Research On Model Predictive Control Of Non-isolated Two-switch DC-DC Converter

With the aggravation of fossil energy crisis and environmental pollution,the 14 th Five-Year Plan proposes to accelerate the promotion of green and low-carbon development.The non-isolated two-switch DC-DC converter adapts to the development requirements of our country’s new energy field with its advantages of low switching pressure and high conversion efficiency.Among them,as the typical non-isolated two-switch DC-DC converter,the two-switch Buck-Boost(TSBB)converter has the characteristics of wide input range and the same polarity of input and output voltage.It has received extensive attention in fuel cells,new energy power generation and other applications that require up-down voltage conversion.Model predictive control(MPC)has developed in industrial process control.This method does not require complicated parameter trial and error process.The rapid development of microprocessors makes it possible to apply model predictive control to the field of power electronics.In this paper,the typical non-isolated two-switch DC-DC converter is taken as the research object,and its continuous control set MPC(CCS-MPC)is researched to solve the problems in the application of traditional CCS-MPC.First of all,the working principle of TSBB converter is analyzed,and the mathematical model of TSBB converter in continuous time domain is established.The traditional CCS-MPC is introduced,the predictive model of the converter is established,and the optimal control law is derived through the cost function.Simulation and experimental results prove that the method can ensure the control accuracy of the converter,especially the Buck mode has excellent control performance.Secondly,an improved CCS-MPC is proposed to solve the steady-state error between the control output and the expected reference under the traditional CCS-MPC due to the inaccuracy of the model.The integral part is applied to the cost function,and the steady-state error of the output voltage of the system is effectively reduced through the optimal control of the cumulative prediction error.Simulation and experimental results prove the effectiveness of the proposed method.Furthermore,in view of the fact that the traditional CCS-MPC samples multiple state variables such as voltage and current,which leads to the high hardware cost,the CCS-MPC with a low current sensor count is proposed.For the Buck mode of the converter,the current sensorless CCS-MPC is proposed,which simplifies the predictive model from the internal connection of the predictive model,making it easy to design the current observer and eliminate the use of current sensors completely.For the Boost mode and Buck-Boost mode,the CCS-MPC without load current sensor is proposed to reduce the number of current sensors used in these two working modes.Simulation and experimental results verify that the proposed CCS-MPC with a low current sensor count can ensure the performance is similar to the traditional control strategy,reduce hardware cost and improve system reliability.Finally,the CCS-MPC based on three-step discretization is researched to solve the problem that the median value of the discrete period may be mismatched with the predicted value due to the one-step discretization of traditional CCS-MPC.By refining the interval of duty cycle of converter’s controllable switches,the converter prediction model is reestablished,and the feasibility of the researched method is analyzed based on simulation results.Analysis shows that the researched control method can improve the control accuracy of the converter to a certain extent.But compared with the improvement of its control performance,the calculation amount generated by this method has a greater increase.Therefore,the first improved strategy proposed in this article should be adopted first.