Variable Predictive Horizon Nonlinear Model Predictive Control And Its Application In Photovoltaic Grid-connected System

In today’s society,clean energy is gradually replacing traditional fossil energy.In the field of power electronics,major breakthroughs have also been made in the research of photovoltaic power generation technology.With the continuous expansion of photovoltaic distributed grid-connected systems,the control of grid-connected inverters is also innovating and developing.The classical control technology is constantly improving in the application process,but these traditional control strategies face difficulties in designing parameters and the inability to guarantee the rapidity of dynamic response.When a voltage sag fault occurs on the grid side,the distributed grid-connected system is usually kept in grid-connected operation,and the low voltage ride-through technology is used to quickly inject compensation current into the grid side to raise the grid-connected side voltage.The key to low voltage ride-through technology is output current tracking.Model Predictive Control(MPC)technology has been widely used in grid-connected inverters because of its fast trackability,and can be combined with low voltage ride-through technology to achieve control objectives during fault ride-through.Low voltage ride-through with model predictive control can effectively compensate voltage and track current,however,the accuracy and speed of model predictive control will be affected by the system delay.Nonlinear Model Predictive Control(NMPC)technology expands the predictive horizon from single to multiple to improve tracking performance.In this paper,the application and development of model predictive control technology in low voltage ride-through are analyzed in detail.On the basis of NMPC,the expansion method of predictive horizon and the form of value function are improved,and the predictive horizon is improved to dynamic optimization form,the upper limit of the predictive horizon has be raised without increasing the operational burden.The predictive horizon rolling optimization methods,which include predictive horizon incremental self-tuning NMPC and error perturbation self-tuning NMPC are proposed.The former needs to calculate the optimal value function and switching vector in the three time domains each time,and the latter only needs to compare the current error in each calculation cycle to select the optimal predictive horizon and the optimal switching vector,which avoids the calculation time delay to a certain extent,saves the calculation space and reduces the switching frequency appropriately while ensuring fast tracking.In this paper,the low voltage ride-through control simulations using control strategies of MPC,NMPC,predictive horizon incremental self-tuning NMPC and error disturbance self-tuning NMPC are carried out on MATLAB/Simulink platform respectively,which verifies the principle and performance is compared and analyzed.Based on d SPACE,an experimental platform for grid-connected inverter is built,and the feasibility of the model predictive control strategy is verified under normal grid-connection,also the NMPC,predictive horizon incremental self-tuning NMPC and error disturbance self-tuning NMPC were tested under fault conditions.Through the comparison of the experimental results,it is verified that the error perturbation auto-tuning NMPC has the smallest tracking error and the fastest tracking speed.