ADVANCED ENSEMBLE MACHINE LEARNING PREDICTION TO ENHANCE THE ACCURACY OF ABRASIVE WATERJET MACHINING FOR BIOCOMPOSITES

Abstract

Listen

Translate article

The optimization of Abrasive Waterjet Machining (AWJM) for natural fiber composites remains challenging due to the heterogeneous nature of these materials, which often results in inconsistent machining outcomes such as surface roughness and kerf angle. This study addresses these issues by focusing on boiled eggshell-filled Roselle fiber polyester composites, aiming to enhance machining precision and efficiency. To achieve this, an ensemble machine learning (EML) model, incorporating k-nearest Neighbors, Random Forest, Long Short-Term Memory (LSTM) networks, and Artificial Neural Networks (ANN), was developed and combined with Response Surface Methodology (RSM). The AWJM input parameters explored include standoff distance (1–3 mm), traverse speed (50–150 mm/min), and abrasive flow rate (100–300 g/min). The optimized settings an abrasive flow rate of 300 g/min, traverse speed of 150 mm/min, and standoff distance of 1.95 mm yielded minimized SR of 3.34 microns, maximized MRR of 891.13 mm³/min, and minimized KF of 2.57 degrees. These results underscore the significant impact of standoff distance and traverse speed on KF and MRR, respectively. This research provides a data-driven approach to optimizing AWJM parameters for high-precision applications, advancing sustainable manufacturing practices for composite materials.