process performance monitoring in the presence of confounding variation baibing li, elaine martin...
TRANSCRIPT
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
PROCESS PERFORMANCE MONITORING IN THE PRESENCE
OF CONFOUNDING VARIATION
Baibing Li, Elaine Martin and Julian MorrisUniversity of Newcastle
Newcastle upon Tyne, England, UK
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Techniques for Improved Operation
Enhanced Profitabilityand
Improved CustomerSatisfaction Modern Process
ControlSystems
Process Optimisation
Process Monitoring for Early Warning
andFault Detection
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Mechanistic models developed from process mass and energy balances and kinetics provide the ideal form given:
process understanding exists time is available to construct the model.
Data based models are useful alternatives when there is:
limited process understanding process data available from a range of operating conditions.
Hybrid models combine several different approaches.
Process Modelling
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Industrial Semi-discrete Manufacturing Process
Consider a situation where a variety of products (recipes) are produced, some of which are only manufactured in small quantities to meet the requirements of specialist markets.
Thirty-six process variables are recorded every minute, whilst five quality variables are measured off-line in the quality laboratory every two hours.
A nominal process performance monitoring scheme was developed using PLS from 41 ‘ideal’ batches, based on 3 recipes.
A further 6 batches, A4, A10, A29, A35, A38 and B32, that lay outside the desirable specification limits were used for model interrogation.
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Industrial Semi-discrete Manufacturing Process
-10 -5 0 5 10
-6
-4
-2
0
2
4
6
A
C
A
A A
B
A
B
A
B
C
B
A A
A A
C C
A
C C C
A
A A
A
C A A A
A
A
B
A
A
B
A
A A A
t1
t2
-6 -4 -2 0 2 4 6
-6
-4
-2
0
2
4
6
A
C
A A A B
A
B
A
B
C
B A A
A A
C C
A C C C
A A A A
C
A
A A A A
B
A A
B
A
A A
A
t3
t4
Latent variable 1 V Latent variable 2 Latent variable 3 V Latent variable 4
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Industrial Semi-discrete Manufacturing Process
0 10 20 30 400
5
10
15
20
T2 s
tatis
tic
0 10 20 30 400
10
20
30
40
50
SP
E
-10 -5 0 5 10
-6
-4
-2
0
2
4
6
A
C
A
A
A A
B
A
B
A
A
B
C
B
A A
A A
C C
A
C C C C
A
A A
A
A
C
B
A A
A
A
A
A
A
B
A
A
B
A
A A A
4
10
29
32
35 38
t1
t2
Bivariate Scores Plot Hotelling’s T2 and SPE
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Industrial Semi-discrete Manufacturing Process
By applying ordinary PLS, the variability between recipes dominates the model and hence masks the variability within a specific recipe that is of primary interest.
Two solutions to this have been proposed
The multi-group approach (Hwang et al,1998)
Generic modelling (Lane et al, 1997, 2001)
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Process Modelling
Traditionally two types of variables have been used in the development of a process model/process performance monitoring scheme:
Process variables (X) Quality variables (Y)
In practice, a third class of variables exists:
Confounding variables (Z).
A confounding variable is any extraneous factor that is related to, and affects, the two sets of variables under study (X) and (Y).
It can result in a distortion of the true relationship between the two sets of variables, that is of primary interest.
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Confidence ellipse including confounding
variation
Trajectory of confounding
variable Confidence ellipse excluding confounding variation
X
X X X X XX
Mal-operation
Global Process Variation
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
To exclude the nuisance source of variability, a necessary condition is that the derived latent variables, , and , are not correlated with the confounding variables:
and for .
The idea of constrained PLS is to apply the constraints given by equation to ordinary PLS.
0tZ hT 0uZ h
T Ah ,,1
ht hu
2T minarg hhh twEtt
2T minarg hhh uqFuu
2T minargT
hhh twEt0tZ
2T minargT
hhh uqFu0uZ
Constrained PLS
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Standard constrained optimisation techniques can be used to solve the equations in each iteration.
An algorithm has been developed that enhances the efficiency of the constrained PLS algorithm.
The other steps of the constrained PLS are as for ordinary PLS.
The resulting latent variables can then be used for process monitoring with the knowledge that they are not confounded with the nuisance source of variability.
Any unusual variation detected from these latent variables can then be assumed to be related to abnormal process behaviour.
Constrained PLS
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Simulation Example
Consider a process where the confounding variation is a result of recipe changes.
Recipe A - Observations [1, 50]. Recipe B - Observations [51, 100]. Recipe C - Observations [101, 150].
Measurements on three process variables and two quality variables were made over 150 time points.
Non-conforming operation occurred at time points 1, 2, 51 to 54, and 101, 102.
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Simulation Example - Scatter Plot
-5 0 5 10 15-5
0
5
10
15
1 52
2 51
53
54
101 102
Process variable 1
Pro
cess
var
iabl
e 2
Recipe ARecipe BRecipe C
The process variables x1 and x2 for recipes A, B, C
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Simulation Example - Bivariate Scores
-4 -3 -2 -1 0 1 2 3 4
-4
-3
-2
-1
0
1
2
3
4
1 52
2 51
5354 101 102
t1
t2
-8 -6 -4 -2 0 2 4 6 8-6
-4
-2
0
2
4
6
1
52
2
51
53
54
101
102
t1t2
Ordinary PLS Constrained PLS
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Simulation Example - T2 Chart
0 50 100 1500
2
4
6
8
10
12
T2 s
tatis
tic
0 50 100 1500
10
20
30
40
50
60
70
80
90
T2 s
tatis
tic
Ordinary PLS Constrained PLS
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Orthogonal Signal Correction (OSC)
Wold et al.’s (1989) OSC algorithm operates by removing those wavelengths of the spectra that are unrelated to the target variables.
It achieves this by ensuring that the wavelengths that are removed are mathematically orthogonal to the target variables or as close to the orthogonal as possible
Although OSC and the PLS filter have similar bilinear structures, the objective and methodology of OSC in terms of extracting the systematic part, T, differs to that of the constrained approach.
The OSC algorithm is based on PCA where at each iteration, that variation associated with the response variables is removed.
The filter in constrained PLS is based on the PLS algorithm. The process signal, X, is related to the confounding information, Z, through PLS.
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Simulation Example - Comparison with OSC
0 50 100 1500
2
4
6
8
10
12
T2 s
tatis
tic
0 50 100 1500
10
20
30
40
50
60
70
80
90
T2 s
tatis
tic
OSC - Ordinary PLS Constrained PLS
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Simulation Example - Comparison with OSC
-4 -3 -2 -1 0 1 2 3 4-4
-3
-2
-1
0
1
2
3
4
1
52
2
51
53
54
101
102
t1
t2
-8 -6 -4 -2 0 2 4 6 8-6
-4
-2
0
2
4
6
1
52
2
51
53
54
101
102
t1t2
Constrained PLSOSC - Ordinary PLS
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Continuous Confounding Variables
In some processes there exist ‘recipe’ or ‘operating condition set-point’ variables that are varied continuously during production to meet changing customer requirements.
The variation caused by these continuously varying recipe variables, i.e. confounding variables, is usually not of direct interest for process monitoring.
In this situation the effect of the confounding variables should be removed so that the detection of more subtle process changes and malfunctions is not masked.
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Continuous Confounding Variable
Consider a process where the confounding variation is a result of a continuously changing variable.
The confounding variable continuously takes values in the interval
[0, 1].
Measurements on three process variables and two quality variables were made over 100 time points.
Samples 1, 2, 51 and 52 are representative of non-conforming operation.
Non-conforming operation was generated by adding a disturbance term to process variables one and two but not to process variable three
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Continuous Confounding Variable
Scatter plot of the process variables x1 and x2
-5 0 5 10 15-6
-4
-2
0
2
4
6
8
10
12
14
16
1
2
51
52
x1
x2
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Ordinary PLS - Three Latent Variables
Hotelling’s T2Squared Prediction Error
0 20 40 60 80 1000
2
4
6
8
10
12
14
16
T2
statis
tic
0 20 40 60 80 1000
0.5
1
1.5
2
2.5
3
SPE
X-block comprising process and confounding variables
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
SPE Contribution Plot
1 2 3 40
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Process Variables
Sca
led
Con
trib
utio
n V
alue
s
SPE contribution plot for observation 51
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Ordinary PLS - Two Latent Variables
0 20 40 60 80 1000
2
4
6
8
10
12
T2 s
tatis
tic
0 20 40 60 80 1000
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
SPE
X-block comprising only process variables
Hotelling’s T2 Squared Prediction Error
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
OSC based Ordinary PLS
0 20 40 60 80 1000
2
4
6
8
10
12
14
T2
statis
tic
0 20 40 60 80 1000
0.5
1
1.5
2
2.5
3
SPE
Hotelling’s T2Squared Prediction Error
Two latent variables
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Constrained PLS
0 20 40 60 80 1000
10
20
30
40
50
60
70
80
90
100
T2 s
tatist
ic
0 20 40 60 80 1000
2
4
6
8
10
12
14
16
18
SPE
Two Latent Variables
Hotelling’s T2 Squared Prediction Error
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Scores Contribution Plot
1 2 3-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Process Variables
Sca
led
Con
trib
utio
n V
alue
s
1 2 3-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Process Variables
Sca
led
Con
trib
utio
n V
alue
s
Scores contribution plot for observation 51
Latent variable 1 Latent variable 2
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Industrial Semi-discrete Manufacturing Process
Returning to the industrial semi-discrete batch manufacturing process the advantages of the constrained PLS algorithm over ordinary PLS.
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Industrial Semi-discrete Manufacturing Process
0 10 20 30 400
5
10
15
20
T2 s
tatis
tic
0 10 20 30 400
10
20
30
40
50
SP
E
-10 -5 0 5 10
-6
-4
-2
0
2
4
6
A
C
A
A
A A
B
A
B
A
A
B
C
B
A A
A A
C C
A
C C C C
A
A A
A
A
C
B
A A
A
A
A
A
A
B
A
A
B
A
A A A
4
10
29
32
35 38
t1
t2
Bivariate Scores Plot Hotelling’s T2 and SPE
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
-10 -5 0 5 10
-6
-4
-2
0
2
4
6
A
C
A
A
A A
B
A
B
A
A
B
C B
A
A
A A
C
C
A
C C C
C
A
A A
A A
C
B
A
A A
A A
A
A
B
A A
B A
A
A
A
35 38
4 10
29
32 35 38
t1
t2
-5 0 5-15
-10
-5
0
5
10
15
A C A
A
A A B A
B
A
A
B
C B
A A A
A C C A
C C
C
C
A A A
A
A
C
B
A A
A
A A
A
A
B A
A
B A A A
A
4
10
29 32 35
38
t3
t4
LV 1 versus LV 2 LV 3 versus LV 4
Constrained Partial Least Squares
Industrial Application
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
0 10 20 30 400
100
200
300
T2 s
tatistic
0 10 20 30 400
200
400
600
800
1000
SP
E
Hotelling’s T2
Squared Prediction Error
Constrained Partial Least Squares
Industrial Application
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
0 5 10 15 20 25 30 35-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Process Variables
S
cale
d C
ontr
ibut
ion
Val
ues
Contribution Plot
Industrial Application
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Constrained PLS - Conclusions
Constrained PLS possesses the following important characteristics:
It removes that information correlated with the confounding variables.
The information excluded by constrained PLS contains only variation associated with the confounding variables.
The derived constrained PLS latent variables achieve optimality in terms of extracting as much of the available information as possible contained in the process and quality data.
Centre for Process Analytics and Control Technology (CPACT)University of Newcastle, UK
Acknowledgements
The authors acknowledge the financial support of the EU ESPRIT PERFECT No. 28870 (Performance Enhancement through Factory On-line Examination of Process Data).
They also acknowledge colleagues at BASF Ag. for stimulating the research, in particular Gerhard Krennrich and Pekka Teppola.