Study On Controller Of Variable-Speed Variable-Pitch Wind Turbines Based On Neural Network

As an important kind of new sustainable development energy, wind power generation not merely can save the routine energy resources, but also may decrease pollution of the environment. Thus it can bring preferable economic benefits and society benefits. At present, wind power generation has caused more and more concern by governments all around the world. Because wind energy has lower density, instability and randomicity, wind turbines have strong nonlinear multivariable with many uncertain factors and disturbances. Therefore control technology is crucial to the efficiency and reliability of wind turbines.Variable-speed variable-pitch control system of MW wind turbines is the core control technology, has characteristics of nonlinear pure delay. In the mean time, wind load changes frequently in the actual system, and the system has grave random disturbance, making it difficult to establish the ideal control model. So common PID control will not satisfy the control need. Neural network has strong nonlinear mapping ability, self-learning ability, associative memory capacity, and parallel information processing and excellent fault-tolerant performance. PID controller which is improved by neural network will have a better control effect in complex wind turbines, be effective in improving the situation of instability due to the changes of system structure and parameters.This paper mainly engages in research on controller of variable-speed variable-pitch wind turbines. Firstly, structure and basic theory of wind turbines are introduced, traditional control technologies of wind turbines are analyzed, and a novel self-adaptive neural network PID control strategy is proposed to make up traditional PID control's lack. Secondly, the mathematical model of variable-speed variable-pitch wind turbines is set up on the basis of dynamic characteristic of wind power system. Thirdly, aim to variable-speed variable-pitch system, its self-adaptive neural network controller is established, and the comparison with common PID is given through algorithm simulation. According to the results from the simulation, self-adaptive neural network controller has more sensitive dynamic response, smaller overshoot than common PID controller. It can achieve the greatest wind energy capture below rated wind speed, and maintain wind turbines rated output power above rated wind speed, and have better real-time performance and robustness.