Research And Implementation Of Non-Intrusive Load Monitoring Method Based On Minute Meter Data

Under the background of vigorously developing global energy Internet,China is orderly promoting the construction of smart power grid.Realizing two-way interactive smart electricity consumption has become the main development goal of power system,and the key to carry out smart electricity consumption is to obtain household electricity load information and conduct intelligent analysis.Since non-intrusive load monitoring does not need to enter households,load monitoring can be realized only by obtaining the total electricity load information of users and conducting intelligent analysis,which has become the mainstream direction of load monitoring research.However,the existing non-intrusive load monitoring methods need to be achieved through transformation or new monitoring devices,which still face problems such as high investment,high cost,wide range of involvement and difficult popularization.Therefore,this thesis makes full use of existing electric meter and data acquisition environment and introduces the user community characteristics,research minute load monitoring characteristics of multi-dimensional data acquisition and processing methods,put forward the non-intrusive load monitoring method based on multidimensional characteristics,realize the rapid and exact residential electricity equipment load decomposition and identification,so as to support the application of energy use strategy optimization,detection of illegal appliances and expand the non-metering function of residential smart meters.The main work of the thesis includes:(1)A multi-dimensional characteristics data collection and processing method for minute-level load monitoring is proposed.The method makes full use of existing residents in electric meter and data acquisition environment,realize the transformation and upgrade of the software for the power load characteristics of minute level meter data collection,using some filling strategies and standardizing the power load data.On this basis,the social characteristics of residential users are introduced to make up for the lack of capture of some residential electricity load characteristics caused by the data collection of minute-level electricity meters,and the collection and processing methods of residential user social characteristics are given.(2)A non-intrusive load monitoring method based on multi-dimensional features is proposed.The method respectively based on residential electricity load characteristics and attention mechanism building load features regression subnet,based on the residential electricity load switch state classification subnet,building electrical equipment based on residential users social features to build social regression subnet.After that,the social features and electric switch state are fused.Then,through the gating mechanism and the load characteristics of fusion,this model output the result of load decomposition,so as to improve the accuracy of minute-level non-invasive load decomposition and identification.(3)The non-intrusive load monitoring prototype system was designed and implemented,including data management module,model management module and online identification module,and part of the system interface was displayed.The prototype system displays the collected data in a visual way,and the model configuration parameters can be manually set,and the decomposition results are displayed on the page,which improves the usability of the non-invasive load monitoring model.