INTELLIGENT FAULT-DIAGNOSIS BASED ON WEIGHTED SYMPTOM TREE MODEL AND FAULT-PROPAGATION TRENDS

This paper presents a fault detection and diagnosis methodology based on the weighted symptom tree model and pattern matching between the co ming fault propagation trend and the simulated one. At the first step, backward chaining is used to find the possible cause candidates for t he faults. The weighted symptom tree model(WSTM) is used to generate t hese candidates. The weights are determined by dynamic simulations. Us ing WSTM, the methodology can generate the cause candidates and rank t hem according to the probability. At the next step, the fault propagat ion trends identified from the partial or complete sequence of measure ments are compared to the standard fault propagation trends which have been generated using dynamic simulation and stored a priori. A patter n matching algorithm based on a number of triangular episodes is used to effectively match those trends. The proposed methodology has been i llustrated using two case studies and showed satisfactory diagnostic r esolution.