Non-invasive Monitoring And Identification Of Loads Switching Behavior

The traditional load monitoring system is also called intrusive load monitoring,which requires the installation of sensors and other hardware equipment at each load monitored.The electrical quantity collected by every monitoring device needs to be processed and analyzed by transmission line or wireless transmission to the central centralized control unit.The efficiency of the work is very low and can not meet the needs of the development of power system.At present,non-intrusive load monitoring(NILM)has been widely studied,which needs to install monitoring equipment only in the power service entrance(ESE)of substation,residential or commercial buildings.Through decomposition the voltage and current signal of ESE,the type and operation situation of the individual load can be obtained,which greatly saves the cost of hardware,installation and maintenance.It facilitates the monitoring of load and improves the economic benefits.In this paper,the load model is built to simulate the actual electrical behavior based on Matlab/Simulink platform.When the load switches,the load electric quantity will change suddenly.Then the unknown events are separated and the steady-state power,current harmonic and instantaneous power spectrum energy characteristics are extracted and establish the feature database.In order to solve the problem that it is difficult to detect and separate the load change information during load switching,a detection algorithm based on the change of current intensity and steady power is proposed in this paper,and the transient power detection algorithm is started at the same time.In this algorithm,the new electrical information is subtracted from the electrical quantity before the transient process starts,and the influence of background load is eliminated,then the information of switching load is obtained.The method of feature extraction is used to construct the load feature database,and the separated load information is decomposed to obtain the steady state and transient characteristics of the total load.An improved binary particle swarm optimization algorithm for discrete variables is proposed,which optimizes the matching between fitted data and load signals,and determines the optimal load switching state.In this paper,different load switching behavior of a large number of simulation results verify the effective of the monitoring of various load switching signal separation,feature extraction and identification methods.It provides a strong support for understanding the user's electricity behavior,making energy saving plan and realizing the intelligent power consumption.