Intelligent tutoring system framework for operator training in process fault diagnosis
Abstract
Abnormal Situation Management (ASM) involves the timely detection, diagnosis and correction of abnormal process conditions. Industrial statistics estimate the economic impact due to abnormal situations to be about $20 billion/year in the U.S. petrochemical industries alone. Process fault diagnosis, which forms the first step in ASM, deals with detection and analysis of the root causes of abnormal behaviors. Although modern control systems achieve a high degree of automation, the process operator still has the overall responsibility for the safe and economic operation of the process, including process fault diagnosis. Diagnostic problem solving forms a major part of an operator's activities in complex chemical processes. It cannot be overemphasized that training operators on fault diagnosis and ASM will help minimize process upsets, and thereby improve the reliability and safety of the plant. However, current practices in operator training for ASM, in general, are quite inadequate in a number of aspects, thus necessitating the development for more effective alternative approaches. Towards that goal, a novel Process Operator Intelligent Training System (POINTS) framework has been presented in this thesis. This intelligent tutoring environment is composed of an interactive, dynamic simulation of a chemical process plant, a computer model of expert operators, an intelligent computer-based tutor and trainee model that allow for automatic evaluation and coaching of the trainees in order to improve their diagnostic skills, and suitable graphical user interfaces. Causal models of the process are represented as diagraph models. To make the training system more effective, the process-related information/explanations including digraph-based symptom-cause relationships are supplied through intelligent hypermedia interface. It helps operators to organize system knowledge and operational information including causal relationships. Specific training applications were developed for a CSTR case study and the AMOCO Model IV FCCU plant. These establish the feasibility of implementing integrated intelligent training programs as POINTS applications. This demonstrates the possibility of using general-purpose toolkit modules to design and configure flexible systems which are relatively easy to use and support. Possibilities of using this system as an on-line assistant of process operators and for the training of remedial actions are also discussed.
Degree
Ph.D.
Advisors
Venkatasubramanian, Purdue University.
Subject Area
Chemical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.