Incipient wear fault diagnosis using a modified change detection method

To recognize the incipient wear of machine parts, the normalization pretreatment and scale independence processing were proposed to improve the conventional method. The frictional vibration signal was analyzed by the improved method to recognize the incipient wear fault. The results show that there are two types of changes in the wear process, i.e., (i) pure dynamic change and (ii) dynamic change accompanied by amplitude change. The analysis on vibration signal and surface topography demonstrates that the former type indicates incipient wear fault, and the latter type indicates wear failure. Preventative maintenance should be carried out as soon as the pure dynamic change is detected. This method makes sense to prevent catastrophic system failure.