Integrated Control System for Three-level Inverters based on Artificial Neural Network

Abstract

In the actual operation of power grid, inverter is the access of source and power grid, and is an important control link affecting power quality. The three-level inverter has very strong nonlinearity, and the loading of nonlinear load produces large harmonic, which seriously affects the transmission quality. The artificial neural network has good fault tolerance and adaptive ability when dealing with the nonlinear system. Based on the artificial neural network algorithm, the three-level inverter control system is designed and simulated. The results show that the neural network control can deal with the nonlinear and uncertainty of the three level inverter control system. With constant voltage and constant frequency, the fluctuation of the output voltage and frequency is within controllable range. The three-level inverter controlled by artificial neural network shows smaller fluctuation and faster adjustment speed. The steady-state error of artificial neural network control is smaller than that under the control of PID. The use of artificial neural network control can greatly improve the dynamic and static capacity of the inverter system.