Scott
If you are already collecting vibration data in this fashion then you would normally utilize anomaly detection. This would require some type of training period and a system capable of doing it. We'll sometimes generate custom ML models for various applications or use AWS (although MS and IBM cloud services provide similar) for anomaly detection of overall data. An expert system may actually suit your situation as anomaly detection (ML/AI) will not necessarily identify existing defects (actually, it won't and don't let them tell you otherwise).
However, your opportunity may fit better with an expert ESA system. Not load or operating dependent, will detect incoming power, machine, and load conditions - electrically and mechanically - and vibration analysts can pick up on additional prognostics (ie: verification). As @Mark Kingkade mentioned, we utilize the Framatome EMPATH / ECMS ESA system. Happy to discuss.
------------------------------
Howard W Penrose, Ph.D., CMRP
Random Past SMRP Chair (2018), 2019+ Govt Relations Smart Grid, Infrastructure and Cybersecurity Working Group Chair,
Chair Technical Standards wind, solar, energy storage, American Clean Power (formerly AWEA), and
President
MotorDoc LLC
Lombard, Illinois
------------------------------