Robust Model Predictive Control For Grid-connected Inverter Based On Improved Phase-locked Loop And Inductance Identification

In recent years,with the rapid development of the social economic and the increasing demand for energy,the problem of energy shortage is becoming more and more serious.Therefore,adhering to green development,focusing on improving the ecological environment and developing clean and efficient renewable new energy are the top priorities of today’s society.As the core of new energy power generation,grid-connected inverter technology is more and more widely used in new energy grid-connected power generation systems.How to accurately identify the circuit parameters quickly and rationally improve the inverter control technology has become a key scientific and technological problem to be solved.This paper focuses on the control of robust predictive model control for grid-connected inverters.The robustness of system parameters is improved from many aspects,and then the control performance of gridconnected inverter is also improved.The research contents include: traditional robust model predictive control、improved phase locked loop(PLL)design of PI-free controller、improved inductance identification method research and robust model predictive control based on improved PLL and inductance identification,etc.Firstly,the principle of traditional model predictive control and the sensitivity of traditional model predictive control strategy to inductance parameters are analyzed.Aiming at the problem that the control performance will decrease when the inductance parameters are mismatched,three traditional robust model predictive control strategies are introduced,including model predictive control based on weight coefficient correction,model predictive control based on current error correction and model predictive control based on sliding mode theory.The effectiveness of the method is verified by simulation.It is also pointed out that the qualty of PLL performance and the matching degree between inductance parameter and actual inductance value are important indexes that affect the stable operation of grid-connected system.Secondly,the traditional PLL that relies on PI controller,is difficult to debug and it has weak adaptability to frequency changes,so an improved PLL without PI controller is proposed.The improved PLL has very fast dynamic convergence speed and high estimation accuracy.It also shows strong adaptability when the power grid frequency changes suddenly.Thirdly,the traditional inductance identification method needs the actual frequency of grid voltage,and cannot take measures to enhance its robustness to frequency deviation,which leads to large error of inductance identification.Two identification methods based on sliding mode observer are proposed.These two methods can not only improve the robustness of the system to the inductance when the inductance error occurs,but also reduce the inductance recognition error and improve the robustness of the system to the frequency by selecting the appropriate sliding mode gain when the frequency error exists.What’s more,three traditional inductance identification methods(recursive least square method with forgetting factor,gradient method and direct calculation method)are studied.Those methods are show the superiority of the new inductance identification method.Fourthly,a model predictive control strategy based on improved PLL and inductance identification method is proposed.Experimental results show that the model predictive control strategy can have double robustness to inductance and frequency.Finally,the improved PLL and the improved inductance identification method are tested in detail on the experimental platform.The effectiveness and feasibility of the robust model predictive control strategy based on improved PLL and inductance identification method are verified.