Artificial Intelligence-Assisted Prediction and Multi-Objective Optimization of Mechanical Properties and Failure Performance of Fiber-Reinforced Composite Materials

Abstract

Fiber-reinforced composite materials are increasingly used in aerospace, automotive, marine, renewable-energy, transportation, and structural engineering applications because of their high specific strength, stiffness, lightweight characteristics, and design flexibility. However, the mechanical response and failure behavior of these materials are governed by complex interactions among fiber type, fiber volume fraction, matrix properties, fiber orientation, laminate architecture, manufacturing conditions, and loading characteristics. Conventional experimental approaches for identifying optimum composite configurations are often time-consuming, resource-intensive, and unable to efficiently capture highly nonlinear material–structure relationships. Artificial intelligence (AI) and machine learning (ML) provide an alternative data-driven approach for predicting mechanical properties, identifying failure mechanisms, and optimizing composite configurations.

This study proposes an AI-assisted framework for the prediction and multi-objective optimization of mechanical properties and failure performance of fiber-reinforced composite materials. The proposed framework integrates experimental or validated literature-derived composite datasets with multiple ML algorithms, including multiple linear regression, decision tree, random forest, support vector regression, gradient boosting, XGBoost, and artificial neural networks. Input parameters include fiber type, fiber volume fraction, fiber orientation, laminate configuration, matrix characteristics, and selected manufacturing and loading parameters, while tensile strength, flexural strength, impact performance, and failure mode are considered as primary response variables. Data preprocessing, feature engineering, cross-validation, hyperparameter optimization, and independent testing are incorporated to improve model reliability and minimize overfitting.

In addition to regression-based prediction, classification models are developed to identify dominant failure mechanisms such as fiber breakage, matrix cracking, interfacial debonding, fiber pull-out, and delamination. Explainable artificial intelligence techniques are incorporated to quantify the contribution of individual material and structural parameters to predicted performance. The resulting predictive models are subsequently integrated with a multi-objective optimization procedure to identify composite configurations that simultaneously improve mechanical strength and resistance to failure while considering practical design constraints.

The proposed framework establishes a systematic connection between composite material design, mechanical-property prediction, failure classification, explainable AI, and optimization. The study is intended to reduce dependence on extensive trial-and-error experimentation and provide a rapid computational approach for materials selection and laminate design. The framework can also support future development of physics-infor med and hybrid AI models for reliable composite design. The research therefore demonstrates the potential of AI-assisted methodologies to accelerate the development of high-performance fiber-reinforced composite materials while improving the interpretability and engineering applicability of data-driven material optimization.