Surrogate Model-Based Parameter Tuning of Simulated Annealing Algorithm for the Shape Optimization of Automotive Rubber Bumpers

A design engineer has to deal with increasingly complex tasks on a daily basis, for which the available time is shrinking. Market competitiveness can be improved by using optimization if process automated. If there limited information about behavior of objective function, global search methods such as simulated annealing (SA) should used. This algorithm requires selection number parameters based task. procedure reducing spent tuning SA computationally expensive, simulation-driven was developed. The applicability method demonstrated solving shape problem rubber bumper built into air spring structures lorries. Due time-consuming function call, support vector regression (SVR) surrogate model used test performance algorithm. To perform SVR training, samples were taken maximin Latin hypercube design. an adaptive space and different cooling schedules implemented. Subsequently, fine-tuned trained model. An optimal found adapted negligible error from technical aspect.