Non-intrusive Load Event Detection and Identification Algorithm for Concealed Power Consumption of Railway Signal Power System

Railway signal power system supplies power to all kinds of signal equipment, which belongs to the first level power load and requires high reliability and continuity of power supply. In view of the potential hidden power consumption problems, non-intrusive load monitoring (NILM) can realize the identification of illegal electrical appliances. This paper uses an improved bilateral cumulative sum (CUSUM) event detection algorithm based on sliding window, In variable length transient event detection and recognition window Wdr the minimum interval variance algorithm is used to determine the start and end points of the event, and the method of determining the threshold is improved, the multi start process of the event is considered as well. Then SVM model after second order particle swarm optimization is used to identify the load of switching events. Finally, compared with other event detection and load identification algorithms, this algorithm has higher detection and identification accuracy, stronger stability, and is more suitable for railway concealed power monitoring.