Fault Diagnosis of Rotating System Based ICA-SVM

ICA (independent component analysis) is used to the fault diagnosis of rotating system. The architecture of diagnosis system based on denoising and blind source separation is proposed. SVM (support vector machine) network is adopted for fault study training and recognition. Using the signal preprocessed by ICA based different contrast function in denoising and blind source separation, the constraint knowledge newly can be easily added to the diagnosis system. The vibration source signal is got by ICA with constraints, and the kernel function of SVM is radical base function (RBF). Simulation results show that the method in this paper has good performance