A new automated interval structure belief rule base-based fault diagnosis method for complex systems

The fault diagnosis of complex systems is highly important for ensuring the safe operation and production of large-scale equipment. The belief rule base (BRB) is a rule-based expert system that can address nonlinear causal relationships and uncertain information and has been widely used in fault diagnosis. However, the difficulty in accessing expert knowledge due to system complexity and environmental variability makes deriving the BRB structure and parameters a major challenge. Therefore, a fault diagnosis method for complex systems based on the automatic construction of an interval belief rule base structure (A-IBRB) is proposed. Firstly, in the initialization phase of the model, a reference interval is constructed using the error-constrained k-means++ algorithm, and the belief degree in the diagnostic results are initialized using the Gaussian interval method (GIBM). Secondly,the selection strategy of key points in the interval is introduced, enabling the rules to be constructed adaptively based on data samples. Furthermore, the model is derived using evidential inference (ER) algorithm. Finally, a projected covariance matrix adaptive evolutionary strategy (P-CMA-ES) algorithm with correlation coefficient and step size adjustment is designed to optimize the model. A case study of gearbox fault detection and bearing fault diagnosis is used to verify the validity and reliability of the proposed model.