Improved bearing sensing for prognostics: from vibrations to optical fibres

Abstract

Bearings are vital elements in rotating machinery. Despite their wide use, bearings are prone to failures that can result in irreversible damage. The detection of bearing damage in its incipient stages and monitoring of the fault severity are required for the optimisation of maintenance decisions. The classical methods for bearing monitoring are based on an analysis of the vibration signals usually captured by an accelerometer located on the machine case. Two difficulties arise when diagnosing bearings via vibrations: the effect of the transmission path to the sensor distorts the signals and the low signal-to-noise ratio characterising the weak bearing signals in the presence of the accompanying strong surrounding noise originating from the vibrations of other rotating components in the machine. The goal of the presented work was to study the possibility of overcoming the current problems of bearing prognostics by locating the sensors as close as possible to the bearing. Two types of sensor were selected: micro electro-mechanical system (MEMS) accelerometers and optical fibres to measure strain. Due to their small dimensions, these sensors can be embedded into the system close to, or even inside, the bearing. The measured signals from bearings with various spall widths show an improved signal-to-noise ratio, demonstrating the power of these two local sensing methodologies. By avoiding transmission path effects, the results show the clear detection of early defects. The results of this study open new options for monitoring and detecting the early signs of failure in critical bearings.

Repeatability of FBG-based damage detection on bearings

“…In order to investigate the repeatability of the fault detection methodology proposed in literature [1][2][3][4][5] for measurements with sensors installed directly on the bearing, we performed two repeatability experiments. First, we used one healthy bearing and one faulty bearing and installed them in the rotating setup 5 times, measuring the strain signals in the same conditions, and second, we used one healthy bearing and one faulty bearing and sequentially installed 5 optical fibers sensors on each bearing and performing strain measurements in the same conditions after removing the previous fiber.…”

“…Due to their size, these types of sensors cannot be instrumented directly onto the bearing itself. They are therefore often placed close to the component of interest, such as on the bearing housing 1,2 , as shown in Figure 1. This implies that they cannot accurately detect and locate the fault on the bearing.…”

“…Although OFS have been demonstrated to be able to detect bearing faults, this has mostly been done by placing the fibers on the bearing housing [1][2][3][4][5] . The bearing faults that have been studied are typically also quite large (>2 mm) and therefore not representative of early damage 1,2 .…”

“…Although OFS have been demonstrated to be able to detect bearing faults, this has mostly been done by placing the fibers on the bearing housing [1][2][3][4][5] . The bearing faults that have been studied are typically also quite large (>2 mm) and therefore not representative of early damage 1,2 . Moreover, the demonstrations are often one single measurement, where no investigation of the repeatability and robustness of the method has been performed.…”

Severity Estimation of Faulty Bearings Based on Strain Signals From Physical Models and FBG Measurements

“…Fault diagnosis of bearings via FBG sensors has not been thoroughly studied in the literature (Khmelnitsky et al, 2015;Wei et al, 2016;Alian et al, 2019). Numerous researchers have developed physics-based models to investigate the RE-edge impact reaction (Arakere et al, 2009;Branch et al, 2009Branch et al, , 2013First Author et al This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 United States License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.…”