Data cleaning method of distributed photovoltaic power generation based on clustering algorithm

Abstract

Due to China’s current energy development strategy, distributed photovoltaic (PV) power generation shows continuous growth. Since PV equipment greatly impacts the distribution network, achieving accurate monitoring of operational data is of greater significance. This paper firstly adopts k-means based clustering algorithm to cluster data into more small clusters and make full use of the characteristics of time series to achieve outlier determination; secondly, it proposes an abnormal data classification processing method to improve data utilization and data cleaning accuracy; finally, it interpolates and fills the rejected abnormal values to ensure data integrity and further improve PV power generation data monitoring.