Fault-Consequence Digraph를 이용한 이상진단 전문가시스템의 지식 기반 구현

윤병석; 오전근; 윤인섭

HWAHAK KONGHAK, Vol.29, No.1, 19-32, February, 1991

Export Citation

Fault-Consequence Digraph를 이용한 이상진단 전문가시스템의 지식 기반 구현

Knowledge Base Representation for the Fault Diagnostic Expert System Using the Fault-Consequence Digraph

윤병석, 오전근, 윤인섭

초록

본 연구의 목적은 나프타 분해로의 실시간 이상진단 전문가시스템(Operation Aiding expert SYStem:OASYS)의 지식모델로서 이상전파 유향그래프(Fault-Consequence Digraph:FCD)를 사용하여 시스템을 구현하고, 이를 통하여 이 모델의 지식모델로서의 적합성을 살펴보는데 있다. FCD모델은 공정에서 이상이 발생했을 때, 그 원인으로부터 전체 공정으로 증상이 파급되어가는 증상상태(symptom pattern)를 나타내므로써 이상진단에 효과적으로 이용될 수 있다. 시스템의 구현시 하드웨어로는 SUN3/260, 소프트웨어로는 intelliCorp상의 KEE를 사용하였다. OASYS 시스템에서는 226개의 증상과 314개의 이상원인 후보가 이용되었다. 구현된 시스템을 실제 사고사례와 가상사례의 데이터를 이용하여 약 20개의 사례연구를 한 결과, 이상발생 초기에 정확한 이상원인후보를 찾아내는데 성공하였고, 이로써 FCD 모델이 이상진단 전문가시스템의 지식모델로서 작성이 편리하고 진단효율이 높은 모델이라는 것이 검토되었다.

The objective of this study is to implement the real time fault diagnostic expert system for the naphtha furnace process and to prove the effectiveness of Fault-Consequence Digraph(FCD) as a knowledge model. The FCD represents a quite simple graphical knowledge model of a hypothetical fault candidate. Since each fault occurring in the chemical process plant possesses the distinctive symp-tom pattern, the generated symptom pattern of a fault from a FCD can be checked whether it really occurs or not. The SUN3/260 workstation and the AI tool, KEE of IntelliCorp have used in this study, which implements the FCD model. From qualitative reasoning in a naphtha furnace, 226 symptoms were generated and 314 fault-candidates were prepared. To implement this suggested FCD model, each FCD model has been tested by logical fault simulation. As a result of more than 20 case studies, the FCD model has been proven to be an appropriate and quite effective tool for the fault diagnostic expert system.

[References]

Charniak E, McDermott D, "Introduction to Artificial Intelligence," Addison-Wesley, Inc. (1985)
Han JH, Yoon ES, A Study on the Knowledge-Based Expert System for Cement Process Control and Diagnosis, M.S. Thesis, Seoul National University (1986)
Hoskins JC, Himmelbrau DM, Comput. Chem. Eng., 12, 881 (1988)
Iri M, Aoki K, O'Shima E, Matsuyama H, Comput. Chem. Eng., 3, 489 (1979)
James JMA, Scheeline HW, "Ethylene," Stanford Research Institute (1978)
Oh JK, A Real-Time Operation Aiding Expert System of Naphtha Furnace, AIChE Annual Meeting (1989)
"KEE User's Guide Manual," Intellocorp, Inc. (1988)
Kramer MA, Palowitch BL, AIChE J., 33, 1067 (1987)
Rich SH, Venkatasubramanian V, Comput. Chem. Eng., 11, 111 (1987)
Rich SH, Venkatasubramanian V, AIChE J., 35, 943 (1989)
Shum SK, Davis JF, Punch WF, Chandrasekaran B, Comput. Chem. Eng., 12, 27 (1988)
Venkatasubramanian V, Rich SH, Comput. Chem. Eng., 12, 903 (1988)

[Referenced By]

[1] Charniak E, McDermott D, "Introduction to Artificial Intelligence," Addison-Wesley, Inc. (1985)

[2] Han JH, Yoon ES, A Study on the Knowledge-Based Expert System for Cement Process Control and Diagnosis, M.S. Thesis, Seoul National University (1986)

[3] Hoskins JC, Himmelbrau DM, Comput. Chem. Eng., 12, 881 (1988)

[4] Iri M, Aoki K, O'Shima E, Matsuyama H, Comput. Chem. Eng., 3, 489 (1979)

[5] James JMA, Scheeline HW, "Ethylene," Stanford Research Institute (1978)

[6] Oh JK, A Real-Time Operation Aiding Expert System of Naphtha Furnace, AIChE Annual Meeting (1989)

[7] "KEE User's Guide Manual," Intellocorp, Inc. (1988)

[8] Kramer MA, Palowitch BL, AIChE J., 33, 1067 (1987)

[9] Rich SH, Venkatasubramanian V, Comput. Chem. Eng., 11, 111 (1987)

[10] Rich SH, Venkatasubramanian V, AIChE J., 35, 943 (1989)

[11] Shum SK, Davis JF, Punch WF, Chandrasekaran B, Comput. Chem. Eng., 12, 27 (1988)

[12] Venkatasubramanian V, Rich SH, Comput. Chem. Eng., 12, 903 (1988)