【目的】为解决永磁同步电机(PMSM)模型预测电流控制(MPCC)的参数失配问题，采用增量式MPCC与模型参考自适应系统(MRAS)进行参数辨识，以提高控制系统的参数鲁棒性。【方法】首先建立基于增量式模型的MPCC系统以消除磁链参数，同时对增量式MPCC的参数敏感性进行分析，其中电阻参数失配对电流预测效果影响很小，可忽略不计。因此基于dq轴电流，采用MRAS仅辨识dq轴电感，从而避免欠秩问题。【结果】基于Matlab/Simulink与STM32平台进行了仿真分析与实时性验证。结果表明，MRAS可精确辨识dq轴电感参数；虽然增量式MPCC与MRAS控制模型需要设置电阻参数，但电阻参数失配对dq轴电感的辨识结果影响较小；加入的MRAS参数辨识系统并不额外增加较大的计算负担，具有较好的实时性表现。【结论】对于PMSM的dq轴电感、磁链与电阻4个参数，采用增量式MPCC消除了磁链参数，而电阻参数无需精确辨识，仅需通过MRAS辨识dq轴电感。MRAS控制模型和增量式MPCC的电阻参数失配对控制性能基本无影响，从而有效提升了PMSM MPCC参数失配鲁棒性。

[Objective] To solve the parameter mismatch problem of model predictive current control (MPCC) for permanent magnet synchronous motor (PMSM), incremental MPCC with model reference adaptive system (MRAS) is used for parameters identification in order to improve the parameter robustness of the control system. [Methods] Firstly, the MPCC system based on the incremental model was established to eliminate the flux linkage parameter, and the parameter sensitivity of the incremental MPCC model was also analyzed, in which the resistance parameter mismatch had a small and negligible influence on the current prediction effect. Therefore, based on the dq-axis current, MRAS was used to identify only the dq-axis inductance, thus avoiding the under-rank problem. [Results] Simulation analysis and real-time verification were carried out based on Matlab/Simulink and STM32 platform. The results showed that MRAS could accurately identify the dq-axis inductance parameters. And although the incremental MPCC and MRAS control models need to set the resistance parameter, the mismatch of the resistance parameter had less impact on the identification results of the dq-axis inductance. The added MRAS parameter identification system did not additionally increase the large computational burden, and it had a better real-time performance. [Conclusion] For the four parameters of dq-axis inductance, flux linkage and resistance of PMSM, the incremental MPCC eliminates the flux linkage parameter, while the resistance parameter does not need to be identified precisely, and only the dq-axis inductance need to be identified by MRAS. The mismatch of the resistance parameter of the MRAS control model and the incremental MPCC has no effect on the control performance, which effectively improves the MPCC parameter mismatch robustness of PMSM.