Study On Harmonic Suppression Technology Of Microgrid Based On Neural Network

In recent years,with the adjustment of energy structure and the promotion of energy internet strategy in China,microgrid has developed rapidly in the reform of power system because of its advantages of low energy consumption and low pollution.But,most of the power sources in the microgrid are distributed power.A large number of power electronic equipment access in the microgrid.So the microgrid has the problems of randomness,volatility and serious harmonic.The problem of power quality not only restricts the healthy development of microgrid,but also becomes an urgent problem in the field of power energy.In order to suppress the harmonic and improve the power quality of microgrid,this paper studies the parallel APF control system,and analyzes its working principle,traditional classical control method and more advanced intelligent control method.On the basis of this,a controller based on compound artificial neural network is proposed to improve the harmonic compensation speed and precision of APF which improves the power quality of microgrid.According to the principle of APF control,the compound neural network controller presented in this paper is divided into harmonic detection part and current compensation control part.The functions of these two parts are realized by neural network.In the part of harmonic detection,BP neural network which has super nonlinear mapping ability is used in this paper.The system can output the normal fundamental wave stably in the second cycle with the traditional detection method of ip-iq.But using the method that proposed in the paper,the system can output the normal waveform stably from the first cycle.The current compensation control part of compound neural network controller is added to neural network predictive loop before PWM signal generation.The loop adopts RBF neural network which has simple application good convergence.It is widely used in the field of automatic predictive control.Predictive control can effectively solve the delay problem of APF control system.Therefore the paper adopts the prediction algorithm based on RBF neural network.The prediction accuracy is improved by two orders of magnitude compared with the advanced adaptive prediction method by programming the program of Matlab for prediction and simulation.The current predictive compensation control is well realized.The global dynamic simulation is carried out by building the circuit model on the Matlab/Simulink platform.The composite neural network controller which adopts the APF model is placed in the main circuit of harmonic source for filtering test.The results show that the APF has strong anti-interference ability and good dynamic response speed in circuit operation.It suppresses nearly 90 percent of the harmonics in the line.THD dropped nearly to 1/10.It obtains satisfactory filtering effect and realizes the purpose of improving power quality of microgrid in a certain extent.