Tracking fundamental frequency for synchronous mechanical diagnostic signal processing

Vibrational and acoustic signatures of rotating machinery are often dominated by harmonics of a fundamental frequency related to the rotation speed (i.e., RPM) of one or more of the machines components. Many standard signal processing approaches such as cyclostationary models are developed assuming a constant fundamental frequency. Even slight variations in the RPM can cause significant reduction in signal processing gain and fault detection/classification performance. Methods to track the fundamental frequency, developed originally for USA Army battlefield acoustic signatures, are presented, analyzed and applied to the NRaD firepump dataset. In addition, a visual aid technique, termed the cyclostationary image is presented that provides insight into the fundamental frequency variations and features of the time domain signature.