Combination of FFT & ICA methods for faults analysis of rotating machine

Abstract

The vibration analysis using Fast Fourier Transform FFT is a common and widely used method, because of the similarity between faults signatures in this analysis a question is raised to what is the most effective ways to distinguish between different faults. In order to overcome this difficulty other method combined with the FFT was used. In this paper, a new combination will be present to overcome this situation. Independent Component Analysis (ICA) is used in combination with FFT to identify faults in rotating machines. The vibration is measured through multichannel vibration data acquisition system. The signals are then analyzed using ICA and finally, FFT is applied on ICA components. The extraction features give the best signature to identify each fault from others. This method is used for detecting more general faults occur in rotating machine (bearing fault, misalignment, unbalance, shaft fatigue), and can identify the similarity between faults. The interaction between different types of faults can be solved effectively by using ICA.