Non-invasive load identification based on time partition and IACO-SVM

Abstract

Non-invasive load decomposition technology is an important part of smart grid technology system. Aiming at the problem that the identification accuracy of existing decomposition technology is low for loads with similar power or small power, a non-invasive load decomposition method based on Improved Ant Colony Optimize algorithm (IACO) -Support Vector Machine (SVM) is proposed. In this paper, a density-based clustering algorithm with noise is used to cluster the power characteristics of the load to obtain the power characteristic template, and the typical working time interval of the load is solved to obtain the time characteristic template of the load. The power and time characteristics are considered comprehensively. Then the parameters of support vector machine were obtained by cross validation method, and then the search space of ant colony algorithm was determined in these parameters. Finally, the optimal parameters were selected in this area, and the load decomposition model was established by using the optimal parameters. Simulation on AMPds2 public data set verifies the effectiveness of the proposed method.