Research on modeling of wind turbine based on LS-SVM

The generator power have related with the wind turbine torque heavily. The wind speed, the rotor speed, the pitch angle and the inherent parameters of wind turbine can influence the wind turbine torque. Different torque curves in different operation can be simulated by Simulink software of Matlab. However, this method needs wind turbine´s parameters which aren´t usually obtained to construct the model and simplifying of the model process maybe bring with errors. The paper proposed - an intelligent torque model with LS-SVM algorithm, this model only needs the input-output training samples to overcome the problem of parameters unavailable. Compared with BP network, this algorithm supports the training of small samples and has good abilities in calculating speed, approximating precision and forecasting effects. Simulation results demonstrate the validity of the model. Simulation based on practical data of wind power is also given and can provide a beneficial reference for the prediction theory of wind turbine.