Research on Clustering Algorithm of Load Decomposition Considering Harmonic Characteristics in Power Safety Monitoring

With the development of smart grid, power grid security monitoring is becoming more and more important. Non-intrusive load monitoring is helpful to understand the operating status of electrical equipment and is of great significance to the economic and safe operation of the power grid.This paper provide a non-intrusive load safety monitoring method for smart grid based on clustering algorithm. Based on improved generalized likelihood ratio detection, this paper introduces the voting window to establish an event detector model, and introduces event detection metrics to determine the value of related parameters and obtain the best event detector. In view of the difficulty in distinguishing electrical appliances with similar power in load decomposition, this paper uses DFT to extract the harmonic characteristics of bus current signals, and establishes a load feature library combined with active power. Then affinity propagation clustering algorithm is used to establish the load feature library to realize the load decomposition. Finally, the effectiveness of the proposed method is verified on REDD data sets.