Full Parameter Optimized Belief Rule Base for Flywheel Health State Assessment With Interpretability

Abstract

The stable operation of the flywheel system is crucial for the on-orbit safety of astronautic engineering, so it is very important to conduct a flywheel system health state assessment. In the assessment modeling process, the model not only needs to address uncertainty to ensure the accuracy of the assessment results but also needs to have a transparent and reasonable assessment process and interpretable assessment results. Therefore, based on the in-depth study of the belief rule base (BRB) modeling method, this paper proposes the interpretability criterion of the optimized model and constructs a new BRB model for assessing the health state of the flywheel system with interpretability with full parameter optimization (IP-BRB). Firstly, based on the traditional interpretability criterion, the interpretability criterion of the optimization model is defined. Then, based on the projection covariance matrix adaptive evolution strategy (P-CMA-ES), the IP-BRB is proposed. Finally, the effectiveness of the model in the health state assessment of the flywheel system is verified by a case study of the assessment of bearing components in a flywheel system. The comparative results show that the IP-BRB model has certain advantages in terms of the accuracy of the assessment results and interpretability of the assessment process.

Keywords

• Expert knowledge
• belief rule base
• health state assessment
• interpretability
• parameter optimization