Adaptive Model Predictive Control Strategy For Voltage Source Inverters

Inverters,as key interfaces to transfer electric energy from DC ports to AC ports,play an important role in industrial fields,such as photovoltaic systems,motor drives,electric vehicles,energy storage systems and so on.Reasonable and effective control strategies are the key to ensure that the inverter system can achieve stable and efficient performance in power conversion.Model predictive control,as a nonlinear optimal control method,is characterized by the intuitive modeling,fast dynamic response and strong robustness,and has been widely applied in inverter control fields.In real application,system parameter uncertainties or fewer sensors may result in information loss,which will have adverse impacts on the control effect.Aiming at the research on current tracking of inverters with information loss,this paper proposes an adaptive model predictive control scheme,where corresponding observers are designed to determine the lost information and correct the system deviation.The proposed method can not only reduce the adverse effects caused by information loss,but also monitor the lost information in real time.The specific work of this paper is as follows:For the inverter current tracking with parameter uncertainties,this paper selects single-phase inverters as examples for the research conduction.Based on the system model of the inverter,the impact on the system caused by the interface parameter uncertainties is represented as unknown periodic deviation,and two different observers are designed for the identification of the deviation.One is the linear time invariant observer,which is responsible for determining the instantaneous values of unknown deviations.The design process for this observer is simple,but complex post-processing units are necessary for the calculation of the interface inductance and resistance.The other is a linear time-varying observer,where the parameters should be determined are the amplitude of the sine and cosine components of the unknown deviation.On the basis of the amplitude,the process for computing the interface parameters is much easier.The convergence of this observer is guaranteed by Lyapunov theorem and Lasalle invariant set principle.Moreover,based on the accurate estimated value of the interface parameters,the maximum tracking current is also discussed in this paper.Then the accurate sinusoidal current tracking is realized by model predictive control strategy.In view of the grid voltage loss caused by fewer sensors,three-phase inverters are chosen as control objects for the research of the proposed adaptive model predictive control scheme.One phase voltage of the grid is collected by a voltage sensor for the phase detection,which provides the necessary information for the observer design.Then the lost two phase voltage is decomposed,and linear time-varying observers are designed for the observation of the amplitude of each component.Based on the estimated grid voltage,model predictive control method is applied for the optimal control signal selection according to the predictive model and cost function so that the control goals can be achieved.The performance of the controller is studied under different grid conditions,including sinusoidal grid,distorted grid,and distorted grid with non-standard frequency.Furthermore,the current tracking effects with grid voltage loss and integrity are compared,and there are similar results,which demonstrates the effectiveness of the proposed adaptive model predictive control method.Finally,the adaptive model predictive control method for inverters in the above two cases has been verified by simulation and experiments.The simulation part is carried out in the real-time simulation system built by MATLAB/Simulink tools.The experimental part is carried out on the rapid control prototype platform,which is combined by PC and NI CompactRIO.The LabVIEW environment in PC is responsible for the design of the bottom control program and man-machine interface,which is convenient for real-time monitoring and storage of experimental results.