Envelope deformation in computed order tracking and error in order analysis

Abstract

The defect diagnosis of rolling bearings operating under time-varying rotational speeds entails an integrative approach involving envelope analysis and computed order tracking that converts a vibration signal from the time domain into the angle domain to eliminate the effect of speed variations. When a signal is resampled at a constant angular increment, the amount of data padded into each data segment will vary, depending on the rate of change in the rotational speeds. This leads to changes in the distance between the adjacent impulse peaks and consequently the results of order analysis. This effect is particularly prominent when the rate of speed change is significant. This paper presents a quantitative analysis of key factors affecting the accuracy of order analysis on rotating machines under variable speeds. An analytical model is established, simulated, and experimentally evaluated. The effects of speed variation, instantaneous speed, angular interval between impulses, and the peak time of the impulse are specified. It is concluded that the error in the order analysis will increase as the acceleration increases. Furthermore, the error is larger under low speeds.