Abrasive water jet machining for a high-quality green composite: the soft computing strategy for modeling and optimization

Recently, the use of green composites in various applications has increased rapidly due their superior properties. Green composite processing is indeed essential to manufacture good quality parts. Conventional machining of green composites is extremely difficult due to their inherent properties and high cost. Therefore, processing of these composites by abrasive water jet machining (AWJM) is a highly relevant topic nowadays, implying a need to minimize the process time and surface roughness simultaneously. In this study, experiments were performed to investigate the AWJM parameters (pressure within pumping system, traverse speed, stand-off distance), concerning the two outputs. The limitations of regression equations to model multiple outputs simultaneously were addressed by developing neural networks. The networks based on genetic algorithm and on back-propagation algorithm were utilized to map the outputs and AWJM parameters, in forward and reverse mode, respectively. The network performances were discussed in detail. Three metaheuristics were implemented to optimize the AWJM parameters, including consideration of different output weight fractions: hybrid spider monkey optimization, grey wolf optimization and teaching-learning-based optimization. Their performances were compared regarding the solution accuracy and convergence rate. The adopted optimal AWJM setting was successfully verified in a confirmation run, clearly demonstrating its benefits.