Protection and maintenance systems are necessary for transmission systems to ensure an efficient and reliable power supply when faults occur. However, most fault detection and location methods rely on the electricity measurement provided by current and voltage transformers. In addition, the initiative repair of a power distribution network is a very considerable portion of power grid operation, and the initiative repair efficiency is crucial for power supply enterprises to provide high-quality services. To give full attention to the value of perceived data and improve the accuracy of fault diagnosis, the technical performance of high-speed power line carriers is extensively investigated, and intelligent sensing devices are installed in power distribution network lines, stations, branch boxes, table boxes, and other parts to realize the initiative repair. To improve the accuracy and timeliness of fault diagnosis for the power distribution network, in this paper, we propose an initiative repair and judgment mechanism based on backpropagation (BP) neural network optimization for the power distribution network. After acquiring information data such as power consumption, the mechanism first takes advantage of the filtered data information to train the BP neural network to predict fault information and other data. Finally, the initiative repair of the power distribution network obtains the forecast data of the BP neural network to judge the fault of the power distribution network. The proposed mechanism considerably shortens the time for the initiative repair and improves the traditional repair mode of the power distribution network, thereby improving the initiative repair efficiency of the power distribution network.