Comparing Taguchi-based RSM and ANN for Shredder Blade Geometrical Parameter Optimization

Abstract

In this study, a Taguchi-based RSM in conjunction with an ANN model was utilized to ascertain optimal geometric parameters for the shredding blade employed in a plastic bottle shredder. The shredding process is pivotal in plastic recycling, involving the reduction of waste plastic into smaller fragments to facilitate subsequent transportation and processing. Despite existing research on plastic shredders, further investigations are warranted to optimize shredding blade design. Consequently, a numerical analysis, providing an in-depth insight into understanding the shredder parameters to elucidate the influence of geometric factors was conducted. Subsequent validation was carried out using experimental designs prescribed by the Taguchi-based RSM and ANN models. Both models were then evaluated based on predictive effectiveness and error against simulation data. The predictive outcomes presented that the ANN model resulted in better prediction capacity and lower prediction error than the RSM model, 0.16197 µm and 0.15567 µm, while the numerical validation value was 0.162 µm. Both the original and optimal blades were fabricated and utilized for experiments, illustrating lower wear after measurement using a microscope from ICamScope®. As a result, it is evident from this inquiry that this methodology presents a viable avenue for enhancing the efficiency of plastic recycling machinery and broader industrial applications.