Non-intrusive Load Identification Technology For Residential Users Based On Signal Decomposition

In smart grid,based on the specific information of consumers' energy consumption,power companies provide users with demand response schemes to optimize the operation of distribution system,so as to improve the efficiency of energy consumption.As an important part of power load,residential users' power consumption structure and details are of great significance to promote power consumption and improve the reliability of power grid.It is necessary to obtain energy consumption of various electrical equipment in residential buildings through load monitoring.Non-intrusive load monitoring can identify the energy consumption of a single electrical appliance by installing current and voltage sensors at the power supply entrance and analyzing the measured electrical signals.Current non-intrusive load identification methods mainly rely on extracting and identifying the symbolic characteristics of power load,but lack of separation methods for complete load signals,which leads to incomplete power information.Based on high-frequency data acquisition,this thesis studies the decomposition and recognition of residential users'electricity signals from the perspective of signal decomposition,and obtains load waveforms containing detailed information.This thesis mainly includes the following contents:(1)The research on load identification at home and abroad is summarized and analyzed,and the non-intrusive load identification framework based on high-frequency acquisition structure is studied.In this thesis,load current changes are used to detect the switching events of power equipment.At the same time,combined with the consumer's electricity habits,the problem of multi-dimensional underorder current signal decomposition is optimized to one-dimensional underdetermined problem,and load decomposition and identification methods based on current signal is formed.(2)According to the sparsity of load current signal in frequency domain,the objective function of underdetermined decomposition is established,and the two-step iterative shrinkage threshold algorithm is used to obtain the decomposition signal of switching load,so as to realize load decomposition.By utilizing the characteristic current generated when each load is running alone,a characteristic filter bank is established to filter the decomposition current,and the effective load identification is realized.The measured data verify that the algorithm can obtain the current signal of a single load and accurately judge the load state.(3)By detecting the fixed voltage range before the switching event,the current signal at the corresponding position is extracted,and the difference of the mixed current between this event and the previous event is calculated,so that the load current signal of the last switching event can be obtained.The objective function is constrained according to the user's electricity consumption habits,and the scope of solution is narrowed.The operation state of each load can be determined by optimizing the solution.The example analysis shows that the algorithm can accurately decompose and identify the load.