A Comparison Between Hilbert Transform and a New Method for Signal Enveloping

Envelope analysis of vibration signals is a well known tool for amplitude demodulation and diagnosis of a number of vibration problems in machines and structures. The typical application is the fault diagnosis in the anti-friction bearings and gearboxes. Hilbert transformation (HT) is often used to extract the envelope signals (upper and lower) from a time domain signal. However it is observed that the envelope signals obtained by the HT are not always without any error. In this paper, 4 different signals, 3 simulated; sine, amplitude modulated and random, and measured vibration data on anti-friction bearings are analyzed using the HT. The paper compares the accuracy of the envelope data obtained for these different signals using the HT. A new method called three-point moving window (TPMW) method is also developed to generate envelope signals and applied to the 4 different signals that gives better results.