Prediction of the influence of loading rate and sugarcane leaves concentration on fracture toughness of sugarcane leaves and epoxy composite using artificial intelligence

Abstract

Nowadays, agricultural waste is one of the most common types of waste in the world. In Thailand, one of the most popular types of agriculture is sugarcane farming. Sugarcane farming produces waste from harvesting or processing such as bagasse, leaves, etc. Many researchers have tried to make waste from the agricultural sector become useful, whether using it as a renewable fuel or mixed into new material. To use mixed agricultural waste to make new composite materials for advanced engineering, one factor to consider is the fracture toughness of the composite. The fracture toughness of a material can be calculated in many ways, whether by testing a real material, finite element analysis, or prediction with predictive equations. This research uses the artificial intelligence methods that have become popular over the years to create a model to predict the fracture toughness of sugarcane leaves composites, one of the wastes generated from sugarcane farming. The model was used to predict the fracture toughness of sugarcane leaves and epoxy composite influenced by leaf concentration (%wt.) and loading rate (mm/min). The modeling uses three different artificial intelligence models i.e., Artificial neural network, Generalized regression neural network and Gaussian process regression using data from a limited number of 27 data. The prediction result in the testing period showed the ANN model had an R2 of 0.8818, a MAPE of 3.40%, and an RMSE of 0.0876. The GRNN model had an R2 of 0.9192, a MAPE of 2.81%, and an RMSE is 0.0738. The GPR model had an R2 of 0.9085, a MAPE of 3.41 and an RMSE of 0.0773. As for the confirmation of the prediction model, it was found that the performance of the three models declined as the level of the predictive factors changed, but the performance remained within the acceptable range.