Operation monitoring and decision-making assistance for flexible DC converter valve based on improved LightGBM

Abstract

Aiming at the problem that the fault diagnosis accuracy of flexible DC converter valve is not high, a fault diagnosis method of Flexible DC converter valve of PCA-LSSA-LightGBM is proposed. Using the fault feature library collected by the simulation system, the normalization and standardization processing are carried out first, and then the key fault features are extracted through principal component analysis to reduce the complexity of fault diagnosis. Combined with the operation mode of flexible DC converter valve, a lightweight gradient elevator model is established, and optimizing the parameters of the light gradient boosting machine are found by introducing Singer mapping and the sparrow search algorithm optimized by Levy flight strategy. Experiments show that the comprehensive accuracy rate of PCA-LSSA-LightGBM model diagnosis results is 96.09%, which is higher than that of other fault diagnosis methods, which verifies the correctness and effectiveness of the fault diagnosis of flexible DC converter valve applied in this method.