Research on Improved Machine Learning for Distribution Network Fusion Control

Abstract

In order to improve the accuracy of test sample data of distribution communication network fusion control, an improved machine learning technology for distribution communication network fusion control was established. The distribution communication network port is identified based on the network protocol, and the number of neurons controlled by the fusion of the distribution communication network is determined according to the sample information sent by the port number. After the feature selection layer by layer, the traffic characteristics of the machine learning protocol are selected according to the neural network structure. Moreover, the Relief algorithm is used to calculate the actual number of machine learning protocols and realize the design of improved machine learning technology for the fusion control of power distribution and communication networks. Compared with other three traditional techniques, it is concluded that this technique has obvious advantages in sample processing accuracy.