Fault Analysis Method of Active Distribution Network Under Cloud Edge Architecture

Efficient fault treatment of active distribution network is an important guarantee to ensure the steadystate reliability of the system. In order to improve the accuracy of distribution network fault identification and analysis, a fault processing method based on deep learning is proposed in this paper. This method collects massive heterogeneous data sets using patrol robot to realize real-time perception and accurate acquisition of distribution network status. Relying on the processing mode of distribution network cloud edge collaboration, the principal component analysis method is used at the edge to effectively remove redundant data, providing a complete and reliable data support for the deep network model. Meanwhile, the attention mechanism is added to the cloud to improve the depth confidence network, further realizing the extraction of useful feature information for complex data sets and avoiding the interference of irrelevant information on the recognition results. The simulation experiment is based on the actual active distribution network model. The experimental results show that the fault identification accuracy Acc of the proposed method can reach 0.9255, indicating an excellent distribution network fault identification and analysis ability to support safe operation of active distribution network.