TRANSMISSION LINE FAULT DETECTION AND CLASSIFICATION BASED ON A HYBRID DWT-GAF-ENSEMBLE CNN MODEL

Tóm tắt

This paper proposes a hybrid model combining Discrete Wavelet Transform, Gramian Angular Field, and an ensemble Convolutional Neural Network for fault detection and classification in transmission lines. Voltage and current waveforms from each phase are decomposed and denoised using Discrete Wavelet Transform, then transformed into time-series images using Gramian Angular Field for feature extraction. These images are input into an ensemble Convolutional Neural Network model consisting of six parallel Convolutional Neural Networks, each corresponding to voltage and current signals of the three phases. The outputs are then concatenated to produce the final fault type classification result. The proposed method is tested on a simulated 220 kV transmission line system with two generators under various fault types and locations. Simulation results demonstrate that the proposed model achieves superior performance compared to conventional Convolutional Neural Network, Gramian Angular Field – Long Short Term Memory, and Discrete Wavelet Transform – Support Vector Machine models. It consistently attains over 95% accuracy in fault detection and classification, while ensuring fast processing time.