Data Mining-Based Distribution Network Fault Risk Level Prediction Method

With the growing popularity of distributed power sources and changes in power supply patterns, accurate risk prediction for distribution networks has become crucial for enhancing power supply reliability in power grids. This paper presents an approach to predicting the failure risk level of a distribution network using an adaptive simulated annealing improved particle swarm algorithm optimized SVM (ASAPSO-SVM) with hyperbolic tangent function controlling weight coefficients and linear iterative strategy controlling learning factors. The method is tested using data obtained from the distribution network information management subsystem, resulting in a high prediction accuracy of 98.43%, demonstrating the method’s effectiveness. Furthermore, the proposed ASAPSO-SVM algorithm is compared and analyzed with two other improved SVM algorithms to validate and verify its practical applicability in active distribution network risk prediction, thus providing reliable prediction support for the safe operation of active distribution networks.