Rule Reduction for Belief Rule Base Based on Redundancy

Rule reduction is one of the key focuses in the research of belief rule base (BRB) systems. Reducing its scale can save modeling and computational workload. In the presentation study, many rule reduction methods have been widely applied to BRB. But most methods only focus on unimportant rules with fewer activations, while ignoring some redundant rules that have similar knowledge to other rules. Therefore, this article proposes a rule reduction method for BRB(RR-BRB) based on redundancy. First, a rule redundancy calculation method is designed for the BRB model, and then the rules are reduced based on the size of redundancy. Second, for the problem of incomplete rule base after reduction, similar rules are used to replace the reduced rules for inference. Then, the phenomenon of breaking interpretability during the RR-BRB optimization process is constrained. Finally, the health assessment of B0006 lithium-ion battery and flywheel system bearing components is taken as an example to verify the effectiveness of the method.