A personal-computer-based system has been developed as a prototype of electronic signal-processing and computing systems for diagnosis of turbines, motors, and other rotary machines through analysis of acoustic emissions. The system includes acoustic sensors that respond in the kilohertz-to-megahertz frequency range wherein acoustic energy tends to emanate from bearing defects. The system also includes a relatively inexpensive desktop computer and a 12-bit analog-to-digital converter that can acquire acoustic-sensor readings in as many as four channels at a sampling rate up to 5 MHz per channel.
The reason for choosing the kilohertz-to-megahertz frequency range is that when the digitized acoustic-emission (AE) signals in this frequency range are processed by suitable algorithms to obtain indications of deterioration of bearings, the resulting indications are more reliable than are those obtained by similar processing of the outputs of accelerometers that sense lower-frequency vibrations. The algorithms in question implement envelope analysis (see figure) and another signal-processing technique called "point process spectral analysis" (PPSA).
In a typical application, vibration sources other than a defective bearing that one seeks to diagnose can be expected to give rise to a variety of anomalous transient events, including random spikes, in the AE signal. These anomalous transient events can mask bearing signatures and prevent reliable detection of bearing faults. A common obstacle encountered in analyzing an AE signal is the inability to determine which transient events are related to shaft rotation and which are not. Conventional time-series and spectral analyses are ineffective in distinguishing between useful and useless signal features. In addition, a bearing defect in its early stage generates intermittent transient impulses with weak periodicities. It is difficult to detect these periodicities through conventional envelope spectral analysis. Furthermore, the need to process a high-frequency AE signal in real time imposes a severe computational requirement, which constitutes an impediment to the design of an inexpensive bearing-diagnostic system.
PPSA was summarized in "Acoustic-Emission Bearing-Fault Diagnosis System" (MFS-26468), NASA Tech Briefs, Vol. 21, No. 11 (November 1997), page 84. PPSA provides high computational efficiency in the spectral analysis of transient events in a high-frequency AE signal. In turn, this computational efficiency enables the prototype bearing-diagnostic system to operate with a minimal processing gap (time delay), so that it can generate diagnostic indications in nearly real time. When pulses from sources other than a defective bearing corrupt the bearing-generated transient pulses, PPSA can provide a detection-and-discrimination capability superior to that of envelope analysis.
The prototype bearing-diagnostic system is programmed with an additional post-test vibration-signal-analysis software package called "PC-SIGNAL." This software is effective and easy to use for general applications in analysis of vibrations, monitoring the "health" of machinery, and diagnosis of faults. PC-SIGNAL also includes a number of subprograms designed especially for processing high-frequency AE signals.
This work was done by Jen-Yi Jong of AI Signal Research, Inc., for Marshall Space Flight Center. For further information, please contact the company at www.aisignal.com or (256) 551-0008.
Inquiries concerning rights for the commercial use of this invention should be addressed to
the Patent Counsel
Marshall Space Flight Center; (256) 544-0021
Refer to MFS-31468.
ORIGINAL URL - /Briefs/Aug00/MFS31468.html