Quadratic Optimal Tracking Control Of Wind Turbine System Based On Support Vector Machine

The problem of environmental pollution has become one of the most popular social topics,when the word of PM2.5 appeared into our daily life and air pollution reached to the red warning level.13th Five-Year programming firstly put forwarded that must establish development goal with innovation,coordination,green,opening and sharing,persist on green development and increase the proportion of non-fossil energy.At present,the development of wind power at home and abroad is on the rise,and the capacity of wind turbine is increasing year by year.And there are their first commercial wind farms in some countries.Because of wind energy's relatively mature technology and lower cost,it has become one of the most promising renewable energy source in the world.As a variable resource,wind has some features,such as nonlinear,random,uncontrollable,and so on.Under normal circumstances,wind turbine system is designed according to the fixed output.So realizing short-term wind speed prediction with high accuracy is of great significance for smart grid,power dispatching,wind power and meteorological services.In this article,the research objective is wind turbine system,and the objective is achieving the tracking control and optimal between wind turbine rotor speed and desired rotor speed.The main content of the paper is roughly divided into three major parts as follows.First,in the part of the wind speed prediction,there are three prediction models,ARIMA model,grey model and SVM model.These models are used to simulate the future wind speed,and their simulation results are compared.The simulation data,including wind power,engine speed,rotor speed,wind speed,come from the National Renewable Energy Laboratory(NREL).The result demonstrates that SVM algorithm not only has a solid theoretical foundation,but also has a better predicting performance.Second,this paper will establish the physical model of the horizontal axis wind turbine with gearbox and obtain the dynamic equation of the ONE-MASS model.The dynamic equation includes rotor speed,rotor torque and generator torque.Meanwhile,it demonstrates their relationship.Finally,it's the tracking control of rotor speed,and this part will use the data of wind speed prediction based on SVM.In this article,there are two algorithms to accomplish the goal respectively and their results will be compared.The first algorithm is robust adaptive control,which can accomplish the tracking control with system uncertain parameters and prove validity of controller theoretically.The second one is linear quadratic optimal control.System's controller is divided into two parts.One is parameter identification of rotor torque,and the other is linear part of rotor speed.The method of linear quadratic optimal control is used in the linear part of controller.Accurate identification of rotor torque is obtained by system parameter of wind turbine and the forecasting wind speed based on SVM model.The method solves the problem of setting parameters limit on some identification methods,which guarantees the boundedness and accuracy of rotor torque.Then use the MATLAB simulation,and compare the results in the two respects of tracking performance and energy consumption.The result of comparison is that accurate identification of rotor torque and linear quadratic optimal control can accomplish the ideal tracking control with fast,stable and energy-saving.