Research On Load Identification Approach Based On Neural Network

In power distribution management,load identification technology can carry on electrical matching through the accurate identification of the load to improve the utilization of electrical energy.In safe power management,load identification also can improve the management efficiency of safe power consumption by identifying and limiting prohibited appliances.Therefore,the study of load identification technology is of great significance.This thesis studies the load identification method based on BP neural network.Firstly,the thesis discusses the extraction,analysis and selection of load characteristics that can be used to identify load types in load currents.By collecting a variety of load current sampling data,extracting a number of optional characteristic in time and frequency domain s which can be used for load identification from the load current,using the Relief-F feature selection method to effectively select among the optional load features,the thesis selects 12 time-domain and frequency-domain load features with strong correlation with load types for load recognition.Then,the thesis designs BP neural network for load identification.According to the selected 12 load characteristics,through theoretical analysis and experimental tests,the thesis chooses important parameters for load identification BP neural network,such as the number of network layers,the number of nodes in input layer,the number of neurons in hidden layer,the number of output nodes and the activation functions of hidden layer and output layer and so on,thus confirms the structure of load identification BP neural network.Finally,this thesis optimizes the parameters of load identification BP neural network by using Particle Swarm Optimization(PSO),aiming to the existing defects of BP neural network.The simulation results show that our load identification BP neural network before and after optimization can both perform load identification.Compared with the BP neural network before optimization,the optimized BP neural network by PSO has higher convergence rate in training and accuracy rate for load identification.Compared with the load identification method based on SVM,the optimized BP neural network also improves recognition performance.