Research On Key Technologies Of Model Predictive Control For Permanent Magnet Synchronous Motors

With the merits of high torque density,high power density as well as high efficiency,permanent magnet synchronous motors(PMSM)are widely used in industrial production,transportation,aerospace and defense industry,etc.For PMSM control,model predictive control(MPC)has significant advantages over other control strategies.MPC is an advanced control strategy,which provides high dynamic responses while handling the constraints systematically.However,the research on MPC for PMSM is still in its early stage.Poor parameter-robustness,difficulties in the design of weighting factors as well as relatively high current ripples need to be addressed.In this dissertation,predictive current control(PCC)and predictive speed control(PSC)of PMSM in terms of parameter-robustness improvement,design of weighting factors as well as suppression of current ripples are studied.Finite control set predictive current control(FCS-PCC)is model-dependent and current steady-state errors(SSEs)occur in case of parameter mismatches.The influence of parameter mismatches on FCS-PCC is analyzed and the strategy which reduces the SSEs is investigated.To reduce current SSEs and improve parameter-robustness,integral-FCS-PCC(I-FCS-PCC)strategy is proposed where the cost function is designed in proportional-intergral(PI)form.The integral actions are only activated as the system approaches the predefined operation range.In this manner,fast dynamic performance is maintained.Moreover,the design of the integral coefficients is facilitated.Experimental results show that with the proposed I-FCS-PCC,the parameter-robustness is improved and current SSEs are effectively reduced.Finite control set predictive speed control(FCS-PSC)is model-dependent and speed SSE is commonly observed due to parameter mismatches.To reduce the speed SSE,incremental reduced-order model of PMSM is derived and integral-FCS-PSC(I-FCS-PSC)is proposed subsequently.With the incremental reduced-order model,the prediction process is less dependent on rotor flux linkage.By introducing the integral terms of equivalent speed error rather than the actual speed error,speed overshoots are avoided.Experimental results show that with the proposed I-FCS-PSC,the speed can track its reference accurately without overshoots.Moreover,speed SSE is effectively reduced even in case of parameter mismatches.In FCS-PSC,the design of the cost function is generally time consuming.The selection of norm is critical to the system performance and the tuning of weighting factors is challenging.The reason is the large magnitude difference between the speed and currents.To address this issue,the selection of norm and the design of weighting factors in FCS-PSC are studied.The impact of the norms on the system performance is analyzed.The design of the weigting factors is facilitated through the normalization of the equivalent speed tracking error.Moreover,FCSPSC without weighting factors is proposed,where the speed control is transformed into the control of the q-axis current.In FCS-PSC,the current ripples are relatively large.The reason is that only one switching state is applied to the inverter each sampling period.To suppress the current ripples,modulated predictive speed control(MPSC)strategy is proposed.The constrained optimization problem of predictive speed control is formulated and analyzed.The voltage and current constraints are transformed into respective incremental voltage constraints,which facilitates the handling of constraints.Based on Karush-Kuhn-Tucker conditions,the constrained optimal reference voltages are deduced.The duty-cycle of the inverter switches are calculated based on the optimal reference voltages.With the proposed MPSC strategy,the total harmonic distortions(THD)of phase currents are reduced while high dynamic performance of the speed is achieved.