Research On Non-intrusive Load Identification System

Monitoring the regional power load is an important method for implementing power demand side management,which has significant social and economic benefits.Non-intrusive load monitoring(NILM)uses intelligent sensing methods without entering the user’s home,and decomposes the user’s total load into information about each electrical device.NILM is the mainstream method and development direction of load monitoring,and has important research value.At present,NILM has the problems of low recognition accuracy,large amount of data,and strong dependence on cloud computing technology.This paper designs a non-intrusive load identification device and system based on V-I trajectory image,which can identify load information on the user side without relying on cloud computing,which effectively improves the accuracy and efficiency of load identification.This paper analyzes the requirements of the NILM system for home users and carries out the overall design of the system.The core composition and work flow of the NILM system are analyzed from three aspects: data collection,load feature extraction and load feature recognition,and in-depth research on load feature extraction and recognition.First,in order to improve the load recognition rate,the V-I trajectory image of the load is selected as the load recognition feature,and the recognition rate is improved by optimizing the coding algorithm and the neural network model.Through HSV encoding V-I trajectory image,three characteristic quantities of current,power factor and trajectory coincidence degree are added,which enriches the load characteristic information;in order to realize the V-I trajectory image recognition algorithm and improve the recognition accuracy under resource-constrained conditions,the convolution neural network(CNN)model is improved.Second,in order to solve the problems of large interactive data volume,dependence on cloud computing,and high implementation complexity of the existing NILM system,this paper designs an embedded non-intrusive load identification device for home application scenarios,with simple structure,easy integration,and realize the V-I trajectory image recognition algorithm based on CNN under resource-constrained conditions;only upload the recognition results to the NILM management platform in the cloud,which reduces the effective data transmission volume and improves the system efficiency.Finally,facing home application scenarios,a variety of common household appliances were selected to test the designed NILM equipment and system.The results show that the system can accurately identify the types and working status of common appliances in real time without relying on cloud computing,and the monitoring information can be presented on the NILM management platform;the test results meet the design goals and have good practical value.