Application of Support Vector Regression and Genetic Algorithm to Reduce Web Warping in Flexible Roll-Forming Process

Abstract

In a flexible roll-forming process, a metal blank is incrementally deformed into the desired shape with a variable cross-sectional profile by passing the blank through a series of forming rolls. Because of the combined effects of process and material parameters on the quality of the roll-formed product, the approaches used to optimize the roll-forming process have been largely based on experience and trial-and-error methods. Web warping is one of the major shape defects encountered in flexible roll forming. In this study, an optimization method was developed using support vector regression (SVR) and a genetic algorithm (GA) to reduce web warping in flexible roll forming. An SVR model was developed to predict the web-warping height, and a response surface method was used to investigate the effect of the process parameters. In the development of these predictive models, three process parameters—the forming-roll speed condition, leveling-roll height, and bend angle—were considered as the model inputs, and the web-warping height was used as the response variable. The GA used the web-warping height and the cost of the roll-forming system as the fitness function to optimize the process parameters of the flexible roll-forming process. When the flexible roll-forming process was carried out using the optimized process parameters, the obtained experimental results indicated a reduction in web warping. Hence, the feasibility of the proposed optimization method was confirmed.