Finite Control Set Model Predictive Control Of Active Front End Converter Based Multi-motor System

With the rapid development over the past decade or two, multi-motor systemsbased on the active-front-end (AFE) converters and common DC-bus technology willbecame mainstream products in the future industrial automation field. In themulti-motor systems, the grid-side converters and the multi-motor-side drivers havedifferent control targets and algorithms, which would increase the complexity of thecontroller design of the overall system. In recent years, AFE converters and motordrivers based on finite control set model predictive control (FCS-MPC) have beenwidely discussed. Taking into account the discrete nature of power converters,FCS-MPC predicts the future behaviors for all possible switching states over a timeframe, and the optimal one that minimizes the cost function is selected to obtain thedesired output. FCS-MPC has attracted many experimental applications, and it haswell application prospect in the future multi-motor control field. In this paper, theFCS-MPC algorithm of the AFE based multi-motor system has been discussed, andthe drawbacks of the FCS-MPC have been analyzed. Moreover, the novel controlstrategies are proposed.An effective method for FCS-MPC algorithm simplification is proposed toreduce the running time without affecting the control performance. This method isdivided into two steps: the first step is to transform the multiple current predictionsinto single reference voltage prediction; the second step is to reduce the number ofcost function calculations by introducing the sector distribution and candidate voltagevectors selection method. Finally, with the proposed method, the time consumption inprediction and cost function calculation can be reduced, and the computationalrequirement of the controller will be much lower in multi-motor system.In the design process of the AFE converter’s controller, time delay effect is anissue that cannot be ignored. Compared with the classical linear control strategies, thecontrol period in FCS-MPC is much shorter, so the influence of time delay should beconsidered. In terms of the time delay caused by algorithm calculation, the currentcompensation equations for single period delay and non-integral period delay areconstructed, and the parameter determination method based on the cost function is given; In terms of the time delay caused by signals filtering, with the introduction ofthe state observation theory, a filter time delay observer is constructed based on themathematical models the AFE converter and signal filters. The experimental resultsprove that, the control performance of the FCS-MPC based AFE converter has beensignificantly improved with the proposed time delay compensation algorithm withoutchanging the circuit structure.Unlike the traditional modulation-based algorithm, FCS-MPC needs themathematical models of control objects to predict future currents, so the inaccuracy ofthe mathematical model will affect the control effect of the FCS-MPC. With theextend-state observer theory and the adaptive filtering theory, a hybrid parallelobserver (HPO) for multivariate observation is designed. In the controller of AFEconverter, the hybrid parallel observer can realize the source voltage sensorlesscontrol without knowing the values of the AC-side inductance and resistance, and thenovel starting algorithm ensured good transient performance in the starting process; inthe controllers of multi-motor drivers, the hybrid parallel observer realizes theestimation of the back electromotive force, stator inductance, and stator resistance.The parameter accuracy dependence of the multi-motor system is greatly decreased.