Machine learning-based approaches for fault diagnosis on transmission lines have attracted increasing attention in recent years, yet concerns over their robustness and explainability hamper their practical applications. Moreover, separate algorithms are normally used to solve the fault classification and location tasks. In this work, we focus on these tasks and propose an integrated model based on the convolutional neural network, whose improved performance over separate models is demonstrated by numerical results. Besides, explainability of the proposed model is demonstrated and analyzed. Specifically, the class activation maps and attention maps illustrate that the proposed structure can reduce the degradation of the model performance due to data pollution, enhancing the credibility of the proposed method in practical applications. Moreover, the proposed model outperforms existing machine learning-based models in terms of accuracy and robustness. Such a structure has been proven effective on field data and can be applied to many other tasks of fault diagnosis.

This paper develops a novel graph convolutional network (GCN) framework for fault location in power distribution networks. The proposed approach integrates multiple measurements at different buses while taking system topology into account. The effectiveness of the GCN model is corroborated by the IEEE 123 bus benchmark system. Simulation results show that the GCN model significantly outperforms other widely-used machine learning schemes with very high fault location accuracy. In addition, the proposed approach is robust to measurement noise and data loss errors. Data visualization results of two competing neural networks are presented to explore the mechanism of GCNs superior performance. A data augmentation procedure is proposed to increase the robustness of the model under various levels of noise and data loss errors. Further experiments show that the model can adapt to topology changes of distribution networks and perform well with a limited number of measured buses.