A Universal Fault Diagnostic Expert System Based on Bayesian Network

Fault diagnosis is an area of great concern of any industry to reduce maintenance cost and increase profitability in the mean time. But most of the researches tend to rely on sensor data and equipment structure, which are expensive because each category of equipment differs from the others. Thus developing a universal system remains a key challenge to be solved. A universal expert system is developed in this paper making full use of expertspsila knowledge to diagnose the possible root causes and the corresponding probabilities for maintenance decision making support. Bayesian network was chosen as the inference engine of the system through raw data analysis. Improved causal relationship questionnaire and probability scale method were applied to construct the Bayesian network. The system has been applied to the production line of a chipset factory and the results show that the system can support decision making for fault diagnosis promptly and correctly.