Prediction of natural fiber composite mechanical properties with a multilayer artificial neural network

For environmental and cost reasons that have gained acceptance across the globe, natural fiber–reinforced composites have been adopted in all sorts of applications currently. However, these composites are a production challenges, such as poor adhesion between reinforcement and matrix. Consequently, a comprehensive Artificial Neural Network (ANN) framework to predict the mechanical properties of green composites, especially for Ultimate Tensile Strength (UTS) and Young's Modulus, is adopted, to reduce the experimental work issues. Since the mechanical properties of green bio-fibers for several key parameters of influence could well differ from one fiber to the next, bio-fiber prediction involves a high level of complexity. Therefore, this paper utilizes an Artificial Neural Network (ANN) framework with non-linear activation functions to predict the mechanical properties of natural fibers—specifically Ultimate Tensile Strength (UTS) and Young's Modulus—based on cellulose content, microfibrillar angle, and fiber diameter. The developed ANN system consists of three model techniques for predicting the mechanical properties of natural fiber composites; however, the 3-6-12-18-2 model has given results quite predictably. Results show that the ANN model could help raise the scientific evaluation of green fibers and determine their suitability as a replacement material for effective green product output in future eco-friendly industries.