Aiming at the difficulty of fault feature extraction of fan bearing vibration signals, a fault diagnosis method of the fan bearing based on the multi-scale fuzzy entropy (MFE) feature extraction and combined with the sooty tern optimization algorithm (STOA) optimized support vector machine (SVM) is proposed. Firstly, the original vibration signals are collected and the multi-level fuzzy entropy is calculated. Secondly, the fault feature vector set is constructed as the input of SVM. Finally, the STOA is used to optimize SVM for classification and diagnosis of bearing faults. Simulation based on the bearing vibration data from Case Western Reserve University shows that the bearing fault diagnosis accuracy reaches 99.3%, which proves that the proposed method has high accuracy and effectiveness.