Research On Non-intrusive Load Monitoring Technology

With the release of national low-carbon plans such as "carbon neutralization" and "carbon peak",people pay more and more attention to the carbon footprint.By analyzing the load data,the non-invasive load monitoring technology realizes the identification of the electrical equipment connected to and away from the system,so as to obtain the energy consumption list of the electrical equipment and realize the visualization of the carbon footprint of the electrical equipment.In this paper,a non-intrusive load monitoring method is proposed and a set of non-intrusive load monitoring system is designed through four steps: load data collection,load event detection,load feature extraction and load identification.The non-invasive load monitoring method consists of three parts: load event detection,load feature extraction and load identification,which are progressive.Load event detection is used to detect the switching of electrical equipment.Load feature extraction is used to extract load imprint.Load identification identifies the type and status of equipment that is disconnected from the system through load imprinting and labeling.By analyzing the advantages and disadvantages of the existing non-invasive load monitoring methods,this paper presents a non-invasive load monitoring method.Among them,load event detection is implemented using HHO optimized improved sliding window mean-difference accumulation and algorithm.This method can effectively improve the error detection and miss detection problems in the process of event detection,and improve the accuracy rate of load event detection.Load feature extraction uses CEEMDAN+HT transformation method to process current data by CEEMDAN and HT transformation,and extracts instantaneous frequency,instantaneous amplitude and instantaneous phase as load imprints.This method can effectively solve the mode overlap problem in the process of load feature extraction.Load identification is implemented using ISMA-GRNN algorithm,which can effectively improve the accuracy of load identification.The non-invasive load monitoring system consists of a load data collection subsystem and a load identification subsystem.Load data collection is the precondition of non-invasive load monitoring technology.After the research and analysis of data collection function,this paper uses Art USB3106 A data collection card with LabVIEW to achieve load data collection.A load identification subsystem is constructed by using the compatibility of LabVIEW’s Matrix node module with MATLAB’s M language and the non-invasive load monitoring method verified by MATLAB.The load data collected by the load data collection subsystem is processed and input into the load identification subsystem for load identification.Finally,the results are output through the front panel of LabVIEW to complete the validation of the non-invasive load monitoring method proposed in this paper.