Abstract: A motor fault diagnosis method based on the decision tree algorithm is proposed with the aim of enhancing the efficiency and accuracy of motor fault diagnosis. The method integrated the theory of feature extraction to extract pertinent features associated with the operating states and faults of the motor. The C4.5 algorithm, a decision tree algorithm, was employed for machine learning utilizing the extracted spectral features of motor operation. To mitigate the issue of overfitting in the decision tree model, practical measures including deep optimization pruning strategies and dynamic splitting threshold methods were employed for refinement. Experimental findings indicate that the proposed method attains an impressive accuracy rate of 96% for both prediction and diagnosis, thus highlighting its substantial practical value.