关键词: 故障树, 贝叶斯网络, 机器学习, 芯片烘箱

Abstract: In view of the difficulty of fault troubleshooting of the chip oven equipment, an improved Bayesian network fault diagnosis algorithm is proposed. The SECS/GEM communication protocol is used to obtain fault alarm data from the equipment side, and the fault tree of the chip oven equipment is established. The fault tree mapping is transformed into a Bayesian network, and the diagnostic decision tree that can be used to guide the maintenance work is obtained by reverse reasoning calculation of the Bayesian network. The historical fault sample data is substituted into the algorithm model for verification and analysis. The verification results show that the fault diagnosis according to the diagnostic decision tree can control the diagnosis error rate at about 5%, which meets the requirements of enterprise users and has high application value.

Key words: fault tree, Bayesian network, machine learning, chip oven