A Recovery Algorithm of Power Quality Big Data Based on Improved Differential Kriging

Abstract

The integrity of power quality big data directly affects the state sensing accuracy and operation safety of the power system. Therefore, the recovery algorithm of power quality big data based on improved differential Kolding is studied to improve the big data recovery effect. The trend turning point is used to divide the time series of power quality big data, and the characteristic matrix of time series is constructed. The recovery model of power quality big data is built according to the characteristic matrix. By improving the differential Kriging solution model, the estimated value of the data to be recovered is obtained and the big data recovery is completed. Experimental results show that the convergence speed is the fastest when the initial scaling factor is 0.3. The algorithm can effectively recover the big data of random missing and continuous missing. In different fault recovery scenarios, the signal-to-noise ratio (SNR) is high, the structure similarity value is high, the data recovery accuracy is accurate, and the integrity of the restored data is better.