Research on non-intrusive load event detection algorithm

Abstract

Event detection is an important step of non-intrusive load disaggregation. The accuracy of event detection directly affects the correct rate of load classification identification. Aiming at the shortcomings of existing event detection and feature extraction algorithms, using the steady-state active power as the feature, three improved event detection algorithms are proposed: improved CUSUM bilateral accumulation algorithm, improved sliding chi-square-GOF algorithm, and improved sliding Cepstrum analysis. The algorithm compares and analyzes the event detection accuracy of the three event detection methods. The experimental results show that the accuracy is better than most of the literature results, and the improved sliding chi-square-GOF algorithm can effectively detect large reference power events, avoiding missed detection of this situation.