Refined Identification of Distribution Network Planning Survey Based on Improved Convolutional Neural Network Algorithm

Distribution network planning is an important guarantee for power grid construction and transformation, which can ensure the reliable, stable, economic and flexible development of the system. With the increasing capacity of distribution network, the amount of data to be processed and analyzed in distribution network planning has increased dramatically. Especially for the high voltage, medium voltage and low voltage superior power supply, distribution network structure and operation status, the workload is huge and the complexity is high. If the planning process completely depends on the planners to analyze and calculate, it is easy to have calculation errors, incomplete analysis or other uncertainty errors. Based on the improved convolutional neural network (CNN) algorithm, a new identification algorithm is proposed to improve the measurement accuracy of distribution network planning. The proposed method uses CNN to extract the first k important variables, and then combines them with the previous two methods, one uses nonlinear regression, the other uses linear regression. In addition, we propose a new metric method. In order to evaluate our results, this article uses a large number of real data sets.