Reliability analysis of complex multistate systems based on an evidence-based discrete-time Bayesian network

This paper presents an approach to address the uncertainty of each unit’s state information caused by a lack of data, complex structure, and lack of human cognition level in complex multistate systems in practical engineering. The units in the system having the same operating environment and structure can cause common-cause failures. With the change in the working time of the system, the reliability of the system also changes, which is referred to as the dynamic timing problem. This paper first expands the discrete-time Bayesian network based on the Dempster-Shafer evidence theory. This forms the reliability analysis method of the discrete-time Bayesian network of evidence. Then, the β factor method is used to analyze and compare the reliability of system units belonging to multiple common-cause failure groups and each unit separately at different times. Finally, the accuracy and feasibility of the method are proved by the hydraulic system of the pipelayer. This paper solves the problem of analyzing the reliability of multistate systems with uncertain probability information when the failure units belong to multiple common-cause failure groups at different times. The proposed approach of integrating the system multistate, uncertainty of state probability information of each unit, failure correlation among units, and dynamic timing solves the problem of reliability analysis while avoiding large errors in actual engineering.