Data-Driven Prediction and Experimental Analysis of Wear in Micro/Nano SiO2-Reinforced Aramid Composites

Authors : Raşit Koray Ergün, İsmail Bayar, Hüseyin Köse

Pages : 425-441

Doi:10.55546/jmm.1797773

View : 84 | Download : 140

Publication Date : 2025-12-26

Article Type : Research Paper

Abstract :In this study, the wear behavior of samples obtained by adding micro and nano-sized SiO2 particle additives to aramid fiber reinforced polymer matrix composite materials at different rates were investigated under dry sliding conditions. The influence of both micro- and nanoscale additives was explicitly considered, and the composites were fabricated using a controlled hand lay-up method. The wear performance was analyzed by calculating the mass loss and specific wear rates, and the worn surfaces were examined using scanning electron microscopy (SEM). Wear rates were evaluated under 10 and 15 N loads and 100-200 m sliding distances. Experimental results revealed that the addition of 2-3 wt.% nano SiO2 significantly improved the wear resistance and reduced the mass loss by approximately 55-70% compared to the neat composite. SEM images revealed the presence of abrasive grooves, localized adhesion and material transfer, and micro-scale cracking associated with matrix fragmentation and particle pull-out. Then, predictive models such as Artificial Neural Network, Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbors and XGBoost algorithms were used to predict wear behavior. The wear data obtained were analyzed with the mentioned models by performing hyperparameter optimization with Grid Search method. Model performances are evaluated according to Mean Squared Error (MSE) and Coefficient of Determination (R2) values with 5-fold cross validation. The best results were obtained with an MSE of 83.4 and an R2 of 0.92 for Artificial Neural Network, an MSE of 82.7 and an R2 of 0.92 for Decision Tree, and an MSE of 83.4 and an R2 of 0.92 for K-Nearest Neighbors. The results show that hyperparameter optimization plays a decisive role in model performance and ANN, DT and KNN models provide high accuracy in terms of wear prediction.
Keywords : Aramid kompozit, SiO2, Aşınma davranışı, Makine öğrenmesi, Hiperparametre optimizasyonu

ORIGINAL ARTICLE URL