Research On Non-intrusive Residential Load Disaggregation In Energy Efficiency Monitoring Of Distribution Network

Monitoring and analysis of residential power loads are of great significance for saving energy,reducing energy consumption,peak load shifting,and alleviating social energy consumption in regional distribution network.Existing resident consumer power load monitoring studies can be divided into Intrusive and Non-Intrusive approaches from data acquisition.Intrusive Load Monitor(ILM)is a data acquisition method that install data collection devices for all users' power equipment,so as to collect detailed and precise load information of electrical equipment which can be used to analyze user's electricity consumption behavior.Non-intrusive load monitoring(NILM),also known as non-intrusive load disaggregation,can disaggregate the total electricity load into the load of electrical equipments by using only a single smart meter or other intelligent power collection equipment,some prior information(such as features and status data when each power device is working)and some algorithm models,so as to carry out the analysis and make decision.Compared with Intrusive mode,Non-Intrusive Load Monitor has the advantages of low cost,easy maintenance and simple device.Its disadvantage is the low disaggregation accuracy and the need of higher data sampling frequency.Considering the above problems,based on the spectral graph theory,this paper presents a method of non intrusive disaggregation of residential power load by using low frequency sampling data.Firstly,a graph structure is established using difference of adjacent sample signals from total load signals and the prior information of graph signal is defined by classifying difference of adjacent samples from appliance load signals.Secondly,unknown graph signals from appliance load signals are reconfigured based on Graph total variation(GTV)function obtained with graph Laplace transform.Fuzzy regularization method is used to figure out normalization problem in signal smoothing process.Finally,Interval time points between adjacent non-zero values in the reconfigured graph signals are assigned to corresponding state data of the appliance,then time series signals of each appliance are reconfigured.AMPds data set is used to carry out simulation.The proposed method has been tested in terms of sampling frequency,prior information,etc.At the same time,the contrast experiments are done with non-intrusive load disaggregation method based on factor hidden Markov model.The simulation results were analyzed using indicators such as the total load breakdown accuracy and the signal similarity of a single consumer device,etc.Result shows that the proposed method is effective and feasible.This method can achieve higher accuracy load decomposition under the condition of lower sampling frequency and less prior information.