Research On Control Strategy Of Maximum Wind Energy Capture In Wind Turbine

With the development of wind power projects in the low wind speed area,it is very important to study the maximum wind energy capture control strategy of wind turbines to improve the generating capability of wind turbines at low wind speed.There are some common control methods of maximum wind power capture for wind turbines,such as the optimal tip speed ratio method,the mountain climbing method,the torque feedback method and the power feedback method,etc.However,these methods all have some difficulties,including hard measurement of wind speed and obvious error of measuring wind speed in real time,long response time caused by large inertia of the wind turbine,the loss of power generation caused by the changing state of the wind turbine,and so on.All these difficulties affect the power generation capacity of the wind turbines in the low wind speed range.Aiming at the above problems,the variable speed constant frequency doubly fed wind turbine is taken as the research object in this paper.Based on BP neural network model prediction,a self-studying algorithm for maximum wind energy acquisition is proposed.The BP neural network model,which has the best prediction effect,is set up under the turbulence wind condition corresponding to the randomly selected learning data samples.The overall reliability of the BP neural network model for predicting the control state of wind turbines is verified in the whole wind speed range corresponding to the maximum wind energy capture area.Under the different turbulent wind conditions,the maximum wind energy capture control simulation is carried out with the self-studying algorithm.And the control results of the self-studying algorithm are compared with the control results of the Kopt,control method.The main contents and conclusions of this paper are as follows:(1)Determination of the basic parameters of the BP neural network modelThe learning data samples are selected randomly.After a lot of training and validation,the basic parameters of the BP neural network model is determined,which has the best prediction effect.The basic parameters include the learning rate,the step size of the batch calculation,the form of the output function,the scale of the prediction time and the number of the hidden layer neuron.etc.(2)verification of the overall reliability of the BP neural network modelUnder other turbulent wind conditions with the same average wind speed,and other turbulent wind conditions with the different average wind speed,the control parameters of the wind turbine are predicted using the built BP neural network model which has the best prediction effect.It is proved that the BP neural network model has high overall reliability in the whole wind speed range corresponding to the maximum wind energy capture area.(3)Analysis of the control effect of the self-studying algorithm for maximum wind energy acquisitionUnder the different turbulent wind conditions,the maximum wind energy capture control simulation is carried out with the self-studying algorithm.And the control results of the self-studying algorithm are compared with the control results of the Kopt control method.The reliability and stability of the self-learning algorithm are verified.And it is proved that the maximum wind energy capture control with the self learning algorithm can improve the power generation in the whole wind speed range corresponding to the maximum wind energy capture area,especially the power generation at low wind speed.