Equipment Fault Diagnosis Based on State Space Model and Improved Hidden Semi-Markov Model

A method of equipment fault diagnosis based on State Space Model and Hidden Semi-Markov Model(HSMM) is proposed,it can make good use of the state date that generated till the present time.To the problem that exited in the ancient research,such as the state durations always submit to exponential distribution,the distribution of the state durations is hard to calculate based on the small sample,system,and state residual durations is always equated with the state durations,the HSMM in this paper is improved in the following two aspects.Firstly,the distribution of the state durations is continuous and parametric,and the prior distribution is setted to Weibull distribution;Secondly,the state residual durations is proposed based on the starting time of the state.The forecast probability based on improved HSMM is deducted,the formula of the fault diagnosis criteria using the state date that generated till the present time and the process of the application of this method are given in the paper.The rationality and the validity of this method was verified through an example.