Combined PSO-RBF model and multi-objective alpha evolutionary algorithm for improving efficiency and quality in GaN water-jet-guided laser processing

Water-Jet Guided Laser (WJGL), an emerging technology, offers notable benefits for processing hard and brittle materials. However, the diversity and complexity of processing parameters present significant challenges to simultaneously achieving high efficiency and superior quality. To overcome these limitations, a multi-objective approach that integrating a Particle Swarm Optimization-Radial Basis Function Network (PSO-RBF) model with a Non-dominated Sorting Alpha Evolutionary (NSAE) algorithm was developed. The proposed NSAE algorithm was validated on standard benchmark functions (Kursawe, Viennet and ZDT series) and compared with existing algorithms (NSGA-II, MOGWO, SPEA, MOGOA and NSWOA), exhibited enhanced computational efficiency and accuracy. Single-factor experiments were conducted to elucidate the effects of process parameters on material removal rate (MRR) and heat-affected zone (HAZ). An optimal Latin hypercube sampling design was employed to generate a uniform experimental dataset, enabling the PSO-RBF model to effectively capture the nonlinear relationships between process parameters and optimization objectives. Pareto front solutions were subsequently obtained using the NSAE algorithm, and the optimized processing parameters were analyzed to yield insights into process improvement.