Research on Target Prediction Method for Urban Distribution Network Reliability Planning

For the traditional distribution network reliability prediction methods need to rely on accurate distribution network structure and detailed historical data, it is difficult to adapt to the complex distribution network and planning reliability index prediction problems. For this reason, this paper proposes a BP neural network-based prediction method for urban grid reliability indicators under the condition of limited historical sample data. Firstly, we analyze and categorize the factors affecting the reliability index of urban distribution network, focusing on the configuration of primary and secondary systems, the impact of flexible loads on the reliability of the distribution network, and the differentiated needs of different types of power users; secondly, we elaborate the basic principles, relevant control parameters and training and learning methods of BP neural network, and select the ring network rate, cable rate, average power supply radius of the line, and the load transferability rate, The factors of ring rate, cableization rate, average power supply radius of lines, load transferable supply rate, and inter-station contact rate are chosen as the main influencing factors, and the historical data of these factors are taken as inputs, and the corresponding results of the distribution system reliability rate index are taken as outputs. Finally, the feasibility of the reliability index prediction method for urban distribution network planning based on BP neural network proposed in this paper is verified based on the historical sample data of distribution network in a city in southern China.