Research On Robust Predictive Control Of PWM Rectifier Under Distorted Grid

The PWM rectifier can realize bi-directional power flow,controllable power factor,sinusoidal current,and high quality dc-bus voltage.So it has been widely used.In practical applications,the PWM rectifier encounters parameter perturbations and external disturbances.The parameter perturbations are mainly due to changes in the parameters caused by the ambient temperature.External disturbances usually appear as unbalanced or distorted grid.The control performance will deteriorate owing to the change of the parameters,which is reflected in the distorted current at the grid side,the increase of harmonics and the reduction of the power factor.In severe cases,the system may be unstable.The change of parameters has a great influence on the system,especially the control algorithm which highly relys on system parameters,such as predictive control.When the grid voltage is disturbed,the grid side current will also be distorted,which does not meet the grid code.Therefore,the disturbance problem encountered in practical applications of the PWM rectifier needs to be solved urgently.Firstly,two kinds of positive and negative sequence extraction methods,delayed signal cancellation and generalized Fourier transform are introduced,which lays the foundation for the research under distorted grid.Secondly,this paper addresses the problem of parameters change of PWM rectifier from the perspective of parameter identification.The identification model of the least squares method was reformulated under the unbalanced grid and the effectiveness of the improved least squares method was verified by simulation and experiments.In this paper,the method of gradient correction identification is also studied.Due to the slow convergence rate,the gradient correction method is seldom used for real-time parameter identification.Based on the existing research,this paper improves the gradient correction method,which makes it similar to the least square method in the parameter convergence rate,but the calculation is slightly less than the least square method.Finally,the effectiveness of the two parameter identification methods are verified under the distorted grid.Thirdly,this paper improves the robustness of the system from the perspective of model free control and disturbance observation.There are three methods for model free control,which are based on current difference,ultra-local model and extended state observer,respectively.Since the system parameters are not used,model free control is very robust.Under unbalanced grid,the observed disturbances are used to compensate for the errors caused by the disturbances and improve the robustness of the system.Combining model free control and disturbance observation with predictive control respectively,which solves the problem that predictive control has a strong dependence on parameters.Combining with positive and negative sequence extraction methods,robust control is achieved under distorted grid.And the simulation and experimental results validate the effectiveness of proposed methods.Finally,the grid voltage sensorless control is studied under distorted grid.The virtual flux can achieve grid voltage sensorless control and has a certain ability to suppress harmonics.However,the virtual flux based on low-pass filter is not suitable for unbalanced grid.This paper has improved it and combined it with delayed signal cancellation to realize grid voltage sensorless control under distorted grid.