International Fiber and Polymer Research Symposium

| Sign in | Sign Up

Home / 15th International Fiber and Polymer Research Symposium

Investigation of Maximum Stress Estimation of Fiber Reinforced Cementitious Matrix Composites by Using Machine Learning Approach

Authors :

Ugur OZVEREN¹ Mete Simsek² Ezgi Lal BUDAK³ Berkay HEPGULSUN⁴ Ozge KAHRAMAN⁵

1- Marmara University, Department of Engineering Faculty 2- Marmara University, Department of Engineering Faculty 3- Marmara University, Department of Engineering Faculty 4- Marmara University, Department of Engineering Faculty 5- Marmara University, Department of Engineering Faculty

Keywords :

Fiber Reinforced Cement Matrix Composites،Random Forest Regression،Machine Learning،Hyperparameter Optimization،Maximum Stress Estimation

Abstract :

In this study, Random Forest regression model was used as a machine learning approach for the estimation of maximum stress values of Fiber Reinforced Cement Matrix (FRCM) composites and Grid Search and Bayesian optimization methods were compared for hyperparameter optimization. In this context, a dataset consisting of 20 different configurations characterized by four different variables obtained from the literature was used for the estimation of maximum stress value. The input variables in the model were determined as FRCM type, textile layer, pre-impregnation status and short fiber addition. Model performance was evaluated using statistical metrics such as R², MSE, RMSE, MAE and MAPE. The results showed that both optimization methods exhibited similar performance on the test data, but the model trained with the Grid Search method had a more balanced generalization ability. In the feature importance ranking, it was determined that the "Textile Layer" and "FRCM Type" variables were the most decisive factors in maximum stress estimations. This study shows that machine learning techniques are an effective tool in understanding the mechanical behavior of FRCM composites and optimizing design parameters.