Research On Improved Model Predictive Control Strategy For Doubly-fed Wind Power Generation Syste

Doubly-fed wind power generation systems are currently widely used in the wind power market,with the advantages of variable speed and constant frequency operation,flexible adjustment of the reactive power of the power system,and safe and convenient grid connection of generators.However,due to the randomness and strong fluctuation of wind speed,the traditional double-closed-loop PI control strategy cannot meet the timeliness of speed tracking,nor can it guarantee the stability of the DC bus voltage and the quality of output current of the grid-side converter.Therefore,this paper proposes a control strategy combining improved model predictive control and extended state observer with model parameter identification on the speed outer loop and the current loop on the grid side,so as to effectively optimize the maximum wind energy capture and grid-connected operation of doubly-fed wind turbine system.The main work content of this article can be summarized in the following two parts:First,in the design of the speed outer ring controller on the machine side,this paper proposes a control strategy combining model predictive control and extended state observer based on model parameter identification method for the maximum power tracking control target in doubly-fed wind power generation system,which effectively weakens the adverse effects of the tracking process of the model mismatch control target and reduces the mechanical stress of the transmission bearing.The model mismatch caused by turbulent wind speed(external disturbance)is solved by introducing the prediction error correction term in the performance index function to compensate for the predicted output.For the problem of model mismatch caused by motor mechanical parameter perturbation(internal disturbance),the useful disturbance information(rotor inertia,viscous friction coefficient,transmission shaft torque)is extracted in time in the extended state observer by using the online model parameter identification method to compensate the prediction model.At the same time,the extended state observer not only estimates the rotational speed,but also actively suppresses the interference affecting the speed tracking on the control output channel,so as to track the wind speed change as quickly as possible.The simulation results under the two disturbance conditions prove the feasibility and effectiveness of the control strategy proposed in this paper.Second,in the design of the current loop controller on the grid side,this paper proposes an improved current prediction control algorithm for the three-phase inverter to optimize the tracking control of the current and keep the DC bus voltage stable.The error caused by the discretization process of the system model is solved by directly applying the control quantity to the continuous-time system model by using the zero-order retainer.The control delay problem caused by multiple rolling optimization in predictive control is solved by approximating the delay time.The problem of excessive deviation between the prediction model and the actual system caused by uncertainty and other factors is solved by designing an evaluation function with prediction error compensation terms,which improves the adaptability of the prediction model in the actual environment.The simulation results show that the improved current predictive control proposed in this paper has smaller current ripple than the traditional current predictive control,and the correction effect of the prediction error is better,which not only ensures that the load current is close to the sine wave,but also maintains the stability of the DC bus voltage,and effectively improves the grid-connected operation.