[Objective] An improved method based on an optimized control set is proposed to address the issues of harmonic currents and torque ripple in the model predictive control (MPC) system of an open-end winding six-phase permanent magnet synchronous motor under single upper-arm faults. [Methods] Based on the vector space decoupling theory, the voltage vectors under single upper-arm faults and their distribution characteristics in both fundamental and harmonic planes were derived and analyzed. To ensure high voltage utilization, virtual vector synthesis was implemented. Addressing the challenge of selecting appropriate basic voltage vectors using conventional convex hull methods, an optimized control set-based model predictive fault-tolerant control (MPFTC) method was proposed in this study. Through post-fault vector characteristic analysis, an optimized control set was selected from the G5 vector group, and its impact on harmonic currents and torque ripple performance was evaluated. Comparative simulations between conventional MPC and the proposed optimized control set-based MPFTC method were conducted under fault conditions. [Results] The simulation results demonstrated that the proposed MPFTC method effectively reduced the total harmonic distortion of the current from 50.54% to 5.96% under fault conditions, with the third-order harmonic component suppressed below 5%. Additionally, the electromagnetic torque exhibited only a transient drop of 1 N·m and recovered within 10 ms, successfully mitigating the untreated 50 Hz oscillation with an amplitude of ±50 N·m. [Conclusion] The proposed MPFTC method demonstrates significant effectiveness in harmonic suppression and torque ripple control, verifying its strong fault tolerance and engineering application value.