A New Fault Diagnosis Model of Flywheel System based on Belief Rule Base and Fuzzy Fault Tree Analysis

Abstract

The fault diagnosis of the flywheel system is of great significance to ensure the normal operation of spacecraft. The current flywheel fault diagnosis technology still lacks effective solutions to the problems of lack of fault samples and system uncertainty resulting in low model accuracy. Belief Rule Base (BRB) is used as a semi-quantitative modeling method for complex systems, which has good processing performance for small sample analysis and uncertain information in the system. However, traditional BRB cannot solve the problem of lack of fault knowledge of the flywheel system and the complicated operating environment, which leads to unclear fault diagnosis indicators and difficulty in establishing logical relationships. To tackle this problem, From the standpoint of mechanism analysis, Fuzzy Fault Tree Analysis (FFTA) can help BRB improve its knowledge base. Therefore, based on FFTA and BRB, a fault diagnosis model for flywheel systems is proposed in this paper. The Bayesian network is employed as the conversion mechanism in the model, and FFTA transformation rules matching to BRB are presented, as well as BRB belief rules. In addition, to to decrease the model’s influence of uncertainty the of uncertainty in the model, an optimized model based on Projection Covariance Matrix Adaptation Evolutionary Strategies (P-CMA-ES) is established for the developed flywheel system fault diagnosis model. By raising the friction torque of the flywheel system, The effectiveness of the fault diagnosis model of the flywheel system is verified.