Optimization of Specific Energy, Scrap, and Surface Roughness in 3D Printing Using Integrated ANN-GA Approach

Abstract

3D printing technology is fast emerging as a solution to convert cyber models to physical models quickly for visualization and feedback in Industry 4.0 environment. Energy efficiency, surface roughness, and material wastage are important performance responses and the effects of 3D printing parameters on these conflicting responses need to be studies to further improve the technology. Multiobjective optimization is a tool to obtain the right balance among conflicting performance responses. This paper aims to find the optimal values of infill, layer height, printing speed, extruder temperature, and scale to optimize specific energy, scrap, and surface roughness, simultaneously. Experiments were performed based on a Taguchi L-27 orthogonal array using PLA filament. A predictive model has been developed using artificial neural network (ANN) integrated with a genetic algorithm (GA) for obtaining Pareto solutions. Technique for order preference by similarity to ideal solution (TOPSIS) is used to obtain the most preferred solution from the Pareto solutions and analytical hierarchal process (AHP) is used to determine weights of the three objectives. The proposed methodology is expected to help the practitioners to rank and customise the decisions proactively in conflicting scenarios before the product is 3D printed, thereby improving sustainability and/or meeting product quality requirements