Research On Predictive Control For The Three-phase Inverter And Its Parallel System

Predictive control is a class of optimization control algorithms originated fromthe industrial production process control, which has the advantages of low precisionrequirement of system modeling, strong robustness, online rolling optimization, andflexibility of comprehensively processing system control goals and constraints. Thethree-phase inverter and its parallel system is a category of converter systems, whichis deeply researched and widely applied nowadays, and the excellent controlperformances of it are the basic requirements of the green power system. But theresearch of combining predictive control with three-phase inverter and its parallelsystem is a novel topic. In this paper, the predictive control algorithms for thethree-phase inverter and its parallel system are designed. Processing characteristics ofrobustness of system model and integrated optimization of system control, thepredictive control algorithms are used to improve the output voltage wave quality andto reduce the tracking error of three-phase inverter.They are also applied in thethree-phase inverter parallel system to improve the control accuracy and the currentshare performance. Lastly the engineering practice of inverter parallel technology forDC600V train auxiliary power system is done.The predictive control research of three-phase voltage source inverter on thetracking problem of output voltage is done by the algorithm of model predictivecontrol based on control increment. The DC increment model of three-phasethree-wire voltage source inverter is built in the dqo rotating coordinates, based onwhich the predictive model of system is constructed. The optimize performance indexof predictive controller is designed according to the system control requirement, andthen expression of the optimal control increment is given. The expectation parametersof predictive controller are selected by the root locus method. The validity andsuperiority of the proposed model predictive control algorithm for three-phaseinverter is validated by the simulation and experiment.The conservative characters of traditional finite control set model predictivecontrol (FCS-MPC) for converters are analyzed, and then a multi-step predictionFCS-MPC (FCS-MPCMSP) is proposed, in which both optimal and suboptimalswitching function sets are considered in one control cycle, and the selected switchingfunction set is optimal in two control cycles. The online calculation burden of FCS-MPCMSP algorithm is analyzed by theory analysis and real-time simulation test.The contrast simulation and experimental results show that the wave quality of outputvoltage of the three-phase inverter controlled by the designed FCS-MPCMSPalgorithm is improved and controller performance of system is enhanced.In order to solve the problem that the system modeling error of traditionFCS-MPC algorithm is neglected, a finite control set model predictive controlalgorithm with modeling error compensation (FCS-MPCMEC) is proposed based onthe feedback adjustment mechanism. The idea of FCS-MPCMEC algorithm isembedded in the FCS-MPCMSP algorithm, and a contrast simulation ofFCS-MPCMSP algorithm with and without modeling error compensation is done, thevalidity of the compensation algorithm is tested.The research on three-phase inverters parallel system predictive control is doneby the algorithm of model predictive control based on control increment. The statespace model of the three-phase voltage source inverters parallel system is built in thedqo rotating coordinates, based on which the predictive model of the three-phaseinverters parallel system is constructed. The optimize performance index of predictivecontroller is built following the control requirement of inverters parallel system, aftersetting it and achieving hot-swap design of parallel inverters, a predictive controlstructure diagram of inverters parallel system is given. The expectation parameters ofpredictive controller are selected by the root locus method. The superiority andrealizability of the designed model predictive control algorithm for the three-phaseinverters parallel system is validated by the simulation and experiment.The droop control of inverters parallel system and the model predictive control ofthree-phase inverter are combined. A control structure of three-phase parallel inverterwireless predictive control is built. The reference voltage signal of model predictivecontroller is provided by the droop controller, and the optimize performance indexesof the model predictive controller are designed by the tracking error of the parallelinverter’s output voltage with the reference voltage, then the wireless modelpredictive control of three-phase inverters parallel system is achieved.The inverter parallel technology is applied to the DC600V train auxiliary powersystem, which is able to improve its reliability and redundancy. A solution based onPQ (active power and reactive power) droop control inverter parallel technology isproposed for the DC600V train auxiliary power system. The control structure andcalculation process of the algorithm are given. The validity and feasibility of thedesigned control algorithm is tested by the simulation and experiment. In order to improve the current share performance of the parallel inverters in the train auxiliarypower system, the droop control strategy is changed. The integral of the parallelinverter output active power is introduced into the system’s droop control strategy.The small-signal model of parallel inverter’s power control loop is built by themethod of small-signal analysis, then the impact of each parameter in droop controlleron system stability and dynamic performance are studied using the root locus method,the parameters with excellent system performance are selected for the droop controllerof the inverter parallel train auxiliary power system. The feasibility and superiority ofthe improved droop control strategy is validated by the contrast simulation andexperiment. The semi-practical simulation platform based on the dSPACE real timeemulator for the three-phase inverters parallel train auxiliary power system is builtfollowing the V-cycle mode of controller design. So engineering basis for in-depthresearch and system test of the inverter parallel train auxiliary power system isestablished.