A Health State Assessment Method for Complex System Considering the Importance of Multi-Source Information

Health status evaluation of complex system involves multi-source heterogeneous data (MSHD) from multi-subsystem and multidisciplinary, MSHD fusion with high confidence is a key step of health status assessment. This paper proposes a health status assessment method by combining hierarchical Bayesian network and random forest algorithm considering the importance of MSHD. In the proposed method, the importance of the multi-source heterogeneous data is firstly analyzed based on deep neural network (DNN) and random forest algorithm. Then, a hierarchical Bayesian network based on the hierarchical structure of the system is established. Considering the uncertainties from data monitoring and performance degradation, the health status of each node of hierarchical Bayesian network is evaluated and updated based on Bayesian information updating technology. The validity of the proposed method is illustrated by the health status assessment of industrial robots. This method can improve the assessment accuracy and robustness, which provides a reference for maintenance decision making of complex system.