Research And Application Of Non-intrusive Household Electrical Appliance Load Identification Method

With the development of the economy and society,the pressure on energy supply is increasing.To solve this problem,we can solve this problem through two aspects:increasing revenue and reducing expenditure.Among them,open source refers to the expansion of the power system,and throttling refers to the improvement of the use of electricity,such as power demand side management.At the same time,smart grid is the current development direction,and non-intrusive load identification is one of the key technologies of smart grid.In order to establish a two-way interaction between users and the power grid and achieve the goal of demand-side response,it is necessary to monitor the load of household appliances.Since non-intrusive load identification has the characteristics of simple installation and maintenance,low cost,high applicability,and strong privacy,it has good social and economic benefits,and has now become the focus of research in this field.This article first explains the background and significance of non-intrusive load identification research,and then introduces in detail the current domestic and foreign research status of this direction and the existing problems of existing methods,and then explains the technical route of non-intrusive load identification.Finally,in-depth research and application are carried out from three aspects of data collection,feature extraction and load identification in the key technology of non-intrusive load identification.This article takes the load of commonly used household electrical appliances as the research object,and simulates the actual usage of household electrical loads under daily conditions through the switching of various electrical appliances.Designed and developed a set of laboratory collection device with AD7606 sampling chip as the core to collect the voltage and total current data at the home on-site and generate the load characteristic library of common household electrical appliances.Starting from the analysis of the steady-state current of the household electrical load,a recognition method based on wavelet packet energy characteristics and improved BP neural network is proposed.The steady-state current signal is decomposed by wavelet packet,and the high-frequency and low-frequency characteristics of the signal are extracted at the same time to make full use of Based on the information contained in the steady-state current,the wavelet packet energy feature is introduced.Compared with the wavelet transform,the high-frequency feature in the collected signal is fully utilized to make the load feature more accurate.And use the improved BP neural network for training and testing.It is proved that this method has good advanced nature,reliability and stability.Starting from demand-side management,analyze and apply the collected information of a large number of residents.First,use fuzzy clustering algorithm(FCM)to perform cluster analysis on user non-intrusive load monitoring data,extract device usage characteristics,and then use decision tree model to realize the connection between user characteristics and device characteristics,and use depth-first search to make decision trees Pre-pruning,excluding decision tree nodes that do not meet the actual situation,and identifying typical users.The experimental results show that this method can more effectively reflect the demand response potential of various users,so as to provide more targeted guidance for power demand side management.From the perspective of engineering practicability and economy,a non-intrusive load identification engineering application scheme based on edge computing is proposed.This scheme is mainly composed of three parts:intelligent identification terminal,cloud server and main station software platform.The data processing unit is the core of edge computing.After the intelligent identification terminal completes data collection,measurement,and load identification,it establishes communication with the cloud server through the 4G module.The main station software platform Dian Dang Jia establishes a connection with the cloud server through the MQTT communication protocol,and can conduct more in-depth mining and application of the data on the cloud server.The main station software platform can be used to display load identification results,energy consumption information,safety warnings,etc.,and at the same time issue control commands to hardware nodes under certain specific conditions.