Effective optimization strategies for abrasive water jet machining of glass-carbon fiber reinforced composites: A comparative study of evolutionary optimization techniques

Abstract

This study applied evolutionary optimization techniques and neural networks to predict optimum machining parameters of Abrasive Water Jet Machining (AWJM) for machining Glass-Carbon Fiber Reinforced Composite (GCFRC) materials. Several researchers have employed different optimization techniques; however, evolving computational capabilities further open avenues to optimize such parameters. Five evolutionary techniques, namely Artificial Neural Network (ANN), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Simulated Annealing (SA), and Nonlinear Least Square Error (LSE), were applied to minimize surface roughness ($\mathrm{Ra}$) and maximize kerf width ($\mathrm{Kw}$) and material removal rate ($\mathrm{MRR}$). The output performance parameters ($\mathrm{Ra}$, $\mathrm{Kw}$, and $\mathrm{MRR}$) were formulated as a linear mathematical function of machining parameters: tool feed rate ($\mathrm{TFR}$), cutting speed rate ($\mathrm{CSR}$), and stand-off distance ($\mathrm{SOD}$). Though this study solves linear optimization problems, the proposed technique will be a strong tool for solving complex associations of machining and performance parameters in the future. A dataset of machining parameters and subsequent performance parameters was adopted from the available literature. The results indicated that the LSE method outperformed other techniques, yielding the lowest Root Mean Square Error (RMSE) in predicting $\mathrm{Ra}$, $\mathrm{Kw}$, and $\mathrm{MRR}$, thus ensuring high machining accuracy. LSE technique reported relatively least RMSE values of 0.37 µm, 0.149Mm, and 237.23 mm³/min for $\mathrm{Ra}$, $\mathrm{Kw}$, and $\mathrm{MRR}$, respectively. SA and PSO displayed identical and competitive RMSE values, slightly higher than LSE (up to 20 % higher). ANN and GA techniques were not effective relative to other considered techniques. LSE, SA, and PSO provide superior performance in optimizing AWJM parameters. The significant contribution of this research is the proposed optimization technique, offering a clear direction for solving complex associations between the performances and machining parameters of AWJM. This work also provides a foundation for future research to optimize such associations for other machining setups.