An uncertain information fusion method for fault diagnosis of complex system
Huawei Wang, Jing-Lun Zhou, Zu-Yu He, Sha Ji-chang · 2004
Fault diagnosis is becoming extremely important for safety and high reliability of complex systems. But the fault diagnosis for complex system is the decision with uncertainty under small sample. The characteristics of complex system fault diagnosis require utilizing all kinds of information adequately. BN provides a flexible means of representing and reasoning with probabilistic information. Uncertainty and dependences are easily incorporated in the analysis. In the article, the application of Bayesian networks (BN) for monitoring and diagnosis of complex system is described. Furthermore, we propose leaky noisy-OR model to reduce the data requirements in BN inference. The advantages of BN model for complex system fault diagnosis are demonstrated through example.