Modeling and multi-objective optimization of abrasive water jet machining process of composite laminates using a hybrid approach based on neural networks and metaheuristic algorithm

Although the abrasive water jet (AWJ) has proven to be a suitable process for machining composite materials, it has some limitations related to dimensional inaccuracy and surface defects. As the performance of the AWJ process mainly depends on the machining parameters, an optimal selection of them is crucial to achieving an improved quality of cut. In this context, the present study reports an experimental investigation to assess the influence of AWJ machining parameters on kerf taper angle (θ) and surface roughness ( R a ) of E glass/Vinylester 411 resin laminates. The experiments are carried out using a full factorial design by varying the water pressure, traverse speed, abrasive flow rate, and standoff distance. A first-ever attempt is made in this paper to optimize the AWJ process using a hybrid approach combining artificial neural networks (ANNs) with a recently proposed metaheuristic algorithm known as multi-objective bonobo optimizer (MOBO). The results show that standoff distance and abrasive flow rate were the most significant control factors in influencing θ and R a , respectively. The developed ANN models are capable to predict the output responses with high accuracy and the solutions from the Pareto front provide a sufficient performance with a trade-off between θ and R a . The corresponding levels of the optimal process parameters are 430 g/min for the abrasive flow rate, the range of 140–180 mm/min for the traverse speed, 280 MPa for the pressure, and 1.5 mm for the standoff distance.