Recognition of Voltage Sag Sources Based on Phase Space Reconstruction and Improved VGG Transfer Learning

Abstract

The recognition of the voltage sag sources is the basis for formulating a voltage sag governance plan and clarifying the responsibility for the accident. Aiming at the recognition problem of voltage sag sources, a recognition method of voltage sag sources based on phase space reconstruction and improved Visual Geometry Group (VGG) transfer learning is proposed from the perspective of image classification. Firstly, phase space reconstruction technology is used to transform voltage sag signals, generate reconstruction images of voltage sag, and analyze the intuitive characteristics of different sag sources from reconstruction images. Secondly, combined with the attention mechanism, the standard VGG 16 model is improved to extract the features completely and prevent over-fitting. Finally, VGG transfer learning model uses the idea of transfer learning for training, which improves the efficiency of model training and the recognition accuracy of sag sources. The purpose of the training model is to minimize the cross entropy loss function. The simulation analysis verifies the effectiveness and superiority of the proposed method.

A systematic review of real-time detection and classification of power quality disturbances

“…A remarkable improvement provided by deep learning, and the main motivation to apply these types of algorithms to PQD detection and classification, is the ability of models to automatically extract the best set of features from raw data to conduct classification. Convolutional Neural Networks (CNN) have been widely used in multiple PQDs and voltage sag classification [103,111,116,119,121,125,130,131,158,164,167,171,179,183]. Other deep learning models have been particularly used in the classification of voltage sags according to the root causes, e.g., deep feedforward ANNs [124], Long Short-Term Memory (LSTM) [129,159,180] (2023) 8:3 [165,175], and independently recurrent neural networks [182].…”

“…The use of powerful image classification techniques after the 2D transformation of signals is also a promising field of research. This approach allows the use of tools from the ever-increasing potential of the image processing and classification field, e.g., the attention mechanism to improve classification accuracy and transfer learning to reuse pre-trained models [167]. Also, a 2D transformation of signals allows the use of deep learning tools such as CNN.…”

SOGI-FLL Grid Frequency Monitoring with an Error-Based Algorithm for a Better Response in Face of Voltage Sag and Swell Faults

“…DG can use different grid monitoring techniques for obtaining online estimates of the grid parameters and stay synchronized with the grid. However, the performance of these estimators is perturbed by faults in the grid, such as voltage sags and swells, which in turn implies a worse DG performance of DG when injecting power to the grid [5][6][7][8][9][10][11][12][13][14][15]. Therefore, during grid faults, a fast and accurate detection of the grid voltage parameters is essential to keep a high quality in the DG's operation [6].…”