Application and Comparative Analysis of Traditional Machine Learning and Deep Learning in Transmission Line Fault Classification

With the continuous expansion of power scale, safe, reliable and stable power supply plays a vital role in social and economic development. The stability of power transmission system directly affects the security and stability of power grid. The stability of the operation state of the transmission line directly affects the security and stability of the power grid. Once the transmission line fails, it will endanger personal safety and cause significant economic losses. If some behaviors can be detected and classified in time, they can be repaired in time to prevent further deterioration. With the increase of computing power, deep learning has better performance than traditional machine learning in many aspects. however, the performance of deep learning in transmission line fault diagnosis and classification has not been effectively proved. Four classifiers based on traditional intelligent algorithms: traditional machine learning algorithm: SVM, adaptive boosting, decision tree (DT), Gaussian Process (GP) and four classifiers based on deep learning: Full Progressive Network (FCN), Deep Gated Current and Progressive Network Hybrid Model (GRU_FCN), Perception Time, ResCNN, are used respectively to identify transmission line fault diagnosis and fault classification. The performance of traditional intelligent algorithm and depth learning is obtained by comparing the results. It is found that the classification results of the classifier based on depth learning are more accurate. Applying the deep learning algorithm to the fault diagnosis of transmission line short circuit can improve the stability of transmission line operation.