Predicting the mechanical behavior of carbon fiber-reinforced polymer using machine learning methods: a systematic review

Abstract

Considering the complexity of the mechanic analysis in advanced composite materials, studies in the literature have demonstrated the use of machine learning (ML) methods, aiming to predict the mechanical properties in high-reliability levels. ML models have been also used in medical applications, biological sciences, and data control systems, presenting prospects in analyzing and modeling mechanical/thermal behavior for engineering applications. For this purpose, this chapter aims to conduct a systematic review of ML methods on the mechanical properties of structural composites. The analysis of the ML approach parameters and efficiency are also highlighted. A systematic review was performed using the PRISMA methodology to identify the main discoveries in recent studies. A total of 490 studies were initially identified from 2013 to 2022. Then, each article was selected and described by specific inclusion/exclusion criteria. The main findings were presented and discussed, and the gaps are identified to open up further investigations yet to be understood and exploited.