Artificial Intelligence and Machine Learning–Driven Modeling for Predicting Mechanical, Thermal and Electrical Properties of Advanced Materials for Next-Generation Engineering Applications

Abstract

Accurate prediction of material properties is essential for accelerating materials design and reducing dependence on computationally intensive simulations and experimental trials. This study proposes an Artificial Intelligence (AI) and Machine Learning (ML)–based framework for the simultaneous prediction of mechanical, thermal, and electrical properties of advanced materials. The proposed approach integrates multi-source datasets with systematic preprocessing and feature engineering to develop a unified multi-output regression model.

Multiple machine learning algorithms, including Random Forest, Support Vector Machine (SVM), Artificial Neural Networks (ANN), and Extreme Gradient Boosting (XGBoost), are trained and evaluated using standard performance metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and coefficient of determination (R²). Experimental results show that XGBoost achieves the best performance, with MAE = 0.093, RMSE = 0.141, and R² = 0.942, outperforming other models across all property domains. The strong agreement between predicted and actual values, along with near-normal error distributions, confirms the robustness and generalization capability of the proposed framework.

Feature importance analysis identifies compositional descriptors as the dominant contributors, while structural and electronic features provide complementary influence. Correlation analysis further reveals strong interdependencies among predicted properties (correlation coefficients ≈ 0.93–0.95), validating the effectiveness of the multi-output learning approach. The proposed framework provides a scalable and efficient solution for multi-property materials prediction and offers significant potential for data-driven materials discovery and optimization in advanced engineering applications.