] and the value is usually
taken as 1 tanh(119909) represents hyperbolic tangent function
4 Application Case Analyses
A model experiment platform was formed and shown inFigure 2 to simulate the existence of leakages for ERJD TheDTS monitoring system of Sentinel DTS-LR produced bySensornet was applied The DTS monitoring and heatingsystem was shown in Figure 3 A schematic diagram of themodel was given in Figure 4 Multimode four core armoredcable of 50125 um was used as the monitoring optical cablewith a type of ZTT-GYXTW-4A1a
TheZTT-GYXTW-4A1a was a special optical cable whichcould be used to be heated And its structure section wasdrawn in Figure 5
Journal of Sensors 5
Monitoringopticalcable
Modelexperiment
pool of ERJD
Figure 4 Schematic diagram of the model
Water-blockingmaterial
Polyethylenejacket
Heatingwire
Fiberpaste
Opticalfibre
Loosetube
Steel-plasticcompound
Figure 5 Structure section of ZTT-GYXTW-4A1a
The leakage location was washed by a water pipe and thevelocity of seepage was controlled The monitoring data wasnoted as the stable velocity was formed
For simplicity just one leakage point was introduced inthe model and the sampling distance was 1 m and the sam-pling interval was 2 hours for 5 days Thus each measuringpoint obtained 60 groups of data First the original tempera-ture data should be denoised and a typical point was selectedas an example analyzed After calculation the decompositionlevels of the wavelet packet were 3-layer Figure 6 was shownto compare the raw signal with the wavelet denoising signal
The first principal component was calculated and listed inFigure 7
The corresponding variance contribution rate of the firstprincipal component 120590
1sum120590 was 8721 and it could be
concluded that the contribution of the first principal compo-nent to informationwas largerThe difference of the principalcomponent value between the leakage point presented and
Raw signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Wavelet denoising signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Figure 6 Comparative diagram between raw signal and the waveletdenoising signal
minus02
minus04
minus06
06
04
02
0
5 10 15 20 25 30 35
Distance (m)
Valu
e
Figure 7 Result diagram after PCA method
other points was much larger than the one of the first prin-cipal component Therefore the second and third principalcomponents were selected to the next ICA which meant119894 = 2 One of the two principal components was used toconstruct a separate component of the signal space and theother one was made to build a separate part of the salvagespace The difference of the separate component 1 in theleakage point was more evident than the one of other pointswhose distribution was more stable Thereby the separatecomponent 1 was considered to establish a salvage spaceThe separate component 2 constructed a separate componentsignal space The corresponding grey-scale map of residualspace was shown in Figure 8
From the result of data processing it can be known thatthe problem of the optical fiber temperature monitoring datacould be analyzed by the blind source separation and ICAwas used to process data In Figure 8 the leakage area of14m was larger than the area of 32m And the test distanceof the model was less than 32m Therefore the area of 32m
6 Journal of Sensors
minus05
0
05
1
minus1
minus15
Dist
ance
(m)
Time
Leakage point
0
5
10
15
20
25
30
6050403020100
Figure 8 Grey-scale map of residual space
was not considered Finally the leakage location could be wellimplemented
5 Conclusions
In this paper ERJD was the studied object and its leakageidentificationwas for the research emphasis Based on leakagecharacteristics of monitoring ERJD by DTS the affectingfactors of optical fiber temperaturemonitoring datawere ana-lyzed The wavelet packet denoising method and thresholddetermining method were also studied A blind source sepa-rationmodel of optical fiber temperature datawas finally builtand the corresponding temperature change process of theleakage source was extracted The implementation methodof blind source separation technique was discussed In-depth research on characteristics of optical fiber temperaturedata was carried out and optimal applicable conditions forrealization method of the blind source separation techniqueswere also analyzed The blind source separation techniqueimplementation combining ICA with PCA was finally pro-posed With leakage fiber-optic model experiment of ERJDthe reliability of optical fiber temperature measuring databased on the blind source separation method is validated
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research has been partially supported by Jiangsu NaturalScience Foundation (SN BK2012036) the National Natu-ral Science Foundation of China (SN 51179066 5113900141323001 and 51409167) the Specialized Research Fund forthe Doctoral Program of Higher Education of China (SN20130094110010) the Nonprofit Industry Financial Program
of MWR (SN 201301061 and 201201038) the Open Founda-tion of State Key Laboratory of Hydrology-Water Resourcesand Hydraulic Engineering (SN 20145027612) the ResearchProgram on Natural Science for Colleges and Universities inJiangsu Province (SN 14KJB520016) the Priority AcademicProgram Development of Jiangsu Higher Education Institu-tions (SN YS11001) and the Scientific Innovation ResearchScheme for Jiangsu University Graduate (SN 2013B25514)
References
[1] H-H Zhu A N L Ho J-H Yin H W Sun H-F Pei andC-Y Hong ldquoAn optical fibre monitoring system for evaluatingthe performance of a soil nailed sloperdquo Smart Structures andSystems vol 9 no 5 pp 393ndash410 2012
[2] B Shi C-S Tang L Gao C Liu and B-J Wang ldquoObservationand analysis of the urban heat island effect on soil in NanjingChinardquo Environmental Earth Sciences vol 67 no 1 pp 215ndash2292012
[3] J F Yan B Shi D F Cao H H Zhu and G Q Wei ldquoExper-iment study on carbon coated heating optical fiber for sandleakage monitoring based on distributed temperature systemrdquoin Proceedings of the 6th International Conference on StructuralHealth Monitoring of Intelligent Infrastructure Paper no MS03-06 Hong Kong 2013
[4] S Johansson ldquoLocalization and quantification of water leakagein ageing embankment dams by regular temperature measure-mentsrdquo inProceedings of the 17th International Congress on LargeDams (ICOLD rsquo91) pp 991ndash1005 Vienna Austria 1991
[5] O Kappelmeyer ldquoThe use of near surface temperature mea-surements for discovering anomalies due to causes at depthsrdquoGeophysical Prospecting vol 5 no 3 pp 239ndash258 1957
[6] A H Hartog ldquoDistributed fiber-optic temperature sensorsprinciples and applicationsrdquo inOptical Fiber Sensor TechnologyK T Grattan and B T Meggitt Eds pp 241ndash301 KluwerAcademic Publishers New York NY USA 2000
[7] H Z Su J Y Li J Hu and Z Wen ldquoAnalysis and back-analysisfor temperature field of concrete arch dam during constructionperiod based on temperature data measured by DTSrdquo IEEESensors Journal vol 13 no 5 pp 1403ndash1412 2013
[8] S Yin ldquoDistributed fiber optic sensorsrdquo in Fiber Optic SensorsF T Yu and S Yin Eds pp 201ndash229 Marcel Dekker New YorkNY USA 2000
[9] A D Kersey ldquoOptical fiber sensors for downwell monitoringapplications in the oil and gas industryrdquo in Proceedings of the13th International Conference on Optical Fiber Sensors (Ofs-13rsquo99) pp 326ndash331 1999
[10] S Grosswig A Graupner and E Hurtig ldquoDistributed fiberoptical temperature sensing techniquemdasha variable tool formonitoring tasksrdquo in Proceedings of the 8th International Sym-posium on Temperature and Thermal Measurements in Industryand Science pp 9ndash17 2001
[11] S A Wade K T V Grattan B McKinley L F Boswell andC DrsquoMello ldquo Incorporation of fiber optic sensors in concretespecimens testing and evaluationrdquo IEEE Sensors Journal vol 4no 1 pp 127ndash134 2004
[12] H-H Zhu L-C Liu H-F Pei and B Shi ldquoSettlement analysisof viscoelastic foundation under vertical line load using afractional Kelvin-Voigt modelrdquo Geomechanics and Engineeringvol 4 no 1 pp 67ndash78 2012
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
![Page 2: Research Article Blind Source Separation Model of Earth ...Research Article Blind Source Separation Model of Earth-Rock Junctions in Dike Engineering Based on Distributed Optical Fiber](https://reader033.vdocument.in/reader033/viewer/2022060910/60a48674852e25584376221c/html5/thumbnails/2.jpg)
2 Journal of Sensors
large-scalemonitoring [8 9] In general the use of distributedfiber sensing technology in leakagemonitoring of buildings isstill in its infancy stage for theoretical research experimentalsimulation and engineering application research especiallyfor monitoring the combination between two mediums inERJD The consequent analysis of temperature monitoringdata and processing aspects still needmore studiesThereforean experimental model of ERJD was established and the dis-tributed fiber leakage monitoring technology was developedaccording to the working features of ERJD In-depth studiesof models and methods for leakage identification of ERJDhave important science and engineering applications basedon features of monitoring data
2 Denoising Research of DTS MonitoringData with Wavelet Packet
Wavelet transform method was proposed by Morlet in theearly 1980s and then it changed rapidly to be a very powerfultool for digital signal processing image processing datacompression data denoising and so forth
Wavelet is a breakthrough to Fourier transform analysisIt is a time-frequency analysismethodwith fixedwindow sizeand changeable shape time window and frequency windowLocal features signal can be well characterized in both timedomain and the frequency domain
Its main advantage is the ability to do local refinementand analysis for signal which overcomes the difficulties of notprocessing the nonstationary signals for Fourier transformtechnique and short-time Fourier transform method
During practical applications wavelet transform will dis-perse the stretching factor 119886 and translation factor 120591 in orderto adapt to the computer calculation 119886 takes 1198860
0 1198861
0 1198862
0 119886
119895
0
The sampling interval of displacement amount 120591 isΔ120591 = 11988611989501205910
Thus wavelet function is
120601119886120591(119905) = 119886
minus1198952
0120601 (119886minus119895
0(119905 minus 119896119886
119895
01205910)) (1)
where 119895 and 119896 are positive integersTaking 119886
0= 2 and 120591
0= 1 in practical operation wavelet
basis function 120601119886120591(119905) can be expressed as follows
120601119895119896 (119905) = 2
minus1198952120601 (2minus119895119905 minus 119896) (2)
Corresponding discrete wavelet transform is
WT119909(119895 119896) = ⟨119909 (119905) 120601
119895119896(119905)⟩ = int
119877
119909 (119905) 120601119895119896(119905)119889119905 (3)
The signal can be decomposed into a high frequencyportion named 119863 and a low frequency part marked 119860by wavelet transformation Compared to the wavelet trans-form wavelet packet decomposition analysis is a moresophisticated approach which well disassembles the highand low frequency parts of each layer simultaneously Itgreatly improves the resolution of signal and overcomes theinsufficient of wavelet decomposition in ldquolow-resolution ofhigh-frequencyrdquoThewavelet packet decomposition hasmorebroad application prospects and the 3-layer wavelet packetdecomposition process was expressed in Figure 1
S
A1 D1
AA2 AD2 DA2 DD2
AAA3 AAD3 ADA3 ADD3 DAA3 DAD3 DDA3 DDD3
Figure 1 Transform decomposition process of wavelet packet
As shown in Figure 1 the decomposition relationship of3-layer wavelet packet is as follows
119878 = 1198601198601198603+ 119860119860119863
3+ 119860119863119860
3+ 119860119863119863
3
+ 1198631198601198603+ 119863119860119863
3+ 119863119863119860
3+ 119863119863119863
3
(4)
The denoising procedure of wavelet packet for DTSleakage measurement data is summarized as below
(1) Wavelet packet decomposition Select wavelets anddetermine the decomposition level 119873 then decom-pose 119873-layer wavelet packet for original signal 119878During data processing the wavelet packet decompo-sition level is determined by the wavelet packet noisereduction effect
(2) Thresholds quantization Select the appropriatethreshold quantization method to quantify treatmentfor each wavelet packet coefficients decomposed
(3) Reconstruction of wavelet packet Reconstruct thelow-frequency coefficients of wavelet packet 119873-layerand its corresponding high-frequency coefficients
Among the previous steps the keys are selection of thewavelet packet decomposition level and approach to thethreshold quantization Wavelet packet decomposition levelis generally notmore than 5-layerThe threshold quantizationcan be processed by denoising with the soft threshold andhard threshold method
Coefficient of wavelet packet of hard threshold estimatedis
119908119895119896=
119908119895119896
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816ge 120582
0
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816lt 120582
(5)
Coefficient of wavelet packet built for soft threshold isgiven
119908119895119896=
sign (119908119895119896) (
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816minus 120582)
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816ge 120582
0
10038161003816100381610038161003816119908119895119896
10038161003816100381610038161003816lt 120582
(6)
For (5) and (6) 119908119895119896
represents wavelet packet coefficientand 120582 is preset threshold or threshold value which iscalculated in the form
120582 =
120590radic2 lg119873lg (119895 + 1)
(7)
Journal of Sensors 3
where 120590 is noise variance 119895 is decomposed scale and119873 is sig-nal length
Signal to noise ratio (SNR) and root mean square error(RMSE) are usually selected as a signal denoising perfor-mance evaluation which are calculated by following equa-tions
SNR = 10 lg(sum119873
119899=11199042(119899)
sum119873
119899=1[119904 (119899) minus 119904 (119899)]
2)
RMSE = radic 1119873
[119904(119899) minus 119904(119899)]2
(8)
where 119904(119899) is original signal 119904(119899) represents signal afterdenoising by wavelet packet and119873 is signal length
3 Blind Source Separation Modelsand Methods of DTS TemperatureMonitoring Data
Seepage will lead to a change in the stability temperature fieldin themedia and a partial volatile region can be generated [1011] However the temperature field distribution of structurebody is affected not only by leakage field but also soil char-acteristics natural phenomena optical fiber laying elevationcross artificial cavities and other factors Further the mea-sured value of the optical fiber temperature sensing system isaffected Therefore the temperature field of structure body isa result caused bymultiple-factor under the combined effectsThe positioning of leakage points inside ERJD under nationalcondition can be attributed to the blind source separationproblem for the influence of multiple-factor [12]
Leakages soil characteristics or other factors causingchanges in temperature monitoring data unit have a cor-responding source component which is a function of thedistance between the sampling point and the starting pointSoil characteristics natural phenomena and other factorsshould be possibly separated out of the temperature dataexcept for the leakage factorThe separated data can be a resultof temperature field change only affected by the leakage factoras much as possible And the leakage factor is just the finalconcern
The effects of leakage factor soil characteristics or otherfactors on the measured temperature data can also be con-sidered to be independent units The temperature data inmost regions along fiber are not affected by leakages guttersand other factors which are small probability events on timeand space Based on non-Gaussian distribution and statisticalindependent of the source signal ICA is selected in blindsource separation technique In addition PCA is determinedas a pretreatment method for ICA technique in order toreduce the computational complexity and effective multiple-variable correlation
The key of PCA is the certain degree of relevance andredundancy of information between the variables At thesame time adjacent sampling points have similar work envi-ronment and strong correlation which provide a possibilityto the use of PCA The mathematical model of PCA can beobtained by the following process First a 119901 dimensional
random vector is consisted by 119901 scalars and it is expressed as119883 = (119883
1 119883
119901)1015840 119901 is defined as an arbitrary number 119883 is
used for orthogonal transformation ordering 119884 = 1198791015840119883 and119879 is orthogonal matrix Each component of 119884 is not relevantand the variance of the first variable of 119884 is the largest thesecond variable coming afterThis process can be representedby the following matrix
1198841= 119905111198831+ 119905121198832+ sdot sdot sdot + 119905
1119901119883119901= 1198791015840
1119883
1198842= 119905211198831+ 119905221198832+ sdot sdot sdot + 119905
2119901119883119901= 1198791015840
2119883
119884119901= 11990511990111198831+ 11990511990121198832+ sdot sdot sdot + 119905
119901119901119883119901= 1198791015840
119901119883
(9)
The definition of singular value decomposition (SVD) ofPCA can be described 119872 represents a matrix with 119898 rowsand 119899 columns whose elements belong to the field of realnumbers or complex field There is the following decom-position
119872 = 119880119878119881119879 (10)
where 119880 represents a unitary matrix with 119898 times 119898 order andis called left singular matrix of119872 119881 is a unitary matrix with119899 times 119899 order and is named right singular matrix of119872 119881119879 istranspose ofmatrix119881 the same bellow 119878 represents a positivesemidefinite diagonal matrix with119898 times 119899 order and elementson the diagonal matrix are singular values of 119872 which areequal to the square root of eigenvalues of 119872119879 lowast 119872 It isorthogonal matrix as the elements of unitary matrix are realnumbers Relationship between the main component of 119884and monitoring data matrix119883 can be obtained by
119884 = 119883119879 (11)
Thus
119883 = 1198841198791015840 (12)
where matrix of 119879 is constituted by eigenvectors of diagonalmatrix and119883119879119883 is orthogonal matrix
From the theory of SVD it can be obtained that
119883 = 1198801198781198811015840 (13)
Comparing formula (12) with (13) it can be known that119879 = 119881 then 119884 = 119880119878 = 119883119881
Therefore the main component of matrix is the productof left singular matrix and its corresponding singular valuesand is also equal to the product of monitoring data matrixand the right singular matrix The number of principalcomponents can be determined through the cumulativevariance contribution rate Variance contribution rate andcumulative variance contribution rate are described follows
120593119896=
120582119896
sum119901
119896=1120582119896
120595119898=
sum119898
119896=1120582119896
sum119901
119896=1120582119896
(14)
4 Journal of Sensors
where 120582119896represents the eigenvalues of covariancematrix and
120593119896and 120595
119898are respectively called variance contribution rate
and cumulative variance contribution rateIn order to avoid inconsistencies between dimensions and
units of data the data is typically normalized
119883lowast
119894=
119883119894minus 119864 (119883
119894)
radic119863 (119883119894)
(15)
The mathematical model of ICA can be expressed by
1199091= 119886111199041+ 119886121199042+ sdot sdot sdot + 119886
1119899119904119899
1199092= 119886211199041+ 119886221199042+ sdot sdot sdot + 119886
2119899119904119899
119909119899= 11988611989911199041+ 11988611989921199042+ sdot sdot sdot + 119886
119899119899119904119899
(16)
where 1199091 1199092 119909
119899represent 119899 random variables named
mixed signals 1199041 1199042 119904
119899are 119899 independent components
called source signals 119886119894119895(119894 119895 = 1 2 119899) is mixing coeffi-
cientParallel algorithm calculation steps of several indepen-
dent components are described as follows
(1) standardize the data so that the mean and variance ofeach column are respectively 0 and 1
(2) data whitened of 119911 is gotten from the data obtainedfrom Step (1)
(3) 119898 is selected as the number of independent compo-nent to be estimated
(4) the mixing coefficient of 119908119894is initialized 119894 = 1
2 119898 and every119908119894is changed to be a unit 2-norm
then matrix 119882 is orthogonalised by the method ofStep (6)
(5) update each 119908119894 119908119894larr 119864119911119892(119908
119879
119894119911) minus 119864119892
1015840(119908119879
119894119911)119908119894
the function of 119892 can be selected from formula (17) toformula (19)
(6) matrix 119882 = (1199081 119908
119898)1015840 is orthogonalised 119882 larr
(1198821198821015840)minus12119882
(7) return to Step (5) if the result is not converged
The function of119892 can be selected from the following func-tions
1198921(119910) = tanh (119886
1119910) (17)
1198922(119910) = 119910 exp(minus
1199102
2
) (18)
1198923(119910) = 119910
3 (19)
Leakage location
Figure 2 Model experiment of ERJD
DTS system
Heating system
Figure 3 DTS monitoring and heating system
The corresponding derivative functions are described asfollows
1198921015840
1(119910) = 119886
1(1 minus tanh2 (119886
1119910))
1198921015840
2(119910) = (1 minus 119910
2) exp(minus
1199102
2
)
1198921015840
3(119910) = 3119910
2
(20)
where 1198861is a constant valuing in [1 2] and the value is usually
taken as 1 tanh(119909) represents hyperbolic tangent function
4 Application Case Analyses
A model experiment platform was formed and shown inFigure 2 to simulate the existence of leakages for ERJD TheDTS monitoring system of Sentinel DTS-LR produced bySensornet was applied The DTS monitoring and heatingsystem was shown in Figure 3 A schematic diagram of themodel was given in Figure 4 Multimode four core armoredcable of 50125 um was used as the monitoring optical cablewith a type of ZTT-GYXTW-4A1a
TheZTT-GYXTW-4A1a was a special optical cable whichcould be used to be heated And its structure section wasdrawn in Figure 5
Journal of Sensors 5
Monitoringopticalcable
Modelexperiment
pool of ERJD
Figure 4 Schematic diagram of the model
Water-blockingmaterial
Polyethylenejacket
Heatingwire
Fiberpaste
Opticalfibre
Loosetube
Steel-plasticcompound
Figure 5 Structure section of ZTT-GYXTW-4A1a
The leakage location was washed by a water pipe and thevelocity of seepage was controlled The monitoring data wasnoted as the stable velocity was formed
For simplicity just one leakage point was introduced inthe model and the sampling distance was 1 m and the sam-pling interval was 2 hours for 5 days Thus each measuringpoint obtained 60 groups of data First the original tempera-ture data should be denoised and a typical point was selectedas an example analyzed After calculation the decompositionlevels of the wavelet packet were 3-layer Figure 6 was shownto compare the raw signal with the wavelet denoising signal
The first principal component was calculated and listed inFigure 7
The corresponding variance contribution rate of the firstprincipal component 120590
1sum120590 was 8721 and it could be
concluded that the contribution of the first principal compo-nent to informationwas largerThe difference of the principalcomponent value between the leakage point presented and
Raw signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Wavelet denoising signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Figure 6 Comparative diagram between raw signal and the waveletdenoising signal
minus02
minus04
minus06
06
04
02
0
5 10 15 20 25 30 35
Distance (m)
Valu
e
Figure 7 Result diagram after PCA method
other points was much larger than the one of the first prin-cipal component Therefore the second and third principalcomponents were selected to the next ICA which meant119894 = 2 One of the two principal components was used toconstruct a separate component of the signal space and theother one was made to build a separate part of the salvagespace The difference of the separate component 1 in theleakage point was more evident than the one of other pointswhose distribution was more stable Thereby the separatecomponent 1 was considered to establish a salvage spaceThe separate component 2 constructed a separate componentsignal space The corresponding grey-scale map of residualspace was shown in Figure 8
From the result of data processing it can be known thatthe problem of the optical fiber temperature monitoring datacould be analyzed by the blind source separation and ICAwas used to process data In Figure 8 the leakage area of14m was larger than the area of 32m And the test distanceof the model was less than 32m Therefore the area of 32m
6 Journal of Sensors
minus05
0
05
1
minus1
minus15
Dist
ance
(m)
Time
Leakage point
0
5
10
15
20
25
30
6050403020100
Figure 8 Grey-scale map of residual space
was not considered Finally the leakage location could be wellimplemented
5 Conclusions
In this paper ERJD was the studied object and its leakageidentificationwas for the research emphasis Based on leakagecharacteristics of monitoring ERJD by DTS the affectingfactors of optical fiber temperaturemonitoring datawere ana-lyzed The wavelet packet denoising method and thresholddetermining method were also studied A blind source sepa-rationmodel of optical fiber temperature datawas finally builtand the corresponding temperature change process of theleakage source was extracted The implementation methodof blind source separation technique was discussed In-depth research on characteristics of optical fiber temperaturedata was carried out and optimal applicable conditions forrealization method of the blind source separation techniqueswere also analyzed The blind source separation techniqueimplementation combining ICA with PCA was finally pro-posed With leakage fiber-optic model experiment of ERJDthe reliability of optical fiber temperature measuring databased on the blind source separation method is validated
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research has been partially supported by Jiangsu NaturalScience Foundation (SN BK2012036) the National Natu-ral Science Foundation of China (SN 51179066 5113900141323001 and 51409167) the Specialized Research Fund forthe Doctoral Program of Higher Education of China (SN20130094110010) the Nonprofit Industry Financial Program
of MWR (SN 201301061 and 201201038) the Open Founda-tion of State Key Laboratory of Hydrology-Water Resourcesand Hydraulic Engineering (SN 20145027612) the ResearchProgram on Natural Science for Colleges and Universities inJiangsu Province (SN 14KJB520016) the Priority AcademicProgram Development of Jiangsu Higher Education Institu-tions (SN YS11001) and the Scientific Innovation ResearchScheme for Jiangsu University Graduate (SN 2013B25514)
References
[1] H-H Zhu A N L Ho J-H Yin H W Sun H-F Pei andC-Y Hong ldquoAn optical fibre monitoring system for evaluatingthe performance of a soil nailed sloperdquo Smart Structures andSystems vol 9 no 5 pp 393ndash410 2012
[2] B Shi C-S Tang L Gao C Liu and B-J Wang ldquoObservationand analysis of the urban heat island effect on soil in NanjingChinardquo Environmental Earth Sciences vol 67 no 1 pp 215ndash2292012
[3] J F Yan B Shi D F Cao H H Zhu and G Q Wei ldquoExper-iment study on carbon coated heating optical fiber for sandleakage monitoring based on distributed temperature systemrdquoin Proceedings of the 6th International Conference on StructuralHealth Monitoring of Intelligent Infrastructure Paper no MS03-06 Hong Kong 2013
[4] S Johansson ldquoLocalization and quantification of water leakagein ageing embankment dams by regular temperature measure-mentsrdquo inProceedings of the 17th International Congress on LargeDams (ICOLD rsquo91) pp 991ndash1005 Vienna Austria 1991
[5] O Kappelmeyer ldquoThe use of near surface temperature mea-surements for discovering anomalies due to causes at depthsrdquoGeophysical Prospecting vol 5 no 3 pp 239ndash258 1957
[6] A H Hartog ldquoDistributed fiber-optic temperature sensorsprinciples and applicationsrdquo inOptical Fiber Sensor TechnologyK T Grattan and B T Meggitt Eds pp 241ndash301 KluwerAcademic Publishers New York NY USA 2000
[7] H Z Su J Y Li J Hu and Z Wen ldquoAnalysis and back-analysisfor temperature field of concrete arch dam during constructionperiod based on temperature data measured by DTSrdquo IEEESensors Journal vol 13 no 5 pp 1403ndash1412 2013
[8] S Yin ldquoDistributed fiber optic sensorsrdquo in Fiber Optic SensorsF T Yu and S Yin Eds pp 201ndash229 Marcel Dekker New YorkNY USA 2000
[9] A D Kersey ldquoOptical fiber sensors for downwell monitoringapplications in the oil and gas industryrdquo in Proceedings of the13th International Conference on Optical Fiber Sensors (Ofs-13rsquo99) pp 326ndash331 1999
[10] S Grosswig A Graupner and E Hurtig ldquoDistributed fiberoptical temperature sensing techniquemdasha variable tool formonitoring tasksrdquo in Proceedings of the 8th International Sym-posium on Temperature and Thermal Measurements in Industryand Science pp 9ndash17 2001
[11] S A Wade K T V Grattan B McKinley L F Boswell andC DrsquoMello ldquo Incorporation of fiber optic sensors in concretespecimens testing and evaluationrdquo IEEE Sensors Journal vol 4no 1 pp 127ndash134 2004
[12] H-H Zhu L-C Liu H-F Pei and B Shi ldquoSettlement analysisof viscoelastic foundation under vertical line load using afractional Kelvin-Voigt modelrdquo Geomechanics and Engineeringvol 4 no 1 pp 67ndash78 2012
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
![Page 3: Research Article Blind Source Separation Model of Earth ...Research Article Blind Source Separation Model of Earth-Rock Junctions in Dike Engineering Based on Distributed Optical Fiber](https://reader033.vdocument.in/reader033/viewer/2022060910/60a48674852e25584376221c/html5/thumbnails/3.jpg)
Journal of Sensors 3
where 120590 is noise variance 119895 is decomposed scale and119873 is sig-nal length
Signal to noise ratio (SNR) and root mean square error(RMSE) are usually selected as a signal denoising perfor-mance evaluation which are calculated by following equa-tions
SNR = 10 lg(sum119873
119899=11199042(119899)
sum119873
119899=1[119904 (119899) minus 119904 (119899)]
2)
RMSE = radic 1119873
[119904(119899) minus 119904(119899)]2
(8)
where 119904(119899) is original signal 119904(119899) represents signal afterdenoising by wavelet packet and119873 is signal length
3 Blind Source Separation Modelsand Methods of DTS TemperatureMonitoring Data
Seepage will lead to a change in the stability temperature fieldin themedia and a partial volatile region can be generated [1011] However the temperature field distribution of structurebody is affected not only by leakage field but also soil char-acteristics natural phenomena optical fiber laying elevationcross artificial cavities and other factors Further the mea-sured value of the optical fiber temperature sensing system isaffected Therefore the temperature field of structure body isa result caused bymultiple-factor under the combined effectsThe positioning of leakage points inside ERJD under nationalcondition can be attributed to the blind source separationproblem for the influence of multiple-factor [12]
Leakages soil characteristics or other factors causingchanges in temperature monitoring data unit have a cor-responding source component which is a function of thedistance between the sampling point and the starting pointSoil characteristics natural phenomena and other factorsshould be possibly separated out of the temperature dataexcept for the leakage factorThe separated data can be a resultof temperature field change only affected by the leakage factoras much as possible And the leakage factor is just the finalconcern
The effects of leakage factor soil characteristics or otherfactors on the measured temperature data can also be con-sidered to be independent units The temperature data inmost regions along fiber are not affected by leakages guttersand other factors which are small probability events on timeand space Based on non-Gaussian distribution and statisticalindependent of the source signal ICA is selected in blindsource separation technique In addition PCA is determinedas a pretreatment method for ICA technique in order toreduce the computational complexity and effective multiple-variable correlation
The key of PCA is the certain degree of relevance andredundancy of information between the variables At thesame time adjacent sampling points have similar work envi-ronment and strong correlation which provide a possibilityto the use of PCA The mathematical model of PCA can beobtained by the following process First a 119901 dimensional
random vector is consisted by 119901 scalars and it is expressed as119883 = (119883
1 119883
119901)1015840 119901 is defined as an arbitrary number 119883 is
used for orthogonal transformation ordering 119884 = 1198791015840119883 and119879 is orthogonal matrix Each component of 119884 is not relevantand the variance of the first variable of 119884 is the largest thesecond variable coming afterThis process can be representedby the following matrix
1198841= 119905111198831+ 119905121198832+ sdot sdot sdot + 119905
1119901119883119901= 1198791015840
1119883
1198842= 119905211198831+ 119905221198832+ sdot sdot sdot + 119905
2119901119883119901= 1198791015840
2119883
119884119901= 11990511990111198831+ 11990511990121198832+ sdot sdot sdot + 119905
119901119901119883119901= 1198791015840
119901119883
(9)
The definition of singular value decomposition (SVD) ofPCA can be described 119872 represents a matrix with 119898 rowsand 119899 columns whose elements belong to the field of realnumbers or complex field There is the following decom-position
119872 = 119880119878119881119879 (10)
where 119880 represents a unitary matrix with 119898 times 119898 order andis called left singular matrix of119872 119881 is a unitary matrix with119899 times 119899 order and is named right singular matrix of119872 119881119879 istranspose ofmatrix119881 the same bellow 119878 represents a positivesemidefinite diagonal matrix with119898 times 119899 order and elementson the diagonal matrix are singular values of 119872 which areequal to the square root of eigenvalues of 119872119879 lowast 119872 It isorthogonal matrix as the elements of unitary matrix are realnumbers Relationship between the main component of 119884and monitoring data matrix119883 can be obtained by
119884 = 119883119879 (11)
Thus
119883 = 1198841198791015840 (12)
where matrix of 119879 is constituted by eigenvectors of diagonalmatrix and119883119879119883 is orthogonal matrix
From the theory of SVD it can be obtained that
119883 = 1198801198781198811015840 (13)
Comparing formula (12) with (13) it can be known that119879 = 119881 then 119884 = 119880119878 = 119883119881
Therefore the main component of matrix is the productof left singular matrix and its corresponding singular valuesand is also equal to the product of monitoring data matrixand the right singular matrix The number of principalcomponents can be determined through the cumulativevariance contribution rate Variance contribution rate andcumulative variance contribution rate are described follows
120593119896=
120582119896
sum119901
119896=1120582119896
120595119898=
sum119898
119896=1120582119896
sum119901
119896=1120582119896
(14)
4 Journal of Sensors
where 120582119896represents the eigenvalues of covariancematrix and
120593119896and 120595
119898are respectively called variance contribution rate
and cumulative variance contribution rateIn order to avoid inconsistencies between dimensions and
units of data the data is typically normalized
119883lowast
119894=
119883119894minus 119864 (119883
119894)
radic119863 (119883119894)
(15)
The mathematical model of ICA can be expressed by
1199091= 119886111199041+ 119886121199042+ sdot sdot sdot + 119886
1119899119904119899
1199092= 119886211199041+ 119886221199042+ sdot sdot sdot + 119886
2119899119904119899
119909119899= 11988611989911199041+ 11988611989921199042+ sdot sdot sdot + 119886
119899119899119904119899
(16)
where 1199091 1199092 119909
119899represent 119899 random variables named
mixed signals 1199041 1199042 119904
119899are 119899 independent components
called source signals 119886119894119895(119894 119895 = 1 2 119899) is mixing coeffi-
cientParallel algorithm calculation steps of several indepen-
dent components are described as follows
(1) standardize the data so that the mean and variance ofeach column are respectively 0 and 1
(2) data whitened of 119911 is gotten from the data obtainedfrom Step (1)
(3) 119898 is selected as the number of independent compo-nent to be estimated
(4) the mixing coefficient of 119908119894is initialized 119894 = 1
2 119898 and every119908119894is changed to be a unit 2-norm
then matrix 119882 is orthogonalised by the method ofStep (6)
(5) update each 119908119894 119908119894larr 119864119911119892(119908
119879
119894119911) minus 119864119892
1015840(119908119879
119894119911)119908119894
the function of 119892 can be selected from formula (17) toformula (19)
(6) matrix 119882 = (1199081 119908
119898)1015840 is orthogonalised 119882 larr
(1198821198821015840)minus12119882
(7) return to Step (5) if the result is not converged
The function of119892 can be selected from the following func-tions
1198921(119910) = tanh (119886
1119910) (17)
1198922(119910) = 119910 exp(minus
1199102
2
) (18)
1198923(119910) = 119910
3 (19)
Leakage location
Figure 2 Model experiment of ERJD
DTS system
Heating system
Figure 3 DTS monitoring and heating system
The corresponding derivative functions are described asfollows
1198921015840
1(119910) = 119886
1(1 minus tanh2 (119886
1119910))
1198921015840
2(119910) = (1 minus 119910
2) exp(minus
1199102
2
)
1198921015840
3(119910) = 3119910
2
(20)
where 1198861is a constant valuing in [1 2] and the value is usually
taken as 1 tanh(119909) represents hyperbolic tangent function
4 Application Case Analyses
A model experiment platform was formed and shown inFigure 2 to simulate the existence of leakages for ERJD TheDTS monitoring system of Sentinel DTS-LR produced bySensornet was applied The DTS monitoring and heatingsystem was shown in Figure 3 A schematic diagram of themodel was given in Figure 4 Multimode four core armoredcable of 50125 um was used as the monitoring optical cablewith a type of ZTT-GYXTW-4A1a
TheZTT-GYXTW-4A1a was a special optical cable whichcould be used to be heated And its structure section wasdrawn in Figure 5
Journal of Sensors 5
Monitoringopticalcable
Modelexperiment
pool of ERJD
Figure 4 Schematic diagram of the model
Water-blockingmaterial
Polyethylenejacket
Heatingwire
Fiberpaste
Opticalfibre
Loosetube
Steel-plasticcompound
Figure 5 Structure section of ZTT-GYXTW-4A1a
The leakage location was washed by a water pipe and thevelocity of seepage was controlled The monitoring data wasnoted as the stable velocity was formed
For simplicity just one leakage point was introduced inthe model and the sampling distance was 1 m and the sam-pling interval was 2 hours for 5 days Thus each measuringpoint obtained 60 groups of data First the original tempera-ture data should be denoised and a typical point was selectedas an example analyzed After calculation the decompositionlevels of the wavelet packet were 3-layer Figure 6 was shownto compare the raw signal with the wavelet denoising signal
The first principal component was calculated and listed inFigure 7
The corresponding variance contribution rate of the firstprincipal component 120590
1sum120590 was 8721 and it could be
concluded that the contribution of the first principal compo-nent to informationwas largerThe difference of the principalcomponent value between the leakage point presented and
Raw signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Wavelet denoising signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Figure 6 Comparative diagram between raw signal and the waveletdenoising signal
minus02
minus04
minus06
06
04
02
0
5 10 15 20 25 30 35
Distance (m)
Valu
e
Figure 7 Result diagram after PCA method
other points was much larger than the one of the first prin-cipal component Therefore the second and third principalcomponents were selected to the next ICA which meant119894 = 2 One of the two principal components was used toconstruct a separate component of the signal space and theother one was made to build a separate part of the salvagespace The difference of the separate component 1 in theleakage point was more evident than the one of other pointswhose distribution was more stable Thereby the separatecomponent 1 was considered to establish a salvage spaceThe separate component 2 constructed a separate componentsignal space The corresponding grey-scale map of residualspace was shown in Figure 8
From the result of data processing it can be known thatthe problem of the optical fiber temperature monitoring datacould be analyzed by the blind source separation and ICAwas used to process data In Figure 8 the leakage area of14m was larger than the area of 32m And the test distanceof the model was less than 32m Therefore the area of 32m
6 Journal of Sensors
minus05
0
05
1
minus1
minus15
Dist
ance
(m)
Time
Leakage point
0
5
10
15
20
25
30
6050403020100
Figure 8 Grey-scale map of residual space
was not considered Finally the leakage location could be wellimplemented
5 Conclusions
In this paper ERJD was the studied object and its leakageidentificationwas for the research emphasis Based on leakagecharacteristics of monitoring ERJD by DTS the affectingfactors of optical fiber temperaturemonitoring datawere ana-lyzed The wavelet packet denoising method and thresholddetermining method were also studied A blind source sepa-rationmodel of optical fiber temperature datawas finally builtand the corresponding temperature change process of theleakage source was extracted The implementation methodof blind source separation technique was discussed In-depth research on characteristics of optical fiber temperaturedata was carried out and optimal applicable conditions forrealization method of the blind source separation techniqueswere also analyzed The blind source separation techniqueimplementation combining ICA with PCA was finally pro-posed With leakage fiber-optic model experiment of ERJDthe reliability of optical fiber temperature measuring databased on the blind source separation method is validated
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research has been partially supported by Jiangsu NaturalScience Foundation (SN BK2012036) the National Natu-ral Science Foundation of China (SN 51179066 5113900141323001 and 51409167) the Specialized Research Fund forthe Doctoral Program of Higher Education of China (SN20130094110010) the Nonprofit Industry Financial Program
of MWR (SN 201301061 and 201201038) the Open Founda-tion of State Key Laboratory of Hydrology-Water Resourcesand Hydraulic Engineering (SN 20145027612) the ResearchProgram on Natural Science for Colleges and Universities inJiangsu Province (SN 14KJB520016) the Priority AcademicProgram Development of Jiangsu Higher Education Institu-tions (SN YS11001) and the Scientific Innovation ResearchScheme for Jiangsu University Graduate (SN 2013B25514)
References
[1] H-H Zhu A N L Ho J-H Yin H W Sun H-F Pei andC-Y Hong ldquoAn optical fibre monitoring system for evaluatingthe performance of a soil nailed sloperdquo Smart Structures andSystems vol 9 no 5 pp 393ndash410 2012
[2] B Shi C-S Tang L Gao C Liu and B-J Wang ldquoObservationand analysis of the urban heat island effect on soil in NanjingChinardquo Environmental Earth Sciences vol 67 no 1 pp 215ndash2292012
[3] J F Yan B Shi D F Cao H H Zhu and G Q Wei ldquoExper-iment study on carbon coated heating optical fiber for sandleakage monitoring based on distributed temperature systemrdquoin Proceedings of the 6th International Conference on StructuralHealth Monitoring of Intelligent Infrastructure Paper no MS03-06 Hong Kong 2013
[4] S Johansson ldquoLocalization and quantification of water leakagein ageing embankment dams by regular temperature measure-mentsrdquo inProceedings of the 17th International Congress on LargeDams (ICOLD rsquo91) pp 991ndash1005 Vienna Austria 1991
[5] O Kappelmeyer ldquoThe use of near surface temperature mea-surements for discovering anomalies due to causes at depthsrdquoGeophysical Prospecting vol 5 no 3 pp 239ndash258 1957
[6] A H Hartog ldquoDistributed fiber-optic temperature sensorsprinciples and applicationsrdquo inOptical Fiber Sensor TechnologyK T Grattan and B T Meggitt Eds pp 241ndash301 KluwerAcademic Publishers New York NY USA 2000
[7] H Z Su J Y Li J Hu and Z Wen ldquoAnalysis and back-analysisfor temperature field of concrete arch dam during constructionperiod based on temperature data measured by DTSrdquo IEEESensors Journal vol 13 no 5 pp 1403ndash1412 2013
[8] S Yin ldquoDistributed fiber optic sensorsrdquo in Fiber Optic SensorsF T Yu and S Yin Eds pp 201ndash229 Marcel Dekker New YorkNY USA 2000
[9] A D Kersey ldquoOptical fiber sensors for downwell monitoringapplications in the oil and gas industryrdquo in Proceedings of the13th International Conference on Optical Fiber Sensors (Ofs-13rsquo99) pp 326ndash331 1999
[10] S Grosswig A Graupner and E Hurtig ldquoDistributed fiberoptical temperature sensing techniquemdasha variable tool formonitoring tasksrdquo in Proceedings of the 8th International Sym-posium on Temperature and Thermal Measurements in Industryand Science pp 9ndash17 2001
[11] S A Wade K T V Grattan B McKinley L F Boswell andC DrsquoMello ldquo Incorporation of fiber optic sensors in concretespecimens testing and evaluationrdquo IEEE Sensors Journal vol 4no 1 pp 127ndash134 2004
[12] H-H Zhu L-C Liu H-F Pei and B Shi ldquoSettlement analysisof viscoelastic foundation under vertical line load using afractional Kelvin-Voigt modelrdquo Geomechanics and Engineeringvol 4 no 1 pp 67ndash78 2012
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
![Page 4: Research Article Blind Source Separation Model of Earth ...Research Article Blind Source Separation Model of Earth-Rock Junctions in Dike Engineering Based on Distributed Optical Fiber](https://reader033.vdocument.in/reader033/viewer/2022060910/60a48674852e25584376221c/html5/thumbnails/4.jpg)
4 Journal of Sensors
where 120582119896represents the eigenvalues of covariancematrix and
120593119896and 120595
119898are respectively called variance contribution rate
and cumulative variance contribution rateIn order to avoid inconsistencies between dimensions and
units of data the data is typically normalized
119883lowast
119894=
119883119894minus 119864 (119883
119894)
radic119863 (119883119894)
(15)
The mathematical model of ICA can be expressed by
1199091= 119886111199041+ 119886121199042+ sdot sdot sdot + 119886
1119899119904119899
1199092= 119886211199041+ 119886221199042+ sdot sdot sdot + 119886
2119899119904119899
119909119899= 11988611989911199041+ 11988611989921199042+ sdot sdot sdot + 119886
119899119899119904119899
(16)
where 1199091 1199092 119909
119899represent 119899 random variables named
mixed signals 1199041 1199042 119904
119899are 119899 independent components
called source signals 119886119894119895(119894 119895 = 1 2 119899) is mixing coeffi-
cientParallel algorithm calculation steps of several indepen-
dent components are described as follows
(1) standardize the data so that the mean and variance ofeach column are respectively 0 and 1
(2) data whitened of 119911 is gotten from the data obtainedfrom Step (1)
(3) 119898 is selected as the number of independent compo-nent to be estimated
(4) the mixing coefficient of 119908119894is initialized 119894 = 1
2 119898 and every119908119894is changed to be a unit 2-norm
then matrix 119882 is orthogonalised by the method ofStep (6)
(5) update each 119908119894 119908119894larr 119864119911119892(119908
119879
119894119911) minus 119864119892
1015840(119908119879
119894119911)119908119894
the function of 119892 can be selected from formula (17) toformula (19)
(6) matrix 119882 = (1199081 119908
119898)1015840 is orthogonalised 119882 larr
(1198821198821015840)minus12119882
(7) return to Step (5) if the result is not converged
The function of119892 can be selected from the following func-tions
1198921(119910) = tanh (119886
1119910) (17)
1198922(119910) = 119910 exp(minus
1199102
2
) (18)
1198923(119910) = 119910
3 (19)
Leakage location
Figure 2 Model experiment of ERJD
DTS system
Heating system
Figure 3 DTS monitoring and heating system
The corresponding derivative functions are described asfollows
1198921015840
1(119910) = 119886
1(1 minus tanh2 (119886
1119910))
1198921015840
2(119910) = (1 minus 119910
2) exp(minus
1199102
2
)
1198921015840
3(119910) = 3119910
2
(20)
where 1198861is a constant valuing in [1 2] and the value is usually
taken as 1 tanh(119909) represents hyperbolic tangent function
4 Application Case Analyses
A model experiment platform was formed and shown inFigure 2 to simulate the existence of leakages for ERJD TheDTS monitoring system of Sentinel DTS-LR produced bySensornet was applied The DTS monitoring and heatingsystem was shown in Figure 3 A schematic diagram of themodel was given in Figure 4 Multimode four core armoredcable of 50125 um was used as the monitoring optical cablewith a type of ZTT-GYXTW-4A1a
TheZTT-GYXTW-4A1a was a special optical cable whichcould be used to be heated And its structure section wasdrawn in Figure 5
Journal of Sensors 5
Monitoringopticalcable
Modelexperiment
pool of ERJD
Figure 4 Schematic diagram of the model
Water-blockingmaterial
Polyethylenejacket
Heatingwire
Fiberpaste
Opticalfibre
Loosetube
Steel-plasticcompound
Figure 5 Structure section of ZTT-GYXTW-4A1a
The leakage location was washed by a water pipe and thevelocity of seepage was controlled The monitoring data wasnoted as the stable velocity was formed
For simplicity just one leakage point was introduced inthe model and the sampling distance was 1 m and the sam-pling interval was 2 hours for 5 days Thus each measuringpoint obtained 60 groups of data First the original tempera-ture data should be denoised and a typical point was selectedas an example analyzed After calculation the decompositionlevels of the wavelet packet were 3-layer Figure 6 was shownto compare the raw signal with the wavelet denoising signal
The first principal component was calculated and listed inFigure 7
The corresponding variance contribution rate of the firstprincipal component 120590
1sum120590 was 8721 and it could be
concluded that the contribution of the first principal compo-nent to informationwas largerThe difference of the principalcomponent value between the leakage point presented and
Raw signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Wavelet denoising signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Figure 6 Comparative diagram between raw signal and the waveletdenoising signal
minus02
minus04
minus06
06
04
02
0
5 10 15 20 25 30 35
Distance (m)
Valu
e
Figure 7 Result diagram after PCA method
other points was much larger than the one of the first prin-cipal component Therefore the second and third principalcomponents were selected to the next ICA which meant119894 = 2 One of the two principal components was used toconstruct a separate component of the signal space and theother one was made to build a separate part of the salvagespace The difference of the separate component 1 in theleakage point was more evident than the one of other pointswhose distribution was more stable Thereby the separatecomponent 1 was considered to establish a salvage spaceThe separate component 2 constructed a separate componentsignal space The corresponding grey-scale map of residualspace was shown in Figure 8
From the result of data processing it can be known thatthe problem of the optical fiber temperature monitoring datacould be analyzed by the blind source separation and ICAwas used to process data In Figure 8 the leakage area of14m was larger than the area of 32m And the test distanceof the model was less than 32m Therefore the area of 32m
6 Journal of Sensors
minus05
0
05
1
minus1
minus15
Dist
ance
(m)
Time
Leakage point
0
5
10
15
20
25
30
6050403020100
Figure 8 Grey-scale map of residual space
was not considered Finally the leakage location could be wellimplemented
5 Conclusions
In this paper ERJD was the studied object and its leakageidentificationwas for the research emphasis Based on leakagecharacteristics of monitoring ERJD by DTS the affectingfactors of optical fiber temperaturemonitoring datawere ana-lyzed The wavelet packet denoising method and thresholddetermining method were also studied A blind source sepa-rationmodel of optical fiber temperature datawas finally builtand the corresponding temperature change process of theleakage source was extracted The implementation methodof blind source separation technique was discussed In-depth research on characteristics of optical fiber temperaturedata was carried out and optimal applicable conditions forrealization method of the blind source separation techniqueswere also analyzed The blind source separation techniqueimplementation combining ICA with PCA was finally pro-posed With leakage fiber-optic model experiment of ERJDthe reliability of optical fiber temperature measuring databased on the blind source separation method is validated
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research has been partially supported by Jiangsu NaturalScience Foundation (SN BK2012036) the National Natu-ral Science Foundation of China (SN 51179066 5113900141323001 and 51409167) the Specialized Research Fund forthe Doctoral Program of Higher Education of China (SN20130094110010) the Nonprofit Industry Financial Program
of MWR (SN 201301061 and 201201038) the Open Founda-tion of State Key Laboratory of Hydrology-Water Resourcesand Hydraulic Engineering (SN 20145027612) the ResearchProgram on Natural Science for Colleges and Universities inJiangsu Province (SN 14KJB520016) the Priority AcademicProgram Development of Jiangsu Higher Education Institu-tions (SN YS11001) and the Scientific Innovation ResearchScheme for Jiangsu University Graduate (SN 2013B25514)
References
[1] H-H Zhu A N L Ho J-H Yin H W Sun H-F Pei andC-Y Hong ldquoAn optical fibre monitoring system for evaluatingthe performance of a soil nailed sloperdquo Smart Structures andSystems vol 9 no 5 pp 393ndash410 2012
[2] B Shi C-S Tang L Gao C Liu and B-J Wang ldquoObservationand analysis of the urban heat island effect on soil in NanjingChinardquo Environmental Earth Sciences vol 67 no 1 pp 215ndash2292012
[3] J F Yan B Shi D F Cao H H Zhu and G Q Wei ldquoExper-iment study on carbon coated heating optical fiber for sandleakage monitoring based on distributed temperature systemrdquoin Proceedings of the 6th International Conference on StructuralHealth Monitoring of Intelligent Infrastructure Paper no MS03-06 Hong Kong 2013
[4] S Johansson ldquoLocalization and quantification of water leakagein ageing embankment dams by regular temperature measure-mentsrdquo inProceedings of the 17th International Congress on LargeDams (ICOLD rsquo91) pp 991ndash1005 Vienna Austria 1991
[5] O Kappelmeyer ldquoThe use of near surface temperature mea-surements for discovering anomalies due to causes at depthsrdquoGeophysical Prospecting vol 5 no 3 pp 239ndash258 1957
[6] A H Hartog ldquoDistributed fiber-optic temperature sensorsprinciples and applicationsrdquo inOptical Fiber Sensor TechnologyK T Grattan and B T Meggitt Eds pp 241ndash301 KluwerAcademic Publishers New York NY USA 2000
[7] H Z Su J Y Li J Hu and Z Wen ldquoAnalysis and back-analysisfor temperature field of concrete arch dam during constructionperiod based on temperature data measured by DTSrdquo IEEESensors Journal vol 13 no 5 pp 1403ndash1412 2013
[8] S Yin ldquoDistributed fiber optic sensorsrdquo in Fiber Optic SensorsF T Yu and S Yin Eds pp 201ndash229 Marcel Dekker New YorkNY USA 2000
[9] A D Kersey ldquoOptical fiber sensors for downwell monitoringapplications in the oil and gas industryrdquo in Proceedings of the13th International Conference on Optical Fiber Sensors (Ofs-13rsquo99) pp 326ndash331 1999
[10] S Grosswig A Graupner and E Hurtig ldquoDistributed fiberoptical temperature sensing techniquemdasha variable tool formonitoring tasksrdquo in Proceedings of the 8th International Sym-posium on Temperature and Thermal Measurements in Industryand Science pp 9ndash17 2001
[11] S A Wade K T V Grattan B McKinley L F Boswell andC DrsquoMello ldquo Incorporation of fiber optic sensors in concretespecimens testing and evaluationrdquo IEEE Sensors Journal vol 4no 1 pp 127ndash134 2004
[12] H-H Zhu L-C Liu H-F Pei and B Shi ldquoSettlement analysisof viscoelastic foundation under vertical line load using afractional Kelvin-Voigt modelrdquo Geomechanics and Engineeringvol 4 no 1 pp 67ndash78 2012
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
![Page 5: Research Article Blind Source Separation Model of Earth ...Research Article Blind Source Separation Model of Earth-Rock Junctions in Dike Engineering Based on Distributed Optical Fiber](https://reader033.vdocument.in/reader033/viewer/2022060910/60a48674852e25584376221c/html5/thumbnails/5.jpg)
Journal of Sensors 5
Monitoringopticalcable
Modelexperiment
pool of ERJD
Figure 4 Schematic diagram of the model
Water-blockingmaterial
Polyethylenejacket
Heatingwire
Fiberpaste
Opticalfibre
Loosetube
Steel-plasticcompound
Figure 5 Structure section of ZTT-GYXTW-4A1a
The leakage location was washed by a water pipe and thevelocity of seepage was controlled The monitoring data wasnoted as the stable velocity was formed
For simplicity just one leakage point was introduced inthe model and the sampling distance was 1 m and the sam-pling interval was 2 hours for 5 days Thus each measuringpoint obtained 60 groups of data First the original tempera-ture data should be denoised and a typical point was selectedas an example analyzed After calculation the decompositionlevels of the wavelet packet were 3-layer Figure 6 was shownto compare the raw signal with the wavelet denoising signal
The first principal component was calculated and listed inFigure 7
The corresponding variance contribution rate of the firstprincipal component 120590
1sum120590 was 8721 and it could be
concluded that the contribution of the first principal compo-nent to informationwas largerThe difference of the principalcomponent value between the leakage point presented and
Raw signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Wavelet denoising signal
Time (h)0 20 40 60 80 100 120
4
2
0
minus2
minus4
Tem
pera
ture
(∘C)
Figure 6 Comparative diagram between raw signal and the waveletdenoising signal
minus02
minus04
minus06
06
04
02
0
5 10 15 20 25 30 35
Distance (m)
Valu
e
Figure 7 Result diagram after PCA method
other points was much larger than the one of the first prin-cipal component Therefore the second and third principalcomponents were selected to the next ICA which meant119894 = 2 One of the two principal components was used toconstruct a separate component of the signal space and theother one was made to build a separate part of the salvagespace The difference of the separate component 1 in theleakage point was more evident than the one of other pointswhose distribution was more stable Thereby the separatecomponent 1 was considered to establish a salvage spaceThe separate component 2 constructed a separate componentsignal space The corresponding grey-scale map of residualspace was shown in Figure 8
From the result of data processing it can be known thatthe problem of the optical fiber temperature monitoring datacould be analyzed by the blind source separation and ICAwas used to process data In Figure 8 the leakage area of14m was larger than the area of 32m And the test distanceof the model was less than 32m Therefore the area of 32m
6 Journal of Sensors
minus05
0
05
1
minus1
minus15
Dist
ance
(m)
Time
Leakage point
0
5
10
15
20
25
30
6050403020100
Figure 8 Grey-scale map of residual space
was not considered Finally the leakage location could be wellimplemented
5 Conclusions
In this paper ERJD was the studied object and its leakageidentificationwas for the research emphasis Based on leakagecharacteristics of monitoring ERJD by DTS the affectingfactors of optical fiber temperaturemonitoring datawere ana-lyzed The wavelet packet denoising method and thresholddetermining method were also studied A blind source sepa-rationmodel of optical fiber temperature datawas finally builtand the corresponding temperature change process of theleakage source was extracted The implementation methodof blind source separation technique was discussed In-depth research on characteristics of optical fiber temperaturedata was carried out and optimal applicable conditions forrealization method of the blind source separation techniqueswere also analyzed The blind source separation techniqueimplementation combining ICA with PCA was finally pro-posed With leakage fiber-optic model experiment of ERJDthe reliability of optical fiber temperature measuring databased on the blind source separation method is validated
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research has been partially supported by Jiangsu NaturalScience Foundation (SN BK2012036) the National Natu-ral Science Foundation of China (SN 51179066 5113900141323001 and 51409167) the Specialized Research Fund forthe Doctoral Program of Higher Education of China (SN20130094110010) the Nonprofit Industry Financial Program
of MWR (SN 201301061 and 201201038) the Open Founda-tion of State Key Laboratory of Hydrology-Water Resourcesand Hydraulic Engineering (SN 20145027612) the ResearchProgram on Natural Science for Colleges and Universities inJiangsu Province (SN 14KJB520016) the Priority AcademicProgram Development of Jiangsu Higher Education Institu-tions (SN YS11001) and the Scientific Innovation ResearchScheme for Jiangsu University Graduate (SN 2013B25514)
References
[1] H-H Zhu A N L Ho J-H Yin H W Sun H-F Pei andC-Y Hong ldquoAn optical fibre monitoring system for evaluatingthe performance of a soil nailed sloperdquo Smart Structures andSystems vol 9 no 5 pp 393ndash410 2012
[2] B Shi C-S Tang L Gao C Liu and B-J Wang ldquoObservationand analysis of the urban heat island effect on soil in NanjingChinardquo Environmental Earth Sciences vol 67 no 1 pp 215ndash2292012
[3] J F Yan B Shi D F Cao H H Zhu and G Q Wei ldquoExper-iment study on carbon coated heating optical fiber for sandleakage monitoring based on distributed temperature systemrdquoin Proceedings of the 6th International Conference on StructuralHealth Monitoring of Intelligent Infrastructure Paper no MS03-06 Hong Kong 2013
[4] S Johansson ldquoLocalization and quantification of water leakagein ageing embankment dams by regular temperature measure-mentsrdquo inProceedings of the 17th International Congress on LargeDams (ICOLD rsquo91) pp 991ndash1005 Vienna Austria 1991
[5] O Kappelmeyer ldquoThe use of near surface temperature mea-surements for discovering anomalies due to causes at depthsrdquoGeophysical Prospecting vol 5 no 3 pp 239ndash258 1957
[6] A H Hartog ldquoDistributed fiber-optic temperature sensorsprinciples and applicationsrdquo inOptical Fiber Sensor TechnologyK T Grattan and B T Meggitt Eds pp 241ndash301 KluwerAcademic Publishers New York NY USA 2000
[7] H Z Su J Y Li J Hu and Z Wen ldquoAnalysis and back-analysisfor temperature field of concrete arch dam during constructionperiod based on temperature data measured by DTSrdquo IEEESensors Journal vol 13 no 5 pp 1403ndash1412 2013
[8] S Yin ldquoDistributed fiber optic sensorsrdquo in Fiber Optic SensorsF T Yu and S Yin Eds pp 201ndash229 Marcel Dekker New YorkNY USA 2000
[9] A D Kersey ldquoOptical fiber sensors for downwell monitoringapplications in the oil and gas industryrdquo in Proceedings of the13th International Conference on Optical Fiber Sensors (Ofs-13rsquo99) pp 326ndash331 1999
[10] S Grosswig A Graupner and E Hurtig ldquoDistributed fiberoptical temperature sensing techniquemdasha variable tool formonitoring tasksrdquo in Proceedings of the 8th International Sym-posium on Temperature and Thermal Measurements in Industryand Science pp 9ndash17 2001
[11] S A Wade K T V Grattan B McKinley L F Boswell andC DrsquoMello ldquo Incorporation of fiber optic sensors in concretespecimens testing and evaluationrdquo IEEE Sensors Journal vol 4no 1 pp 127ndash134 2004
[12] H-H Zhu L-C Liu H-F Pei and B Shi ldquoSettlement analysisof viscoelastic foundation under vertical line load using afractional Kelvin-Voigt modelrdquo Geomechanics and Engineeringvol 4 no 1 pp 67ndash78 2012
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
![Page 6: Research Article Blind Source Separation Model of Earth ...Research Article Blind Source Separation Model of Earth-Rock Junctions in Dike Engineering Based on Distributed Optical Fiber](https://reader033.vdocument.in/reader033/viewer/2022060910/60a48674852e25584376221c/html5/thumbnails/6.jpg)
6 Journal of Sensors
minus05
0
05
1
minus1
minus15
Dist
ance
(m)
Time
Leakage point
0
5
10
15
20
25
30
6050403020100
Figure 8 Grey-scale map of residual space
was not considered Finally the leakage location could be wellimplemented
5 Conclusions
In this paper ERJD was the studied object and its leakageidentificationwas for the research emphasis Based on leakagecharacteristics of monitoring ERJD by DTS the affectingfactors of optical fiber temperaturemonitoring datawere ana-lyzed The wavelet packet denoising method and thresholddetermining method were also studied A blind source sepa-rationmodel of optical fiber temperature datawas finally builtand the corresponding temperature change process of theleakage source was extracted The implementation methodof blind source separation technique was discussed In-depth research on characteristics of optical fiber temperaturedata was carried out and optimal applicable conditions forrealization method of the blind source separation techniqueswere also analyzed The blind source separation techniqueimplementation combining ICA with PCA was finally pro-posed With leakage fiber-optic model experiment of ERJDthe reliability of optical fiber temperature measuring databased on the blind source separation method is validated
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research has been partially supported by Jiangsu NaturalScience Foundation (SN BK2012036) the National Natu-ral Science Foundation of China (SN 51179066 5113900141323001 and 51409167) the Specialized Research Fund forthe Doctoral Program of Higher Education of China (SN20130094110010) the Nonprofit Industry Financial Program
of MWR (SN 201301061 and 201201038) the Open Founda-tion of State Key Laboratory of Hydrology-Water Resourcesand Hydraulic Engineering (SN 20145027612) the ResearchProgram on Natural Science for Colleges and Universities inJiangsu Province (SN 14KJB520016) the Priority AcademicProgram Development of Jiangsu Higher Education Institu-tions (SN YS11001) and the Scientific Innovation ResearchScheme for Jiangsu University Graduate (SN 2013B25514)
References
[1] H-H Zhu A N L Ho J-H Yin H W Sun H-F Pei andC-Y Hong ldquoAn optical fibre monitoring system for evaluatingthe performance of a soil nailed sloperdquo Smart Structures andSystems vol 9 no 5 pp 393ndash410 2012
[2] B Shi C-S Tang L Gao C Liu and B-J Wang ldquoObservationand analysis of the urban heat island effect on soil in NanjingChinardquo Environmental Earth Sciences vol 67 no 1 pp 215ndash2292012
[3] J F Yan B Shi D F Cao H H Zhu and G Q Wei ldquoExper-iment study on carbon coated heating optical fiber for sandleakage monitoring based on distributed temperature systemrdquoin Proceedings of the 6th International Conference on StructuralHealth Monitoring of Intelligent Infrastructure Paper no MS03-06 Hong Kong 2013
[4] S Johansson ldquoLocalization and quantification of water leakagein ageing embankment dams by regular temperature measure-mentsrdquo inProceedings of the 17th International Congress on LargeDams (ICOLD rsquo91) pp 991ndash1005 Vienna Austria 1991
[5] O Kappelmeyer ldquoThe use of near surface temperature mea-surements for discovering anomalies due to causes at depthsrdquoGeophysical Prospecting vol 5 no 3 pp 239ndash258 1957
[6] A H Hartog ldquoDistributed fiber-optic temperature sensorsprinciples and applicationsrdquo inOptical Fiber Sensor TechnologyK T Grattan and B T Meggitt Eds pp 241ndash301 KluwerAcademic Publishers New York NY USA 2000
[7] H Z Su J Y Li J Hu and Z Wen ldquoAnalysis and back-analysisfor temperature field of concrete arch dam during constructionperiod based on temperature data measured by DTSrdquo IEEESensors Journal vol 13 no 5 pp 1403ndash1412 2013
[8] S Yin ldquoDistributed fiber optic sensorsrdquo in Fiber Optic SensorsF T Yu and S Yin Eds pp 201ndash229 Marcel Dekker New YorkNY USA 2000
[9] A D Kersey ldquoOptical fiber sensors for downwell monitoringapplications in the oil and gas industryrdquo in Proceedings of the13th International Conference on Optical Fiber Sensors (Ofs-13rsquo99) pp 326ndash331 1999
[10] S Grosswig A Graupner and E Hurtig ldquoDistributed fiberoptical temperature sensing techniquemdasha variable tool formonitoring tasksrdquo in Proceedings of the 8th International Sym-posium on Temperature and Thermal Measurements in Industryand Science pp 9ndash17 2001
[11] S A Wade K T V Grattan B McKinley L F Boswell andC DrsquoMello ldquo Incorporation of fiber optic sensors in concretespecimens testing and evaluationrdquo IEEE Sensors Journal vol 4no 1 pp 127ndash134 2004
[12] H-H Zhu L-C Liu H-F Pei and B Shi ldquoSettlement analysisof viscoelastic foundation under vertical line load using afractional Kelvin-Voigt modelrdquo Geomechanics and Engineeringvol 4 no 1 pp 67ndash78 2012
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of
![Page 7: Research Article Blind Source Separation Model of Earth ...Research Article Blind Source Separation Model of Earth-Rock Junctions in Dike Engineering Based on Distributed Optical Fiber](https://reader033.vdocument.in/reader033/viewer/2022060910/60a48674852e25584376221c/html5/thumbnails/7.jpg)
International Journal of
AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
RoboticsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Active and Passive Electronic Components
Control Scienceand Engineering
Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
International Journal of
RotatingMachinery
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporation httpwwwhindawicom
Journal ofEngineeringVolume 2014
Submit your manuscripts athttpwwwhindawicom
VLSI Design
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Shock and Vibration
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Civil EngineeringAdvances in
Acoustics and VibrationAdvances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Electrical and Computer Engineering
Journal of
Advances inOptoElectronics
Hindawi Publishing Corporation httpwwwhindawicom
Volume 2014
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
SensorsJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Chemical EngineeringInternational Journal of Antennas and
Propagation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Navigation and Observation
International Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
DistributedSensor Networks
International Journal of