Files
Abstract
Mechanical failure of rotating systems creates machinery issues in fatigue and unplanned maintenance on traditional and routine processes. Big data has provided an opportunity to investigate the feasibility of continuous predictive maintenance. Processes are explored for continuous data collection using Siemens MindSphere—real time visibility for manufacturing intelligence allows for proactive monitoring. Machine learning algorithms are a promising method to collect and diagnose signatures through vibration data and effectively facilitate machine monitoring-- exploratory data analysis currently favors FFT spectra transformation for predicting abnormalities in the frequency domain. By researching the relationships that exist between failure signatures to mechanical failure, FFT data is used to assess the type of failure. An architecture is proposed that may allow for rotating machinery to be diagnosed in real time for a dynamic testbed. Implementation of virtual modeling via a digital twin is also explored for considerations of future work in simulation assets.