Statistic-based method to monitor belt transmission looseness through motor phase currents

Abstract

Belt–pulley systems are widely used in the industry due to their high efficiency and their low cost. However, only few works exist about the monitoring of their degradation. This paper details the impact of belt looseness on electrical measurements under steady and transient state in order to identify spectral signatures. This analysis enlightens the advantage of the transient state to detect belt looseness because it exacerbates belt slip. An innovative methodology is then proposed based on the application of a square-wave speed reference in order to monitor belt looseness. A statistical-based indicator is defined from the phase currents in order to automatically detect drifting of the indicator. A normalization process is also applied to increase the detection robustness. The proposed indicatoris evaluated on a 30-kW induction machine and a direct current machine coupled with two trapezoidal belts for three speed and four load conditions. It reaches very good results with almost 90% correct detections for 1% false alarms. These results are way better than those obtained with a classic spectral analysis during the steady state. Moreover, results demonstrate that higher load conditions are more accurate for the monitoring of belt looseness. Index Terms—Belt–pulley systems, fault diagnosis, induction motors (IMs), robustness, spectral analysis, statistical-based indicator, torque–speed segmentation.