Neural Network-Based Fault Detection Technology for Power Transmission and Distribution Lines

With the large-scale integration of distributed energy resources, the fault characteristics of power transmission and distribution lines have become increasingly complex, and traditional detection methods can no longer meet the requirements of high precision and rapid response. To address this, focusing on the characteristics of strong nonlinearity and obvious temporal features of line fault transient signals, this paper conducts research on fault detection based on neural networks. Firstly, it analyzes the voltage and current waveform characteristics of different fault types within short-term transients and constructs a multi-condition fault sample dataset. A CNN-LSTM model integrating one-dimensional convolutional neural network (1D-CNN) and long short-term memory (LSTM) network is designed to realize the coordinated output of fault type identification and fault distance estimation. Finally, the model performance is verified by PSCAD/EMTDC simulation. The results show that this method has obvious advantages in both identification accuracy and positioning precision.