A Parallelized Clustering Method for High-dimensional Power Multilevel Data Resources in load changes Mode

Conventional parallelized clustering methods for high-dimensional electric power multilevel data resources are based on a single data feature, which cannot meet the demand of parallelized clustering. Therefore, a parallelized clustering method for high-dimensional power multilevel data resources under load changes mode is designed. The sparse features of high-dimensional electric power multilevel data resources are extracted, and the high-dimensional electric power data are changed into low-dimensional observation data to reduce the data dimensions and solve the optimal sparse subspace, so as to satisfy the demand of parallel clustering of electric power resources. The parallelized clustering center of high-dimensional electric power multilevel data resources is established based on load changes, and the high-dimensional electric power multilevel data resources are optimized by parallel genetic algorithm, and whether the electric power data belongs to the value domain is detected, so as to realize the accurate clustering of data resources. Comparison experiments are used to verify that the method has better parallelized clustering effect and can be applied in real life.