Study For Wind Turbine Generator Control System Based On Wind Speed Estimation

Modern wind power technological research focuses mainly on two important aspects - improving electrical power generation efficiency and reducing power generation cost. In practice, we make generators run at maximum output power point to achieve the maximum output power by installing aerodromometer and other sensors to detect wind speed. On the one hand, this measure can achieve the purpose of improving power generation efficiency. On the other hand, the incorporation of the anemometer leads to a slower dynamic response, and an increase of system equipment cost and maintenance cost, also a reduction of the overall efficiency of doubly fed induction generator (DFIG) system.Therefore, this paper puts forward a non-detection of the maximum wind speed tracking algorithm, using support vector machines (SVM) to predict the wind speed, deducting rotating speed of the rotor corresponding to the maximum power point from the real-time accurate wind speed data, thereby reducing system loss and cost.Firstly, this paper describes the significance of wind power, the current research situation and the superiority of the DFIG. Then, this paper represent the structure of DFIG, establish a mathematical model of three-phase coordinate and two-phase rotating coordinate, and determine the control strategy of both the rotor side controller (RSC) stator flux orientation and grid side controller (GSC) grid voltage orientation. In the wind speed predict algorithm, this article introduces the SVM based on the Statistical Learning Theory and Structural risk minimization principle which can achieve the best balance point between studying accuracy and studying ability. This paper established the model and compared different types of core functions based on describing the principle of SVM. After that, it compares Cross Validation (CV) method with PSO method in order to choose the scheme which verifies using PSO method to optimize two significant parameters in SVM network to make the network fast and accurately. Simulation results show that the improved SVM algorithm has excellent accuracy in prediction.Finally, this article established a DFIG simulation model based on wind speed estimation by using Mathworks MATLAB in view of the mathematical model of 3.6MW DFIG system. The consequence indicates that the system can work at the maximum of wind speed output power point, transmitting active power in permenantly constant voltage and frequency with excellent rapidity and robustness.