Classification of Power Quality Disturbance Using Segmented and Modified S-Transform and DCNN-MSVM Hybrid Model

In this paper, a novel approach to classify the signals of power quality (PQ) disturbance is proposed based on segmented and modified S-transform (SMST), deep convolutional neural network (DCNN), and multiclass support vector machine (MSVM). The idea of frequency segmentation with different adjustable parameters was used in the Gaussian window function. The accurate time-frequency localization and efficient feature extraction of different PQ disturbances then could be achieved. Firstly, the SMST was used to analyze the PQ disturbance signals and obtained two-dimensional (2D) contour maps with high time-frequency resolution. Then, the DCNN was employed to automatically extract features from the 2D contour maps. Finally, the MSVM classifier was developed for the classification of single and complex signals of PQ disturbance. In order to demonstrate the effectiveness and robustness of the proposed model, eight single and thirteen complex waveforms of PQ disturbances were considered without noise and with different noise level, respectively. Extensive simulations were performed and compared to other existing methods. The simulation results show that the proposed method has better performance than several state-of-the-art algorithms in classifying PQ disturbances under different noise level.

Automated summary

In this paper, a novel approach using SMST, DCNN, and MSVM is proposed for classifying PQ disturbance signals. The Gaussian window function with adjustable parameters is used for frequency segmentation. This approach achieves accurate time-frequency localization and efficient feature extraction. SMST analyzes the signals and generates 2D contour maps, which are then processed by DCNN for feature extraction. Finally, MSVM is used for the classification of PQ disturbance signals. Extensive simulations show that the proposed method outperforms several state-of-the-art algorithms in classifying PQ disturbances under different noise levels.