Monitoring Local Faults in Aerospace-grade Gearboxes and Bearings Using an IoT Sensing Platform

This paper presents the design of experiments and corresponding results for how Ridgetop Group Inc. (Ridgetop) utilized a collection of advanced diagnostic and prognostics tools to detect, localize, and assess critical mechanical faults in rotating system components such as a gearbox or a bearing. The technical approach was based on using Ridgetop’s existing product lines to fulfill as many as possible of the eight functional elements of the IEEE 1856-2017 Standard Framework for Prognostics and Health Management of Electronic Systems: (1) Sensors (S), (2) Data Acquisition (DA), (3) Data Manipulation (DM), (4) State Detection (SD), (5) Health Assessment (HA), (6) Prognostic Assessment (PA), (7) Advisory Generation (AG), (8) Health Management (HM). To achieve functional elements 1-2, Ridgetop utilized an Internet-of-Things (IoT) DA toolbox comprising of wireless accelerometers, a Linux based single board computer and gateway system with Multiple Input Multiple Output (MIMO) connectivity, and the Sentinel MotionView software platform for data acquisition, analysis, and sensor to gateway management. This sensing and data acquisition system was deployed on a customized testing apparatus with rotating gears that was built to acquire raw vibration data signals using multiple MEMS-based accelerometers, with a ±58.86 m s
²
measurement range, mounted on the bearing and the shaft. To achieve functional elements 3-6, a custom set of Python scripts was developed to perform Feature Data (FD) extraction based on Time-Domain and Frequency Domain vibration analysis methods that was then post processed with Ridgetop’s Adaptive Remaining Useful Life Estimation (ARULE) software that produces key prognostic estimates for State-of-Health (SoH), Remaining Useful Life (RUL), and Prognostic Horizon (PH). Functional elements 7-8 were not realized in the executed design of experiments where highly accelerated life testing (HALT) was performed on different sets of bearings and gears to extract and analyze the fault-to-failure progression signatures of critical gear faults. The results discussed in this paper have yielded a detailed approach for how all 8 functional elements can be further integrated into a comprehensive Prognostic Health Management (PHM) toolbox using the key prognostic estimates of SoH, RUL, and PH to realize advisory generation and scheduling of health management actions.