Optimal PMU Configuration in Distribution Networks Based on Fault Diagnosis Capacity

In the distribution network, the fault features are not obvious and easy to be affected by noise and transition resistance, making faulty line section detection very difficult. The two-way power flow caused by high proportion participation of renewable energy further threatens the accuracy of fault diagnosis system. A faulty line section detection method based on multi-source information (node unbalanced current and insufficient power flow) is proposed, the convolutional neural network is used because of its excellent generalization analysis ability, PMUs are optimally configured based on their fault diagnosis capabilities, a series of configuration methods are given for flexible choices. The method is based on clear physical characteristics, and simulations verify the effectiveness of faulty line section detection and the flexible configuration of measurement devices in the distribution network.