Data-Driven Modeling for Predicting the Machining Performance of Ramie-Epoxy Based Hybrid Composites Using Machine Learning Algorithms

Abstract

Polymer composites with both fiber and filler reinforcements are finding extensive uses in various application areas where their machining ability such as drilling or milling performance plays a decisive role. In this context, this investigation reports on the application of machine learning (ML) techniques and statistical methods to analyze and predict the drilling performance of ramie fiber reinforced epoxy composites filled with sponge iron (SI) slag particles. These hybrid composites are fabricated using conventional hand lay-up technique and are then subjected to drilling trials using the design of experiments. The 81 different parameter combinations as per the design are used to study the effects of filler content, spindle speed, feed rate and drill bit diameter on the thrust force and delamination factor of these composites using Analysis of variance (ANOVA). The results reveal that the filler content followed by feed rate, drill diameter and spindle speed have significant effect on thrust force and delamination factor. The data generated from experimentation are further processed to predict the thrust force and delamination factor of the composites with an ML approach following decision tree regression (DTR), random forest regression (RFR), gradient boosting regression (GBR) and extreme gradient boosting regression (XGBR) algorithms and to analyze the absurdity among experimental results and predicted responses. It is found that the GBR model outperforms other models in predicting the thrust force whereas the XGBR model shows the best prediction of delamination factor.