Fault detection and diagnosis for missing data systems with a three time-slice dynamic Bayesian network approach

Abstract

A multi-time-slice dynamic Bayesian network with a mixture of the Gaussian output (MT-DBNMG) based data-driven fault identification method is proposed to handle the missing data samples and the non-Gaussian process data. First, via introducing more time slices, a new dynamic Bayesian network structure with multi-time-slice is constructed which can describe the dependence between the current state and historic states. Second, a parameter learning strategy based on expectation maximization algorithm is deduced, from the complete historical data with the non-Gaussianity, to train the parameters of MT-DBNMG. Subsequently, for the missing measurements, an online non-imputation inference method for MT-DBNMG is proposed to conduct fault detection and identification. The effectiveness of the proposed approach is demonstrated by the continuous stirred tank reactor system and the Tennessee Eastman chemical process. The results show that the presented approach can accurately detect abnormal events, identify the fault, and is also robust to unknown noise.