Abstract: Based on nonlinear mapping relationship between fault symptom and fault type in control electric component, RBFNN(radial basis function neural network) approach was presented for fault diagnosis. Fault mechanism and failure behavior of control electric component was analyzed, then featured fault types were extracted from control electric component failures and the extracted features were regarded as fault symptom eigenvector. The process of fault diagnosis principal, fault diagnosis model and fault diagnosis algorithm was given using RBFNN with enough fault feature information. Trained RBFNN was used for fault vector recognition and diagnosis to verify the proposed fault diagnosis model effectiveness and rationality. Simulated result shows that RBFNN can overcome the limitation of local infinitesimal during fault diagnosis process, and the requirement for fast diagnosis rate and high diagnosis precision can be met.