Methodology for Diagnosing Ball Bearing Wear （方法诊断滚珠轴承磨损）.pdf

Methodology for Diagnosing Ball Bearing Wear By: Bill Watts, P.E. and Mike Johnson, P.E. 1 | Introduction Probably no other specific machine fault diagnosis is as common and as important to machine life and reliable operation as ball bearing wear. With respect to both manual analysis and automatic diagnostic systems results, it is crucial that an analyst or system user recognize the logic behind this diagnosis. Although the topic is ball bearing wear, it should be more generally referred to as rolling contact bearing wear. The same discussion applies to cylindrical roller bearings. For slow speed bearings (less than ~300 rpm), the narrow band logic discussed in this paper is sometimes valid; however in many cases other techniques that directly measure shock energy are more accurate. 2| General Characteristics in Spectral Data The geometry of a ball bearing (inner and outer race diameters, ball diameter, and number of balls) determines a small set of calculated fundamental bearing frequencies that are best expressed in terms of multiples (orders) of shaft rotational rate. Bearing wear or damage is generally characterized by one or more fundamental bearing tones (non- integer multiples of shaft rotational rate), harmonics (multiples of those frequencies), and/or shaft or bearing cage rate sidebands, and/or abnormal “humps” of random noise. Some analysts like to rely on knowing the exact fundamental tones and tracking the amplitudes of those specific tones. However, the exact model of bearing in a component is not always known and bearing tone frequencies can shift over time due to bearing deformation. Also consider that often the fundamental tones are low or even non-existent. The primary evidence of bearing wear may lie in higher frequency harmonics, sidebands and/or elevated random noise. Bearing wear or damage can occur and progress in an infinite variety of ways. Thus it is reasonable tha