Multiresponse optimization and network-based prediction modelling for the WEDM of AM60B biomedical material

Abstract

The AM60B magnesium alloy remains a pivotal player in the pursuit of lighter, biodegradable and more sustainable solutions for various biomedical implants because of its excellent machinability, exceptional strength and resistance to degradation. However, AM60B magnesium alloys exhibit poor machinability when machined by conventional methods, since they are susceptible to deformation and degradation at elevated temperatures. The current work focused on multiresponse optimization for AM60B magnesium alloys for wire electrical discharge machining (WEDM), where multiple output variables including the machining rate, surface irregularity, and microindentation hardness, were considered simultaneously. Response surface methodology (RSM) along with artificial neural network (ANN) were utilized to investigate the effect of these parameters on the control of these output characteristics. The findings demonstrated the best combination of input specifications, with a discharge duration of 112.743 µs, a spark gap time of 57.6532 µs, a discharge voltage of 6.63 V and a wire advance rate of 5.39357 mm/min which yielded the best surface irregularity, microindentation hardness, and machining rate from the RSM. ANN and RSM models were effective in simulating the experiments, with predicted values closely aligning with the experimental results. An optimally trained network model exhibited good agreement with a mean error less than 5%. Additionally, the condition of the machined surface, including any cracks, voids, or other flaws, was examined using scanning electron microscope (SEM).