Enhancing dimensional accuracy of small parts through modelling and parametric optimization of the FDM 3D printing process using GA-ANN

Nowadays, almost every manufacturing industry primarily focuses on precision manufacturing, which can now be easily possible due to advanced rapid prototyping techniques to achieve their overarching goals. Small parts fabrication necessitates ever-more-careful workmanship and strict environmental controls as all materials inevitably expand and contract due to environmental changes, which can raise costs and lengthen the production process. One rapid prototyping method is frequently used to print small objects is fused deposition modelling. In the proposed work, the significant FDM printing parameters (no. of contours, infill density, and layer thickness) are optimized for improving the dimensional precision of FDM printed small specimens of 1mm x 2mm x 3mm. Twenty experimental runs were designed by employing a face-centred central composite design (FCCD) methodology to analyse the effect of input variables on the fabricated specimen. For training and optimization, hybrid statistical tools and artificial neural networks (ANN) integrated with genetic algorithm (ANN-GA) are utilized to obtain the optimized combination of input parameters. Validation tests were performed, sequentially to confirm the various created models for the selection of best process parameter. It has been observed that the minimum percentage change accomplished with GA-ANN approach in height, length and breadth is 1.9455 %, at input variables (Infill density: 55.85 %, Layer thickness: 0.1mm, no of contours: 8), 0.29542% at input variables (Infill density: 23.31, Layer thickness: 0.18, no of contours: 7), 0.4648 % at input variables (Infill density: 48.541 %, Layer thickness: 0.1 mm, no of contours: 3), are best forecasted results obtained using GA-ANN approach and the same has been validated experimentally.