Asynchronous motor fault vibration signals have strong nonlinear characters that easily lose their nonlinearity with traditional linear methods, resulting in an impact fault diagnosis effect. Hence, a fault diagnosis method based on kernel principal component analysis (KPCA) and particle swarm optimization support vector machines (PSOSVM) is proposed. First, the kernel function is used to realize the nonlinear mapping from the original space to higher-dimensional space, and perform principal component analysis (PCA) on the mapping data. The nonlinear principal components of the original sample are then obtained; feature extraction and data compression are realized. The SVM uses the kernel principal features for pattern recognition. An optimizing method with a distance ratio and particle swarm algorithm is used for parameter optimization of the KPCA and SVM, respectively. The experimental results indicate that the method can effectively extract nonlinear features of a fault signal and perform well in nonlinear pattern recognition. Compared with the PCA and SVM methods, it has good classification effect and strong timeliness, both of which can quickly and effectively diagnose asynchronous motor faults.