Fault diagnosis of nuclear facilities based on hidden Markov model

Fengwei Yuan, Qian Deng, Jiazhu Zou 

College of Mechanical Engineering, University of South China, Hengyang 421001, Hunan, China

Due to the complex structure of nuclear facilities in a high irradiation environment, people are hard to approach it, In view of these situation a fault diagnosis method based on HMM (Hidden Markov Model) of capturing the audio signal while facilities are operating is proposed. With the strong modelling ability, HMM can be applied to analysing such as audio signal non-stationary time signal. By using this method, the original mechanical structures of nuclear facilities are not destroyed. The proposed sensors were needed as few as possible by the whole diagnosis system and which has a simple structure, low cost structure, the fault diagnosis rate is high and so on. State monitoring and fault diagnosis system of complex  nuclear power equipment can timely and effective to provide running status and potential failure information for operating personnel, which has a vital significance for the safe and reliable operation of nuclear power equipment.