Research Of Variable-pitch Control Based On Prediction Of Wind Speed

Because the randomness of the wind speed and stability of wind power generation system in the process of variable pitch control is easily influenced by wind speed fluctuation,the wind power output fluctuation is very big,therefore was proposed based on wind speed forecasting wind power variant pitch control research,in front of the interconnection of the fan motor speed control.Aiming at the problem of low accuracy of short-term wind speed prediction,a method for improving the short-short wind speed forecasting of Support Vector Machine(SVM)is proposed.Weighted similarity measure method to join in the lithosphere in the sample interval change trend similarity function,research suggests that to improve the SVM sample weight,the prediction error can be from 7.61% to 7.46%,verify the effectiveness of the proposed method.Aiming at the problem of low accuracy of short-term wind speed interval prediction method,the short-term wind speed interval and trend prediction method of Extreme Learning Machine(ELM)based on fuzzy information granulation are presented.For the wind speed sequence of fuzzy information granulation ELM,the SVM prediction model is set up,respectively,the study found that the ELM model prediction error prediction error is lower than the SVM model,and three sequences of average fell 1.9%,0.65% and 1.9% respectively,the superiority of the proposed method is verified.Based on the ultra short term wind speed forecasting and prediction of wind speed trend,the establishment of Recurrent Neural Networks(RNN)pitch predictive control simulation system,realize the control of speed,and RBF neural network predictive control system and variable pitch fan site in the PI control strategy under the actual condition to do comparison study.The control results of fan actual fan speed jitter,it is difficult to achieve the smooth fan grid,and pitch based on RNN distance pitch prediction based on RBF neural network prediction from the motor speed control in the rise are relatively flat and control,to verify the pitch prediction control effectiveness of speed control strategy,through the comparison of RNN and RBF neural network prediction results of variable pitch control,predictive control of motor speed rise can be found relatively more gentle,small jitter variable pitch based on RNN,and in the near rated speed is relatively stable,can realize the connection of the wind turbine is more stable.