Classification and Identification of Voltage Sag Sensitive Consumers Based on Multivariate Features

Abstract

Non-intrusive identification has the advantages of not interfering with consumer privacy and low hardware cost. It has excellent performance in the analysis of consumer monitoring data. The distribution of sensitive equipment can be identified based on the monitoring data to understand the specific needs of consumers for power quality, and provide data support for power supply companies to formulate service strategies. Considering the influence of feature coincidence on load identification, the difference in voltage sag tolerance characteristics of equipment was introduced, and a load decomposition model was constructed by combining multiple features. The non-intrusive load identification is realized by the memory simulated annealing algorithm, and the acceptance probability of the identification result is given based on the probability of equipment state change and the separation of characteristic coincidence loads.