Fault diagnosis method for substations based on fault enumeration tree to generate fuzzy Petri net

To address the issues of complexity and difficulty of automatic modeling in Petri net fault diagnosis models that incorporate temporal information, a fault diagnosis method for substations based on generating fuzzy Petri nets from fault enumeration trees is proposed. To comprehensively traverse all fault combination scenarios in the substation system, a fault enumeration tree traversal method based on Breadth-First Search (BFS) was developed, enabling the rapid simulation of primary and secondary fault combinations in substations. Based on the simulation results, a fuzzy temporal Petri net fault diagnosis model for substations was established, and parallel inference was employed for fast fault diagnosis. Primary and secondary interconnected simulation for a typical 110 kV substation was tested on the CloudPSS cloud simulation platform. The test results indicated that the proposed method incorporated the action time limit information of protection devices and circuit breakers into the fault inference process. By comparing the timestamps for action time limits and alarm information, alongside matrix operations in the classical Petri net model, the method demonstrated strong diagnostic performance, even in the case of distorted alarm information. This approach could be applied to more complex secondary protection systems while ensuring fault diagnosis accuracy. The parallel hierarchical matrix inference algorithm improved the inference efficiency of the diagnostic model, providing significant practical value. Additionally, the inference process included a confidence correction step, which more thoroughly considered the impact of protection failures, false trips, and misreported or omitted alarm information on fault diagnosis, improving fault tolerance in complex fault scenarios.