Modelling of tribological responses of composites using integrated ANN-GA technique

Abstract

In the present article, artificial neural networks (ANNs) and genetic algorithm (GA) methodology were integrated to model tribological characteristics of stir-cast Al-Zn-Mg-Cu matrix composites under two-body abrasion considering large numbers of experimentally generated results. Tribo-responses of wear rate (Wrt), coefficient of friction (COF) and roughness of abraded surface (RAS) were evaluated under wide range of intrinsic ( i.e., particle quantity) and extrinsic ( i.e., abrasive size, load, distance and velocity) input parameters. Characteristics of Wrt, COF and RAS are often mutually contradictory in nature and so, multi-objective optimization technique becomes imperative for selection and design of machine components. Accordingly, those were optimized through Pareto solutions. Sensitivity of different factors was analyzed on each of the tribo-performances and validated via experimental evidences. Amongst the input variables, particle quantity and abrasive size dominated significantly over other variables except load which imparted modest influences. The role of various input parameters was explained through determination of different micromechanisms via exhaustive post wear characterizations, microstructural and surface topography attributes. Lowest values of Wrt and COF with a modest value of RAS were identified at 15 ± 2 wt.% particle quantity.

The Use of Artificial Intelligence in Tribology—A Perspective

“…In addition, various researchers have applied ML and AI approaches to predict and optimize the tribological behavior of different materials and operating conditions with manifold applications in mind [14][15][16][17][18][19][20]. For instance, Alambeigi et al [21] investigated the accuracy of predictive AI models by comparing them with experimental results obtained by testing the dry sliding contact of sintered steels [21].…”

Process optimization for enhanced tribological properties of Al/MWCNT composites produced by powder metallurgy using artificial neural networks

“…Unlike traditional statistical analysis methods, ANN has been reported to be successful in the analysis of complex processes [32]. This machine learning method, which allows to advance the algorithm by using the relationship of the information entering and leaving the training system, is effectively used in different fields such as system design, machining, and tribology [33,34]. Thanks to the trained mathematical model, it can produce meaningful results about new data.…”

High‐Stress Abrasive Wear Analysis of in Situ TiC‐Reinforced Zinc–Aluminum Composites Using Integrated Taguchi–TOPSIS Method

“…As a result, optimization methodologies may be the most practical alternative for determining the optimal processing conditions. Various techniques like Grey relational analysis (GRA), [9,10] genetic algorithm (GA), [26,27] artificial neural network (ANN), [28][29][30] response surface methodology (RSM), [31][32][33][34] fuzzy TOPSIS, [35,36] analysis of variance (ANOVA), [37] can be employed to address a range of these problems.…”