Power Optimization Control Of Variable Speed Variable Pitch Wind Turbines With Data-driven Control Method

Wind energy is one of the most competitive renewable energy.In recent years,wind power has been developing rapidly,and the capacity of wind power installed in the world is increasing.However,the fast-changing wind speed has a great deal of influence on the safety and reliability of wind turbine.Aiming at the problem of automatic generation control(AGC)and its frequency stability under the background of high wind power penetration,the operation control mechanism and optimization control strategy of large-scale grid connected wind turbine has become a research hotspot in the field of wind power.The stochastic fluctuation of wind determines the existence of nonlinear and strong disturbance in wind power system.It is difficult to establish the wind turbine model accurately.The model-based control method is difficult to achieve good control effect.Therefore,based on the data driven control method,the power optimization control problem of variable pitch wind turbine is studied as follows:(1)The variable speed variable pitch wind turbine characteristics and control strategies are analyzed based on wind power system structure and SCADA(Supervisory Control and Data Acquisition)data.The 15 minutes rolling wind speed is forecasted based on 4-hour-ahead time scale using small-world BP neural network(SWBP)algorithm,combined with numerical weather forecast information.Results of wind speed prediction performance evaluation show that the predicted wind speed accuracy based on the SWBP algorithm is high,which is suitable for the information synthesis of the control system.(2)Research on the MFAC-PID control method of variable speed variable propeller wind turbine.Firstly,the control input criterion function is set up to achieve the goal of maximum wind energy capture and smooth power point tracking.Then,a series of equivalent local linearization models are established through pseudo-partial derivative(PPD)obtained from online I/O measurement data,and the model free adaptive control(MFAC)algorithm is obtained.Finally,online control is adjusted on the condition that both MFAC and PID control increment constraints at the same time,and MFAC-PID control algorithm is designed to ensure the output steadily track the set value.Simulation results show that the MFAC-PID algorithm guarantees the convergence of tracking error in all wind speed,and completes the dynamic control based on I/O data.The constant power output and stable regulation under full wind speed are better achieved.(3)Research on multi-target model free adaptive predictive control(MFAPC)for variable speed variable pitch wind turbine.MFAPC strategy is realized by on-line estimation and prediction of PPD based on I/O data using the advantages of both predictive control and MFAC.According to the optimal control target of wind turbines in whole wind speed range,the multi-objective function of optimal output power and minimum control variable is designed.Simulation results verify the convergence and stability of the MFAPC scheme,and the vibration is small in control process,the response speed is fast.(4)Feedforward compensation controller which completely compensates wind speed disturbance is designed with wind speed prediction.Combined with the feedback controller of MFAPC,a feedforward-feedback composite control system of wind speed disturbance suppression is established.By building a closed loop simulation test platform for BP neural network in different working conditions,the simulation is compared with the control strategy of PID,MFAC and MFAC-PID.Results show that the composite strategy achieve advanced control of pitch angle when wind speed changes randomly to reduce load disturbance,which showes good robustness and dynamic characteristics.