Improved System Model for Fault Classification and Diagnosis of Power System Based on Artificial Intelligence and BP Neural Network and Its Application

Fault diagnosis of power system is a crucial part of power system operation and maintenance and safety management. To improve the accuracy and efficiency of fault diagnosis, a transmission lines fault diagnosis model integrating variational mode decomposition and back propagation neural network is designed based on artificial intelligence. The outcomes indicated that the feature extraction model based on variational mode decomposition could accurately distinguish different short-circuit fault features and determine the fault type according to the entropy value. Moreover, the feature extraction operation and the corresponding improvement measures effectively enhanced the performance of the diagnostic model. The fault diagnostic model designed in the study obtained the smallest values on mean absolute percentage error, root mean square error, and mean absolute error, and the lowest error was 0.08. The parameter optimization and updating measures improved the loss and F1 values of the back propagation neural network significantly. In the practical application analysis, the design of the study achieved better false alarm rate and false negative rate of 0.097 and 0.051, respectively. The diagnostic correctness of this model was better than other advanced models, and the average error rate was significantly different. The separation ability and computational efficiency for different faults took the values of 0.936 and 11.972s, respectively. This study helps to realize automated and intelligent power fault diagnosis, reduce the operation and maintenance cost, and further meet the needs of smart grid construction.