Qualitative bond graph approach to intelligent supervisory coordinator

xvi, 165 leaves : ill. ; 30 cm.PolyU Library Call No.: [THS] LG51 .H577P EE 2003 LoIn recent years, the increasing complexity of process plants and other engineered systems has extended the scope of interest in control engineering that was previously focused on the development of controllers for specified performance criteria such as stability and precision. Modern industrial systems require a higher demand of system reliability, safety and low-cost operation which in-turn call for sophisticated and elegant fault detection and isolation algorithms. In this thesis, three basic aspects of the Intelligent Supervisory Coordinator (ISC) for fault detection and localization have been studied. Among the problems addressed in the thesis are: (a) Hybrid simulation; (b) Automatic fault detection; and (c) Qualitative model-based fault diagnosis. Hybrid simulation is a novel simulation technique for dynamic systems that utilizes both qualitative and quantitative knowledge. The automatic fault detection deals with the on-line system monitoring method for system performance classification. Qualitative model-based fault diagnosis is a qualitative method for localizing faulty components in process plants. Based on these studies, an integrated real-time ISC that assist human operators to manage process plants is developed. The studies are accomplished in three phases. In the first phase, a novel hybrid simulation technique is proposed, that alleviates the difficulty in establishing precise mathematical representations of process plants. Dynamic system behaviors are predicted through this hybrid simulation technique. Qualitative representation of bond graph model is adopted to model a dynamic system. System measurements are represented by real numbers rather than qualitative values to improve accuracy. The integration of qualitative and quantitative information enhances the accuracy and effectiveness of qualitative simulation, and at the same time reduces the need for a precise mathematical model. The effectiveness of the proposed hybrid simulation approach is demonstrated by simulation studies of both linear and non-linear systems. In the second phase, the tasks of automatic fault detection and diagnosis are addressed. Fuzzy-genetic algorithm (FGA) is proposed to effect automatic fault detection. The automatic fault detection system (AFD) monitors the system states continuously by fuzzy logic. The optimization capability of genetic algorithms allows the generation of optimal fuzzy rules. System behaviors are represented as four states: normal, malfunction, load disturbance and faulty, and are distinguished by fuzzy logic after tuning its rule table. When a faulty behavior is detected, the AFD triggers the fault diagnosis algorithm. With the previously derived qualitative bond graph model of the system, Genetic Algorithm (GA) is then proposed to search for possible fault components among the system. The proposed fault diagnosis algorithm is tested on an in-house designed and built floating disc experimental set-up. In phase three, the development of the integrated real-time ISC will be discussed and applied to the servo-tank liquid system. The ISC integrates artificial intelligence techniques, like, fuzzy logic and GA; with control engineering in order to perform system simulation, fault detection and diagnosis.Department of Electrical EngineeringPh.D., Dept. of Electrical Engineering, The Hong Kong Polytechnic University, 2003