Fault Diagnosis Expert System of Semiconductor Manufacturing Equipment Using a Bayesian Network
Bo Li,Ting Han,Fuyong Kang
DOI: https://doi.org/10.1080/0951192x.2013.812803
IF: 4.1
2013-01-01
International Journal of Computer Integrated Manufacturing
Abstract:It is well known that fault diagnosis is very important to improve the availability of semiconductor manufacturing equipment. But how to acquire, represent and reason the diagnosis and maintenance knowledge is the key of a fault diagnosis expert system. According to the features of the knowledge source, production rule was chosen as the knowledge representing method. And Bayesian network (BN) with improved causal relationship questionnaire and probability scale methods was proposed as inference machine to diagnose the possible root causes, corresponding probabilities and suggested solutions. Based on above methods, a fault diagnosis expert system was proposed, whose overall structure and key technologies, including knowledge acquisition, representation and inference methods were presented in detail. Furthermore, this expert system was designed by using Unified Modelling Language (UML) method and developed with MS VS.NET and SQL Server 2000. Two cases in a chipset assembly and test factory showed the inferring process by BN and validated the inferring result of the expert system, which proves it accurate and believable.