Double Parallel Rotor Permanent Magnet Synchronous Motors exhibit superior performance and compact size, but the growing trend of electrification imposes higher demands on them. This study proposes a predictive fault-tolerant control integrating a closed-loop identification model and conducts experiments on Double Parallel Rotor Permanent Magnet Synchronous Motors. Results indicated that the proposed closed-loop identification model, along with its fractional-order lead-lag compensator module, effectively optimized motor performance, reducing average tracking error by 78.36%. Additionally, with demagnetization faults, the predictive fault-tolerant control outperformed traditional fault-tolerant control in speed, current, and torque fault-tolerant control, demonstrating superior performance. Through 10 weeks of practical application records, Double Parallel Rotor Permanent Magnet Synchronous Motors achieved a working accuracy of 95%-99% under the closed-loop identification model, with recall rates reaching 92%-96% in faulttolerant scenarios. In both natural and simulated demagnetization fault situations, 97.69% of Double Parallel Rotor Permanent Magnet Synchronous Motors could continue normal operation. This research holds positive significance for the development of motor systems and enhancing their adaptability in the trend of electrification.