The fault identification of the motor rotor system is one of the effective measures to ensure the normal operation of the motor and the stable operation of the power system. This paper proposes a motor fault identification method based on empirical mode decomposition and support vector machine to achieve effective fault identification. Select three types of faults in the rotor system: rotor imbalance, rotor misalignment, and rotor dynamic static friction, compare and analyze them based on the signals of normal rotor operation. Using empirical mode decomposition to decompose the vibration signal of the rotor system, a series of intrinsic mode function components are obtained. The first five order IMF are selected for energy feature extraction, which is used as the feature vector input to establish a support vector machine model. The optimal penalty factor c and kernel function g are found through cross validation and grid search methods to identify the fault type of the rotor system. The experimental results show that the fault recognition method based on empirical mode decomposition and support vector machine can extract the characteristics of fault signals, classify the types of rotor fault signals well, and have high recognition accuracy. It can be well applied to the field of motor fault recognition