Research On Dynamic Weight Of Model Predictive Control Based On FFT And Optimization Application Of Wind Farm Load Control

Wind power generation has become the fastest growing and largest application of new energy power generation methods.The traditional active power control strategy is to perform simple centralized proportional distribution according to the current active power margin of each fan.This distribution method does not take into account the mechanical fatigue of each fan due to the change in the distributed power.In actual operation,changes in fan power will cause such things as changes in the torque of the drive shaft and the torque on the tower.When the rate of change of these variables such as torque and torque is too large,it will aggravate the mechanical fatigue of the unit and cause the life of the fan to decrease.In response to this problem,the current multi-objective optimal allocation method considering mechanical fatigue is attracting the attention of researchers.In the past,some scholars have studied the problem of converting wind farm control into optimal control,and minimized the rate of change of variables that cause increased mechanical fatigue as the optimization goal.The research shows that,under the condition of meeting the power grid dispatching target,by dynamically adjusting the active power reference value of each fan,the change rate of the transmission shaft torque and other variables can be effectively reduced,and the damage equivalent load(DEL)under the same working condition is significantly reduced,and then Reduce mechanical fatigue and extend the life of the unit.The traditional centralized optimal solution will cause the calculation amount to increase exponentially with the increase of the number of wind turbines.When the number of wind turbines increases,it will cause difficulty in solving,so it is not suitable for real-time online control of large-scale wind farms.The optimal solution is obtained by iterating through the distributed controller and the main control unit to obtain the optimized power reference value,which is then transmitted to the stand-alone controller.As a parallel computing unit,the distributed controller greatly reduces the amount of calculation of the main control unit,so it is suitable for real-time control of wind farms of various sizes.At the same time,the iterative solution method based on Fast dual gradient has the advantages of rapid convergence and fewer iterations,which reduces the communication volume between the main control unit and the distributed controller.This paper combines the characteristics of model predictive control(MPC)and fast Fourier transform(FFT),and comprehensively considers the cycle life of the low-speed shaft torque of the tower thrust force,and designs an algorithm based on distributed operation model predictive control.Based on the current state of the wind turbine and the environment,predict and optimize the operating characteristics of a period of time in the future,and transmit the active power reference value of multiple wind turbines to each unit in a distributed form.When the grid dispatching is met,reduce each wind turbine.Loss,which in turn reduces the cost of the wind turbine during operation.This algorithm is validated by simple examples in MATLAB and the wind farm model,which can effectively smooth the fluctuations during operation.