Le, V.T. and Nguyen, T.-T. and Nguyen, V.C. (2024) Wire and arc-based additive manufacturing of 316L SS: predicting and optimizing process variables using BRFFNN, NSGA-GP and TOPSIS approach. Neural Computing and Applications. ISSN 09410643
Full text not available from this repository. (Upload)Abstract
Wire- and arc-based additive manufacturing (WA-AM) technology is a prominent solution to produce components having large-scale dimensions in terms of elevated deposition rates, high material utilization efficiency, and cost effectiveness. Although many research works have explored the WA-AM process of 316L stainless steel, the selection of optimal input parameters for enhancing geometrical characteristics of the WA-AMed 316L stainless steel has not been addressed. In this paper, the variables—voltage (U), current (I), and traveling speed (V), were optimized to obtain the expected attributes of weld beads (WB) in the WA-AM of 316L SS. The empirical models of the width (BW), height (BH), and contact angle (CA) of WBs were developed using a BRFFNN (Bayesian regularized feed forward neural networks) model. To determine the best optimality, the non-dominated-sorting-genetic algorithm based on a grid partition (NSGA-GP) and TOPSIS (technique for order of preference by similarity-to-ideal solution) were adopted. The outcomes indicate that the developed BRFFNN models are adequate to predict the objectives (BW, BH, and CA). The optimized value of I, U, and V is 130 A, 22.0 V, and 0.30 m/min, respectively, which enable BW, BH, and CA to be improved by 22.97, 11.24, and 5.61, respectively. The optimal parameters were used to successfully build a component without major defects, indicating their suitability for producing 316L SS components used in industrial applications. The outcomes have demonstrated the efficiency of the proposed optimization approach, which can also be used to predict optimal parameters of other AM and conventional manufacturing processes. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.
Item Type: | Article |
---|---|
Divisions: | Offices > Office of International Cooperation |
Identification Number: | 10.1007/s00521-024-10375-z |
Uncontrolled Keywords: | Metal castings; Smart manufacturing; Wire, 316 L stainless steel; 316L SS; Bayesian; Bayesian regularized feed forward neural network; Feed forward neural net works; Ideal solutions; Neural network model; NSGA-GP; Optimisations; Wire- and arc-based additive manufacturing, Deposition rates |
URI: | http://eprints.lqdtu.edu.vn/id/eprint/11355 |