The reliable operation of a switched reluctance motor (SRM) drive system requires the position sensor to provide reliable position signals. Therefore, given the problem that the Hall position sensor fails from time to time, which affects the normal operation of the motor, it is of great significance to study an effective position signal diagnosis method to improve the reliability of the system operation. In this paper, a fault diagnosis method combining time threshold and state prediction is proposed. Firstly, the type and location of position sensor faults are analyzed; secondly, the proposed method based on time threshold combined with state prediction is theoretically analyzed, which compares the actual state values of the three position signal combinations with the predicted state values in real time, and threshold constraints are applied in combination with the state value transition time to detect various kinds of faults and achieve fast and accurate results; finally, to validate the proposed method, a three-phase 12/8 structure is used to diagnose faults with a three-phase 12/8 structure. Finally, to verify the effectiveness of the proposed method, an experimental validation of the method is carried out with a three-phase 12/8 structure motor as the research object. The experimental results prove the feasibility and effectiveness of the method in fault detection. In addition, the method can be realized without complex calculations and additional hardware and can be extended to be used in drive systems such as brushless DC motors (BLDC).