Abstract: To handle the speed sensor fault identification for a bearingless induction motor (BL-IM), a fault diagnosis control strategy based on Back Propagation Neural Network (BPNN) was proposed. Firstly, the torque, phase current and other signals of the BL-IM were selected as the basis for the BPNN sensor fault diagnosis. The fault data samples such as the torque of the sensor under different faults were used to continuously train and learn BPNN, so as to improve the accuracy of fault diagnosis and fault classification. Secondly, the speed sensorless fault-tolerant control system was established by using the fractional order model reference adaptive control (FO-MRAS), completing the switch from the fault system to the fault-tolerant control system. Finally, the normal operation of the BL-IM under sensor fault was realized. The simulation and experimental results show that the proposed BPNN fault diagnosis system can realize the accurate identification of speed sensor faults in no-load and on-load operation, and the fault-tolerant control system can significantly reduce the influence of sensor faults on the speed. At the same time, the motor suspension rotor has good suspension characteristics. The security and reliability of the BL-IM can be improved.