Identification of abnormal tribological regimes using a microphone and semi-supervised machine-learning algorithm

Abstract

Functional surfaces in relative contact and motion are prone to wear and tear, resulting in loss of efficiency and performance of the workpieces/machines. Wear occurs in the form of adhesion, abrasion, scuffing, galling, and scoring between contacts. However, the rate of the wear phenomenon depends primarily on the physical properties and the surrounding environment. Monitoring the integrity of surfaces by offline inspections leads to significant wasted machine time. A potential alternate option to offline inspection currently practiced in industries is the analysis of sensors signatures capable of capturing the wear state and correlating it with the wear phenomenon, followed by in situ classification using a state-of-the-art machine learning (ML) algorithm. Though this technique is better than offline inspection, it possesses inherent disadvantages for training the ML models. Ideally, supervised training of ML models requires the datasets considered for the classification to be of equal weightage to avoid biasing. The collection of such a dataset is very cumbersome and expensive in practice, as in real industrial applications, the malfunction period is minimal compared to normal operation. Furthermore, classification models would not classify new wear phenomena from the normal regime if they are unfamiliar. As a promising alternative, in this work, we propose a methodology able to differentiate the abnormal regimes, i.e., wear phenomenon regimes, from the normal regime. This is carried out by familiarizing the ML algorithms only with the distribution of the acoustic emission (AE) signals captured using a microphone related to the normal regime. As a result, the ML algorithms would be able to detect whether some overlaps exist with the learnt distributions when a new, unseen signal arrives. To achieve this goal, a generative convolutional neural network (CNN) architecture based on variational auto encoder (VAE) is built and trained. During the validation procedure of the proposed CNN architectures, we were capable of identifying acoustics signals corresponding to the normal and abnormal wear regime with an accuracy of 97% and 80%. Hence, our approach shows very promising results for in situ and real-time condition monitoring or even wear prediction in tribological applications.

A review of recent advances and applications of machine learning in tribology

“…331 CNNs, either in 1D, 2D, or 3D, which have shown significant potential to extract structural/imaging features to build their relationship with output in various research fields may prove valuable to assist these MD applications in tribology. [332][333][334][335][336] Researchers may find it challenging to train ML models at present since they may be unable to identify a combination of features that are appropriate for their learning objectives. Therefore, other deep learning (DL) methods like restricted self-organizing maps (SOMs), restricted Boltzmann machine (RBM), and autoencoders (both stacked and denoising) may assist researchers for feature map generation.…”

A review of advances in tribology in 2020–2021

“…Wear loss of plasma transfer arc (PTA) coatings on AISI 1020 steel was predicted by using an ordinary feedforward artificial neural network and basic, kernel-based, and weighted extreme learning machine trained with the data of microhardness measurements and laboratory dry sliding wear tests [432]. Pandiyan et al [433] tried to identify abnormal tribology events, such as scuffing, which are difficult to predict using physics-based numerical modelling, www.Springer.com/journal/40544 | Friction Fig. 33 Comparisons of wear volume vs shaft revolutions between experiment, physical simulation, and the data-driven ANN (LSTM) for two types of running-in [431].…”

Aplicación de herramientas de Inteligencia Artificial para la optimización frente al gripado del contenido de aditivos FM, EP y AW en un lubricante

“…Para establecer los parámetros de creación y entrenamiento de la RNA se siguen las indicaciones planteadas en otros trabajos [9,12,23] sobre el número de capas ocultas, número de neuronas, funciones de transferencia, algoritmo de entrenamiento, que permiten obtener una predicción más precisa. Para esta investigación se ha buscado una precisión mayor al 98 %, al igual que en otros trabajos similares [19]. Con la herramienta SVM se ha procedido de la misma manera, es decir, se han seleccionado los parámetros Kernel, función de entrenamiento, orden del polinomio de entrenamiento, BoxConstrain, y KFold para obtener una precisión similar.…”

“…En 2021 se ha propuesto una metodología, apoyada en algoritmos de aprendizaje automático ML (Machine Learning), capaz de diferenciar los regímenes de desgaste mediante señales de emisión acústica, consiguiendo una precisión de identificación del 97 % [19]. Ya en 2022, se ha desarrollado un modelo que permite clasificar los mecanismos de desgaste a partir de imágenes SEM y RNA, obteniendo una precisión de alrededor del 98% para los datos de entrenamiento, en torno al 72% para los datos de validación y alrededor del 73% para los datos de prueba [20].…”