Cluster Analysis Method for Multi-dimensional Power Distribution Network System

Abstract

To improve the pertinence and effectiveness of equipment maintenance work and reasonably reduce the maintenance cost, it is urgent to carry out research on the online diagnosis information collection of key equipment for power distribution. Based on the distributed intelligent system, a smart distributing management system architecture is designed with comprehensive perception and software definition. A layered management and multidimensional display of the power distribution IoT system is studied, and product requirement in the power distribution IoT environment is analyzed. With edge computing, positioning function of product feature target is formulated. By combining demand analysis methods with modular architecture, models of product demanding are built. The primary clustering uses the Ward system clustering method to classify the load characteristics, the secondary clustering uses the fuzzy C-means method, then, the clustering center is provided by the primary Ward clustering results. The method proposed can avoid the sensitivity of the fuzzy C-means clustering method to the initial parameters, and can ensure a better clustering effect.