Andrew D. Dimarogonas
Dept. of Mech. Eng., Washington Univ., Box 1185, St. Louis, MO 63130
Vibration and acoustic monitoring of the operating conditions and incipient failures of rotating shafts and other machine members has been used extensively in the past 30 years. The shaft vibration spectrum was used to obtain information about unbalance, bearing, and other instabilities in rotating machinery since the 1920's. It was further identified in the 1960's as a source of information for the existence of cracks, misaligned couplings, loose parts, and other unwanted operating conditions. Such conditions would mainly introduce asymmetries of the rotating parts which would result in vibration spectra in the acoustic range which are characterized by the response of parametrically excited systems, including sub- and superharmonics. Such signatures have been used for rotating machinery condition monitoring. It was further observed that field singularities, such as cracks, couple the lateral, longitudinal, and torsional vibration of rotating shafts providing additional monitoring tools. Finally, it was found that the nonlinearities associated with closing cracks and gaps and journal misalignments lead to additional information on the condition of the rotating parts, such as higher harmonics. In recent years, the wealth of accumulated observation on the signatures of abnormal operating conditions in rotating machinery have been incorporated in expert systems for rotating machinery monitoring. Such systems have been developed using predicate logic or artificial neural networks. Further, fuzzy sets and interval analysis have been incorporated to account for nontraditional recording of operating conditions.