MACHINE LEARNING MODEL FOR PREDICTING MECHANICAL PROPERTIES OF NATURAL FIBER COMPOSITE MATERIALS

Abstract

The increasing awareness for environmental sustainability has led to emerging eco-friendly materials. Composite materials have various applications in different sectors such as automobiles, healthcare, and daily use products. The usability of natural fiber reinforced composites are essential due to its various advantages over synthetic fiber composites. Machine learning (ML) has evolved in recent times. The use of modern technologies in composites provides the necessary scope and depth in research. In this dissertation, different articles have been evaluated to screen out the data for training the ML model for predicting the mechanical properties of natural fier reinforced composites. This study deployed linear regression as ML algorithm and programming was performed on Pycharm community and Jupyter notebook. Python programming language and python libraries such as sklearn, tkinter, pandas, numpy, matplotlib has been used. The dissertation established relation among mechanical properties. The user selects the fiber and polymer and defines the volume fraction, density, area density, and fiber orientation as per the product requirement. Based on the inputs, the graphic user interface (GUI) displays the mechanical properties (ultimate tensile strength, young’s modulus, compression modulus, compressive strength, poisson ratio, strain at failure (in %)) predicted by the ML model. The interactive interface also displays the accuracy of the ML model.