Adaptive control of wind turbine generator system based on RBF-PID neural network

Wind turbine generator system (WTGS) is a complex multi-varies nonlinear system with the characteristics of time - varying, strong coupling and multi - interference. Wind speed is affected by many factors, the magnitude and direction are random. There are numerous of factors such as temperature, weather, environment and so on that affect wind turbine generator system, result in the WTGS cannot be guaranteed safety operation and constant output power. With the characteristic of strongly nonlinear, delay and uncertainty, the WTGS cannot be given an ideal control using traditional PID controller. Although the neural network control can solve the problems of nonlinear and uncertainty, it belongs to nonlinear approximation in essentially and cannot eliminate the error in steady state. In order to improve wind turbines behavior, an adaptive control method based on RBF-PID neural network is presented in this paper. This algorithm synthesizes the mechanical and electrical characteristics. System identification is part of the controller. At high wind speed, pitch angle is adjusted to keep rated output power using neural network adaptive controller. The RBFNN is used as the identifier of the WTGS. According to the identification information and the given learning speed, PID parameters are modified on line. The Matlab simulation results at random wind speeds show the controller can effectively improve the performance of variable pitch control.