The Space Congress® Proceedings 1992 (29th) Space - Quest For New Fontiers

Apr 21st, 2:00 PM

Paper Session I-B - GSIM: A Low Cost MBR Approach for KSC Paper Session I-B - GSIM: A Low Cost MBR Approach for KSC

Design and Operations Design and Operations

John Wilkinson Lockheed Space Operations Company

Follow this and additional works at: https://commons.erau.edu/space-congress-proceedings

Scholarly Commons Citation Scholarly Commons Citation Wilkinson, John, "Paper Session I-B - GSIM: A Low Cost MBR Approach for KSC Design and Operations" (1992). The Space Congress® Proceedings. 12. https://commons.erau.edu/space-congress-proceedings/proceedings-1992-29th/april-21-1992/12

This Event is brought to you for free and open access by the Conferences at Scholarly Commons. It has been accepted for inclusion in The Space Congress® Proceedings by an authorized administrator of Scholarly Commons. For more information, please contact commons@erau.edu.

GSIM: A Low Cost MBR Approach for KSC Design and Operations

AUTHORS

John "JW" WilkinsonLockheed Space Operations Company

MS: LSO-002Titusville, Fl 32780

(407)383-2200x2797

ABSTRACT

NASA research, over the past 8 years, has developed Model Based Reasoning (MBR) technology to the point where it can now be applied to solve operational problems. GSIM is a tool developed jointly by NASA and LSOC that builds on this research, making MBR technology cost effective and readily available. Using GSIM, high fidelity MBR models can be developed on standard office computers and workstations. Using the intuitive GSIM model development environment, these models can be constructed at the rate of approximately IQx that required for development of lower fidelity math models, such as those used today for launch team training and application software validation. MBR models are not single use investments; but support diverse functions such as design assistance, traditional operations simulation, training, FMEA, real-time system monitoring/diagnosis and data fusion.

This paper presents findings/plans of the GSIM project and discusses long term benefits of wide spread life cycle oriented use of MBR technology at KSC.

DEFINITION

mod-el 4:a miniature representation of something; also: a pattern of something to be made 5:an example for imitation or emulation 11: a description or analogy used to help visualize something...that cannot be directly observed 12:a system of postulates, data, or inferences presented as a mathematical description of an entity or state of affairs. [Webster87]

ob-ject la:something material that may be perceived by the senses. [Webster87]

class 3:a group, set, or kind sharing common attributes: as a:a major category in biological taxonomy ranking above or below the phylum or division. [Webster87]

INTRODUCTION

Model Based Reasoning, or MBR, is a unique and powerful approach for applying computers to solve real-world problems. Its most distinctive characteristic is its separation of data and program. For example, MBR could be applied by an auto manufacturer to provide services for all its makes of vehicles. A separate model could be constructed for each make of vehicle. Application programs could then be developed to perform each kind service that was needed e.g. printing wiring diagrams or implementing a diagnostic service center to be placed in each dealers shop. Each application program would work for all makes of vehicle, (past, present or future) for which a model was developed.

MBR Models capture how something works and what it is. This is done by defining its component parts and the interaction between those parts. MBR Engines are application programs that work on MBR Models. As shown in Figure 1, MBR Models are the fuel required to run any MBR Engine. Thus the maintenance cycles of both models and engines are separate. Both models and engines can be modified or added independently without affecting the other.

A secondary benefit of the use of MBR systems is a dramatic reduction in the quantity of software that must be developed or maintained. Figure 2 illustrates the effect of applying MBR technology to the Launch Processing System application software currently required for Shuttle processing and launch at KSC. In the MBR approach, the primary investment for developing new

2-42

GSIM: A Low Cost MBR Approach for KSC Design and Operations

systems is the MBR Model which roughly equivalent to the schematics & specifications task today.

This leads to a top level organization for MBR systems such as that shown in Figure 3. This organization is in fact the one used by the GSIM (Graphics Based Simulation) project. It consists of MBR tools, engines, models, and classes. MBR engines apply model data to provide a direct service like simulation or diagnosis. MBR tools provide utility services such as import/export to commercial products, model library management and printing. Models are composed of objects and the inter-object connections which is equivalent to the data captured by a schematic. Each symbol on the schematic becomes an object. Example objects might be: a solenoid, a hydraulic pump, or a hydraulic line. Each object belongs to a particular class of objects. This is equivalent to a vendor part number for an object. This same class can be used many times on a single page of a schematic.

Figure 1: MR liiiii engines

Schematics 1 & Specs i|

**-*

yModel _ Data

Model 1 Programs |

single programs |__ Application |, \

Engines || f

J

1 'Goafn AppL PgpK. 1 ! AppL

Conventional & cunenl

ol 1 PpmsJ|

t approach.MBR approach

• Intelligent - reasons, over model

* Data driven

• Reuseable engines

• Multi-use models

• Minimal cost/time for new applications

Figure 2: MBR lowers maintenance and facilitates improvements

Classes let us define once how a class of objects works and reuse it many times.

MBR technology has been under investigation / development over the past 8 years at KSC. The driving force behind this MBR development has been diagnosis, however simulation is one of the natural spin- off benefits. The pilot system was LES (Lox

Expert System). It was developed in 1984- 85 by NASA/KSC, assisted by LSOC, to expedite problem diagnosis dining Shuttle Lox loading operations.[Jamieson85] The follow on system was KATE (Knowledge- based Autonomous Test Engineer) which has been under development from 1986 to present.[New87] KATE adds the capability to control as well as diagnose. Both systems

2-43

GSIM: A Low Cost MBR Approach for KSC Design and Operations

have some simulation and dynamic schematic drawing capabilities as well. The third system was SCAN (Shuttle Connector Analysis Network) which was developed by LSOC (Lockheed Space Operations Company) and is the only one of the three to

MBR Tool

Application Engine

Model

Class

i. Links COTS & other tools to MBR In Reuseable across all applications

•L, Reuseable across all models 1 m Intelligent

~|j Linkable building Mocks Applies MBR technology

™ Reuseable across all engines m Declarative

Defines objedts & connections

T Reuseable across all model objects ||| Level corresponds to LRLTs

i-Wtf$<ww."m

Figure 3: MBR top level organization

be operationally deployed to date. [Wilkinson88] SCAN is used to track the configuration of the Shuttle wiring harnesses and connectors and verify thorough testing after each connector reconnect operation. SCAN is also used at JSC for in-flight analysis; and by Rockwell International at their Palmdale facility for Shuttle fabrication and modification. However, SCAN does not perform MBR diagnosis or simulation. A new commercial system has recently been introduced named MATE (Model-based Autonomous Test Engineer) that provides diagnosis using a limited set of symbols.[Jamieson91]

THE ENGINEERING TASK

Figure 4 depicts the engineering products and services required from design through operations processing and launch which consume major engineering resources. This list is not unique to Shuttle processing but is typical of processing any space vehicle. The traditional route involves separate tasks, separate deliverables and frequently entirely

separate groups of personnel to carry out each task. Since each of these tasks are centered around capturing particular processes and deliverables, work done on one task has little carry over benefit in accomplishing the next. Yet, each part is quite necessary. A change in the Shuttle or GSE (Ground Support Equipment) routinely necessitates changes in multiple or all of the tasks and deliverables.

An MBR view of the engineering task focuses first on capturing system knowledge in a standard and reusable form. This task is the creation of the MBR model. Once completed, MBR engines are used to accomplish or assist in many of the other activities. The MBR model becomes the primary focal point for changes in the vehicle or GSE. MBR engines become the focal point for changes in process requirements. The resulting synergistic effect reduces the cost and time required to implement required changes. It also makes increases in quality and depth of services economical. That is, changing a single MBR engine is far less costly than making the corresponding change in all application documentation or all application programs of the affected type.

THE GSIM CONCEPT

GSIM is intended to make MBR technology affordable and accessible to the engineering community at large. Figure 5 illustrates the driving forces behind the design of GSIM that are considered necessary to support the engineering task described earlier.

As depicted in Figure 6, GSIM combines spreadsheet and simplified CAD schematic metaphors, to provide a readily understandable user interface requiring minimal training. Models are defined using standard configurations of low cost PC and Unix workstations. No special software or hardware is required. The performance objective for office computers is that a subsystem model, e.g. Environmental Control System, should be able to run on a PC compatible 386 running at 25 MHz. Model classes may also be defined across platforms. This level of portability is obtained by establishing a standard interface layer for graphics and other computational

2-44

GSIM: A Low Cost MBR Approach for KSC Design and Operations

mmmmmm Control &1

V^ftixIxjx^XxXjxixjxox

pl'T^

•will Control & Monitor AppsfiiTJiii£:::|:i::::::::::&%^

I FMEA/Engineering Analys

Requirements, Design & User Documentation

MBR-KnowIedge capture

-Highly manual

-Partly automated

-Highly automated

Manual Tasks _

Diagnosis

Control & Monitor Support

Training

Test Plans & Validation

Control & Monitor Apps

Simulation JI FMEA/Engineering Analysis J

— W Schematics & Specs J

Requirements, Design L & User Documentation

Traditional -Process capture

Figure 4: Focus on capture of system knowledge versus process knowledge

GSIM Objectives• Efficient model capture

• No special H/W, S/W or S/W environment requirements

• Model development & testing on existing office and personal computers

• Portable

• High fidelity/efficient performance

»Generic tools customizable without changing tool

•Real-time embeddabte in future control systems

> Data interchange with existing MBR and COTS CAD tools

GSIM ApproachSchematic capture mode entry

Only requirement is C++ compiler

Presently runs on PC/ATs and PS/2s running DOS, Sun/compatables and Apollo running Unix/XWindows

Readily rehostable portable l/F layer

Hi-fi at LRU object level with adequate subsystem model performance on 25 MHz PC

Formalized interfaces used for rehosting and embedding in RT applications

Through tool customization above.

SQL MBR library planned with filters for specific tools

Figure 5: GSIM is designed for efficient low cost exploitation of MBRtechnology.

2-45

GSIM: A Low Cost MBR Approach for KSC Design and Operations

services. All GSIM tools and engines conform to these standards. This permits the GSIM to be ported to a new machine by writing a small library of routines that map GSIM standard interface calls to those available on the new machine,

'Using a schematic capture approacfa, for entry and maintenance of models, allows engineers to efficiently develop and 'test models. They use a graphical and. non-programming method that is quickly learned by users of either CAD or spreadsheet applications. Dining the proof of concept phase of GSIM, a miniature Liquid Oxygen loading model was developed including about 20 new classes and 851 objects in about 2 man- weeks. Current estimates of improved productivity over the current approach are about 10:1. Model fidelity is significantly

higher since models are defined down to the schematic element level or class. Classes are then defined supplying transfer functions for each class. Analog transfer functions are defined to GSIM typically, by entering transfer curves graphically while referencing vendor specifications. Qasses also declare animation and specification data

GSIM's tools and engines are general and modular in nature permitting them to be applied readily to new applications. The Controlled Model Data Store is equipped with import/export services. These services act as filters to convert between the GSIM standard MBR data format of the controlled library and those used by other MBR or COTS tools. This permits each of these other tools to view GSIM MBR standardized data as if it were defined in the format of that tool.

12 3 4

5 6

7

A B C D E F G1.

4•• ••

J

»

•• ••

Figure 6: The GSIM architecture is modular and open.

The GSIM user does not normally dealdirectly with MBR tools and engines.Instead, an environment is established for the

that is a collection of the appropriatetools, engines and glue to do something

Figure 7 outlines the key GSIM

environments and their functions. GSIM environments add a user-friendly and consistent coating and accomplish a set of services. Both tools and engines are designed to support integration into different environments.

2-46

GSIM: A Low Cost MBR Approach for KSC Design and! Operations

Model Development Environment

• Schematic capture

• Maintenance * Validation & CM « Schematic/MBR oriented analysis

Class Development Environment

* Heirarchical classes

•Maintenance • C AD DXF import • Project specific

& shared classesI

Real-time Simulation

• VME standalone

Access to Fanner Ftojects/COTS

•KATE• LES ike 'Online^

analysis »CAD^ • CAD plotters

KATE: LES: LOX

Figure 7: GSIM environments combine MBR engines and tools for

LOCKHEED'S PROGRESS & PLANS

Work on GSIM began in November 1990 in conjunction with a Simulation Host Study for the Launch Team Training System. This study sought to determine if present KSC computational resources and models were sufficient for an expanded launch team training role. GSIM's part in the study was to determine if new technology should be applied to simulation. GSIM's proof of concept work showed that the new modeling technique and environments offered by GSIM, had the higher fidelity desired and a higher productivity rate than anticipated. Also the GSIM MBR approach yielded models that could be used for far more than simulation and training. The part of the Simulation Host Study reviewing present models and computational resources showed that present resources were adequate and current models could be upgraded to accomplish an adequate amount of the training mission. Thus GSIM was no longer funded under the Simulation Host Study.

Lockheed determined that the GSIM technology was a vital approach and began company sponsored IRAD funding in 1992 and continuing at least through 1993. The purpose of the IRAD is to spin off a production tool development task from the Simulation Host Study concept work done on GSIM. In 1992 the Model and Class Development Environments will be developed into production tools. Lockheed believes that a key to the success thus far has been to keep focused on the final delivery objective rather than just the research. This

helps ensure of anproduct. In keepingLockheed will be a ofdesign engineers in. theGSIM developers. The "wiH,review .andaccept incremental stages of GSIMdevelopment.. Further diewill apply GSIM to a diffindtproject actually 'being: bull: by thegroup. In, this way we can deferable 'fieeffectiveness of 'flic allow itsdirection, and, complexion to beits ultimate users. 'Thiswill result instatistics as m the moof gaining user tike 1stages.

CONCLUSION

• Lockheed in lie qtfffoach for toil and of launch at KSC, not only 'far ShoUte, tat: fair mat as

* ReCbcos of fame wade amttlli-iciisc anvoich such as a way to* mute edicienft and effective programs.

The pioneering of MBR las been acconqdisfaed flutoagjh die LBS and IM projects tod If Janueson :ltalisBWifti by Jack CUHher both of NASA. their work, fioDoflMNi MBR pnojeds such m

2-4?

GSIM: A Low Cost MBR Approach for KSC Design and Operations

GSIM would have nothing to follow on. The SCAN project was the first MBR project to make it to production status at KSC. SCAN was led by Bob Giffen, followed by the author, and presently Jim Tulley all of LSOC.

The Model-based Autonomous Test Engineer (MATE) project is the first known commercial application of MBR. Mate is the product of John Jamieson of CIT in Titusville, FL

John Jamieson of NASA, is also the co­ author of the GSIM concept and originator of the GSIM idea.

[WilkinsonSS] Wilkinson, J., Tulley, J., "Automated Analysis of Shuttle Wiring Using SCAN", Proc. 25th Space Congress, Cocoa Beach, Fl

BIBLIOGRAPHY

[JamiesonSS] Jamieson, J.R., Scarl, E.A., & Delaune, C.I., "A Knowledge Based Expert System for Propellant System Monitoring at the Kennedy Space Center", Proc. 22nd Space Congress, Cocoa Beach, H, pp. 1-9.

[Jamieson91] Jamieson, J.R., "Model Based Reasoning for Industrial Control and Diagnosis", International Control Engineering Exposition and Conference, Chicago, n, Feb 24-27 1992.

[New87] New, E., "Knowledge-based Control & Redundancy Management Techniques Used in NASA's Kate Project", Proc SOUTHCON 87, Orlando, Fl, March 87.

[Pepe90] C. O. Pepe & S.L. Fulton,"An Introduction to Model-based Reasoning", AI Expert, Jan. 1990.

[Scarl85] Scarl, E.A., Jamieson, J.R., Delaune, C.I., "Process Monitoring and Fault Location at the Kennedy Space Center", SIGART Newsletter, in press 1985.

[Scarl86] Scarl, E.A., Jamieson, J.R., Delaune, C.I., 'Sensor-Based Diagnosis Using Knowledge of Structure and Function", IEEE Transactions on Systems, Man and Cybernetics, Vol SMC-16, no 6, Nov. 1986.

[Webster87] "Webster's Ninth New Collegiate Dictionary", Merriam-Webster Inc., Springfield, MA, 1987.

2-48