Non-intrusive Load Monitoring based on Multiple Feature Optimization and Genetic Algorithm

Load monitoring is an important part of smart utilization. To address the problem of low accuracy of current non-intrusive load monitoring methods in identifying multi-state loads and loads with similar power, this paper proposes a multi-feature genetic optimization method considering state probability factors. The algorithm selects the active power and the amplitude of the third harmonic current as the research characteristics, and uses clustering by fast search and find of density peaks (CFSFDP) clustering algorithm to construct the load characteristic template. Based on the traditional genetic optimization objective algorithm, the state probability factor is added as an auxiliary feature to further improve the identification degree of similar loads. The performance of the algorithm is evaluated on The Reference Energy Disaggregation Data Set (REDD). The simulation results show that the above method can effectively improve the accuracy and outperform existing algorithms.