A Transmission Line Fault Diagnosis Model Based on Interpretable BRB With Power Set

It is crucial to diagnose transmission line faults quickly and accurately. Belief rule base (BRB) has strong nonlinear modeling ability, allowing for the effective utilization of both qualitative knowledge and quantitative data related to these faults. To ensure accurate diagnosis results, the model must account for uncertainty and ignorance. Additionally, interpretability of the results is essential for improving diagnostic credibility. Therefore, a transmission line fault diagnosis model based on interpretable BRB with power set (PBRB-I) is proposed in this paper. Firstly, transmission line faults and data are analyzed, with the use of the Spearman correlation coefficient for data preprocessing. Secondly, according to the characteristics of transmission lines, interpretable modeling criteria are defined. Then, a power set identification framework is utilized to represent ignorance. Finally, the evidence reasoning (ER) algorithm is applied as a reasoning tool, and a parameter optimization method with interpretable constraints based on the projection covariance matrix adaptive evolutionary strategy (P-CMA-ES) is proposed. In the case study, the PBRB-I model demonstrates an accuracy of 91.11%, and it exhibits high performance stability across different data allocation ratios. It not only shows outstanding accuracy but also effectively expresses ignorance and produces interpretable results.