Identification of Transient Power Quality Disturbances Based on S-transform Feature Extraction and Random Forest Classification

Abstract

AbstractAiming at the problem of cross coupling, complexity and difficulty in classification and identification of transient power quality composite disturbances, an accurate identification method based on S-transform feature extraction and random forest classification is proposed. Firstly, the complex time-frequency matrix of the disturbance signals is extracted by S-transform, and the corresponding 11 characteristic curves are obtained, and the feature set of the disturbance signals is constructed from five aspects as the input of the random forest classifier. Then the random forest classification model is constructed. 70% of the original data is taken as the training set and 30% as the test set. The important features are selected in the original feature set, and the number of base classifiers is determined to optimize the random forest model and train the model. The test data verifies the accuracy of the model. Simulation results in different noise environments verify the accuracy and robustness of the method.KeywordsPower qualityTransient compound disturbanceS transformRandom forest