Model Predictive Control Method Of Cascaded U-Cell Statcom

Compare with traditional multi-level converter,cascaded U-Cell topology can achieve more levels with fewer power switching devices.The equipment composed of cascaded U-Cell has the advantages of lower cost,small size,and higher efficiency.Model Predictive Control(MPC)has higher dynamic performance and stronger robustness compared with the traditional control method.It is easier to implement the MPC control method for cascaded U-Cell STATCOM system which has multiple constraints and variables.This paper focuses on the application of MPC in cascaded Ucell STATCOM.Firstly,the five-level cascaded U-Cell topology is taken as an example to analyze its working conditions in different states.The mathematical model of cascaded U-Cell STATCOM in continuous time domain is established.The three keys of model predictive control are analyzed in detail: the establishment of discrete mathematical model,the design of cost function and the rolling optimization.The simulation model was built in Simulink to verify the feasibility of MPC application in cascaded U-Cell STATCOM.The application of MPC in cascaded multilevel topology has the problems of large computation amount and serious computation delay.On the basis of applying the delay compensation in MPC,the MPC control strategy is layered optimized,which can essentially reduce the amount of calculation in the process of MPC traversal optimization.The traditional MPC by weight coefficient will all control objectives into a control objective to comprehensive solution,and hierarchical optimization MPC control strategy according to the importance of the control targets multiple control objectives can be divided into different priority,high priority of the optimal solution for solving constraint condition of priority control target,thus reduced the MPC control strategy of measurement from exponential to polynomial.When the system model is not accurate,the predicted value of MPC will produce a large deviation,which makes it choose the suboptimal switching vector in the value judgment,and reduces the compensation performance of STATCOM.In this paper,the prediction errors of DC side voltage and grid-connected current are analyzed respectively from three aspects of model parameter mismatch,system equation discretization and sampling,and based on this analysis,a hierarchical error correction method is proposed.Firstly,the predicted value of DC side voltage is corrected,and the predicted value of grid-connected current is corrected based on the predicted value of the corrected voltage.The method can not only reduce the model parameter mismatch and the prediction error caused by sampling,but also reduce the inherent discretization error of MPC.Simulation results show that the quality of grid-connected current can be improved obviously after error correction.Finally,a multi-processor five-level cascading U-Cell STATCOM experimental platform based on DSP+FPGA is designed and built,and the prediction error in MPC control strategy is emphatic verified.Experimental results show that the theoretical analysis of the prediction error in the cascaded U-cell STATCOM is correct and the hierarchical prediction error correction method is effective.In this paper,there are 77 pictures,6 tables and 92 references.