Big data clustering algorithm of power system user load characteristics based on K-means and SOM neural network

Big data clustering algorithm of power system user load characteristics based on K-means and SOM neural network

연구 분야: Databases

학회: Multimedia Tools and Applications

Power system user load characteristic big data has serious dynamic evaluation to power grid. In order to improve the clustering ability of power system user load characteristic big data, a fusion clustering method of power system user load characteristic big data based on K-means and SOM neural network is proposed. K-means clustering model is used to carry out distributed reorganization and integrated operation of power system user load characteristic big data, and cloud grid distribution model of power system user load characteristic big data is constructed. Principal component analysis method is used to cluster power system user load characteristic big data in irregular block distribution areas, and common mode component calculation of power system user load characteristic big data is realized in regional graded power grid. SOM neural network is used to realize convergence judgment and optimization control in the process of power system user load characteristic big data clustering, to avoid the clustering center falling into local optimum, to extract the energy spectrum characteristic quantity of power system user load characteristic big data, and to realize characteristic clustering of power system user load characteristic big data according to parameters such as power system user load characteristic, power flow distribution and capacity, etc. according to K-means clustering distribution and SOM neural network training results. The simulation results show that this method has good automaticity and high clustering accuracy in clustering large data of power system users' load characteristics.