A new optimization algorithm of drawbead on panel die was proposed based on an iterative learning control model, which greatly improved the optimization efficiency. The iterative learning control model was constructed by the forming state function, the forming quality function and the learning updating law function. It was applied to the optimization of drawbead restraining force, and the state function, which was hardly calculated explicitly, was replaced by the numerical simulation to predict the stress-strain state under the given technological parameters. Furthermore, the defect degree of drawbead was defined as the forming quality function to evaluate the forming quality around drawbead. The learning updating law function not only could show the change value of drawbead restraining force based on the forming quality around drawbead, but also could update the learning gain automatically to modify the change extent of drawbead restraining force, which accelerated the convergence speed of the optimization. At last, the rapidity and practicality of the algorithm were verified by the numerical experiment of the inner door panel.