Ball Bearing Diagnosis via Spectral Analysis of Short Vibration Records Using the Maximum Entropy Method

To enable efficeint vibration diagnosis of rotating machinery in industry settings, we investigate a spectral analysis approach for accurately estimating the envelope spectrum from short-duration vibration signals, thereby reducing data traffic. We propose the use of the maximum entropy method as the spectral estimation algorithm for short-record vibration data and demonstrate that it achieves significantly higher accuracy than the commonly used Fast Fourier Transform. Through analysis of both simulated and measured vibration data, we show that this approach can reduce the data required for vibration diagnosis to approximately one-tenth of that required by conventional methods.