Adaptive filling modeling of butt joints using genetic algorithm and neural network for laser welding with filler wire. (December 2017)

• Record Type: Journal Article
• Authors: Zhang, Ke; Chen, Yixin; Zheng, Jian; Huang, Jian; Tang, Xinhua
• Is Part Of: Journal of manufacturing processes. Volume 30(2017)
• Page Start: 553
• Page End: 561
• Publication Date: 2017-12
• Subjects: Adaptive filling modeling -- BP neural network -- Genetic algorithm -- Laser welding with filler wire; Production management -- Data processing -- Periodicals; Manufacturing processes -- Periodicals; Procestechnologie; Productietechniek; Production -- Gestion -- Informatique -- Périodiques; Fabrication -- Périodiques; Manufacturing processes; Production management -- Data processing; Periodicals; 670.5
• DOI: 10.1016/j.jmapro.2017.10.025
• Languages: English
• ISSNs: 1526-6125

Abstract

Abstract: Ensuring a uniform joint gap without mismatch over an entire seam remains a challenge. Varying gaps and mismatches significantly affect the welding qualities, especially for sheet metal. In this paper, an adaptive filling model based on a back propagation neural network (BPNN), combined with a genetic algorithm (GA), is proposed. First, optimized welding parameters suitable for the different gaps and mismatches were determined experimentally. Then, the resultant data were used to establish a neural network model that correlated the optimized parameters with the gaps and mismatches; and, by further using the GA, we optimized the initial weights and thresholds of the BPNN. Gap, mismatch, bead width, and reinforcement were acted on as the input neurons, and laser power, wire feed rate, and welding speed were acted on as the output neurons. Findings: indicated that the production of quality welds with desirable uniform weld appearances is ensured despite any gap variances, from 0.1 to 1.0 mm, and mismatches, from 0 to 1.2 mm. The results of the experiment showed that the proposed model is effective in describing the relationship between the joint geometrical parameters, the desired appearance, and the optimized welding parameters, lay the foundation of the subsequently adaptive control.