Polyphase Order Analysis Based on Convolutional Approach

Abstract

The condition of rotating machines can be determined by measuring of periodic frequency components in the vibration signal which are directly related to the (typically changing) rotational speed. Classical spectrum analysis with a constant sampling frequency is not an appropriate analysis method because of spectral smearing. Spectral analysis of vibration signal sampled synchronously with the angle of rotation, known as order analysis, suppress spectral smearing even with variable rotational speed. The paper presents optimised algorithm for polyphase order analysis based on non power of two DFT algorithm efficiently implemented by chirp FFT algorithm. Proposed algorithm decreases complexity of digital resampling algorithm, which is the most complex part of complete spectral order algorithm.