SDG Multiple Fault Diagnosis by Fuzzy Logic and Real-Time Bidirectional Inference

Abstract

Significant research has been done in the past 30 years to use signed directed graph (SDG) for process fault diagnosis. However, multiple fault diagnosis is still a difficult problem because the number of combinations grows exponentially with the number of faults. The method by real-time inverse inference is suitable for multiple fault diagnosis. However, the choice of thresholds is made based on experience so improper thresholds may lead to missed or wrong diagnosis. In addition, the compensatory response and inverse response in the SDG model usually hamper inverse inference and it also leads to missed or wrong diagnosis. In this work, a SDG multiple fault diagnosis by fuzzy logic and real-time bidirectional inference is proposed. Fuzzy logic is used to determine the states of nodes in SDG model and a bidirectional inference strategy based on assumption and verification is used to overcome influence of compensatory response and inverse response. The poor resolution of SDG based fault diagnosis is overcome by arranging the causes in decreasing order according to the indexes calculated by fuzzy logic. The implementation of multiple fault diagnosis method is done using the integrated SDG modeling, inference and post-processing software platform. Its application is illustrated on an atmospheric distillation tower. The result shows this method provides fast, reliable and accurate multiple fault diagnosis.