Data-driven method for process optimization in electromagnetic-electrohydraulic hybrid high-velocity sheet metal forming

High-velocity forming (HVF) is a flexible process that can essentially enhance the forming limit of lots of low-ductility metals; however, its complicated mechanism for the deformation control hinders its wide applications. To this end, this paper introduces a data-driven methodology to uncover the nonlinear regulation of the deformation control and figure out the optimum process setting toward the preferred deformation behavior. More specifically, we adopt a newly electromagnetic-electrohydraulic hybrid forming process as the research object. Artificial neural networks are adopted to learn the process mapping between three discharge parameters (electromagnetic voltage, electrohydraulic voltage, and discharge timing) and two forming indexes (forming height and thinning rate). And then, the genetic-algorithm-based multi-objective method is used to maximize forming height, meanwhile, minimize thinning rate, and obtain the optimum process parameters. Based on such optimization, two deformation control rules are identified: (1) the greater electromagnetic voltage is preferred for better deformation uniformity; (2) the optimum discharging timing and electrohydraulic voltage are negatively correlated with each other. Based on the optimum parameter setting, the experimentations are conducted to validate such results. A 50% improvement rate on the limiting forming height is observed compared with the conventional electrohydraulic process, which is about 150% higher than the improvement rate (20.9%) obtained in our previous work, showing a substantial process improvement. In a word, the proposed method provides an effective approach for better understanding and exploiting the complicated process control mechanisms in the high-velocity forming process and its variants.