When diagnosing broken rotor bar faults in squirrel-cage motors using motor current signature analyses (MCSA), diagnostic accuracy suffer from insufficient characteristic information. This paper proposes a broken rotor bar fault diagnosis method combining Kernel Principal Component Analysis (KPCA) and Support Vector Machine optimized by Sparrow Search Algorithm (SSA-SVM). Firstly, 45 characteristic quantities suitable for current signal analysis are extracted from time domain waveform and spectrum of the phase current. The KPCA is used for feature extraction and data reduction from original features. Then, SSA is used to optimize the parameters of SVM, and a fault intelligent diagnosis model based on SSA-SVM is constructed. Finally, the selected feature vectors are input into the SSA-SVM model for classification and recognition. The proposed method is experimentally verified on a 3kw motor test platform. The experimental results show that the proposed method can diagnose rotor bar breakage faults in motors under different operating conditions and different severity levels. The fault diagnosis accuracy of the SSA-SVM model is 99.67%, which has higher diagnostic performance compared to other diagnostic algorithms.