Abstract: Model predictive control has been widely used in multiphase motor control systems because of its advantages, such as simple concept, quick response, and easy handling of nonlinear constraints. However, the present model predictive control only considers the electromechanical energy conversion of α-β subspace, and doesn’t consider the harmonic loss caused by the converter dead-time effect and other non-linear factors. In this paper, a dual subspace duty cycle model predictive current control strategy is proposed for dual three-phase permanent magnet synchronous generator rectification system. With the proposed strategy, the α-β subspace and x-y subspace are controlled simultaneously during a sampling cycle, so the harmonic currents are effectively suppressed. Based on this, a virtual vector duty cycle optimization strategy is introduced in the two subspaces to further improve the steady-state control performance. Finally, the feasibility and effectiveness of the proposed control strategy are verified by experiments.