Triboinformatics Approach for Friction and Wear Prediction of Al-Graphite Composites Using Machine Learning Methods

Abstract

Data-driven analysis and Machine Learning (ML) algorithms can offer novel insights into tribological phenomena by establishing correlations between material and tribological properties. We developed ML algorithms using tribological data available in the literature for predicting the coefficient of friction (COF) and wear rate of self-lubricating aluminum-graphite (Al/Gr) composites. We collected data on effects of material variables (graphite content, hardness, ductility, yield strength, silicon carbide content, and tensile strength), processing procedure, heat treatment and tribological test variables (normal load, sliding speed, and sliding distance) on tribological properties and established two-parameter relationships. This data is analyzed using several ML algorithms: Artificial Neural Network (ANN), K Nearest Neighbor (KNN), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), and Random Forest (RF). The trained ML models can predict the tribological behavior from material variables and test conditions, beyond what is possible from two-parameter correlations. GBM outperformed other ML algorithms in predicting friction behavior, while RF had the best prediction of the wear behavior. ML analysis identified graphite content and hardness and as the most significant variables in predicting the COF, while graphite content and sliding speed were the most dominant variables for wear rates.

Current Trends and Applications of Machine Learning in Tribology—A Review

“…Obtained results and the suggested IBA-ANN approach can thus help to save resources when searching for beneficial stress or material combinations with limited experimental database. Very recently, Hasan et al [63,64] compared five different ML techniques when predicting the friction and wear behavior of aluminum base alloys and graphite composites: ANN, kNN, SVM, gradient boosting machine (GBM), and RF. The 852 data sets were obtained from experimental studies in literature.…”

“…A few studies, however, manage to extract real insights and thus additional knowledge from a large and broad database. The comprehensive works in the field of composite materials from Kurt and Oduncuoglu [52], Vinoth and Datta [53], and Hasan et al [63,64] utilizing literature-extracted databases may be highlighted here and can serve as excellent examples. The current showstopper is still the availability of sufficient and comparable datasets as well as the handling of uncertainties regarding test conditions and deviations.…”

A Review of Application of Machine Learning in Design, Synthesis, and Characterization of Metal Matrix Composites: Current Status and Emerging Applications

“…Usage of machine learning techniques to synthesize new MMCs with the desired surface and tribological characteristics is a technology that could become part of the evolving field of ''Triboinformatics'' or ''Intelligent Tribology'' corresponding to the application of machine-learning methods to surface science and engineering. 43,44 Table II lists the literature in which machine learning techniques are applied to study the tribological properties of different MMCs. As can be seen, for tribological properties in addition to attributes related to the samples such as reinforcement volume percentage, size, chemical composition, and post-/processing parameters, tribological test factors like applied load, sliding distance, sliding velocity, temperature rise, hardness of the counter-body, etc., were considered as input variables in the developed model.…”

Data‐driven modeling for predicting tribo‐performance of graphene‐incorporated glass‐fabric reinforced epoxy composites using machine learning algorithms

“…Again, this shortcoming of inadequate database is on account of limited experimentations. Md Hasan et al, [ 22 ] in their work on modeling of aluminium graphite composites, used five ML models. Their findings successfully showed better precision and predictive performance as indicated by the R 2 value of 95.46% for GBM, of 95.03% for RF, of 94.50% for ANN, of 93.11% for Support Vector Machine (SVM) and of 87.7% for K‐Nearest Neighbor (KNN).…”