Deep learning-based inverse design for engineering systems: multidisciplinary design optimization of automotive brakes

The braking performance of the brake system is a target that must be considered for vehicle development. Apparent piston travel (APT) and drag torque are most representative factors evaluating performance. In particular, as two have conflicting relationship with each other, multidisciplinary design optimization (MDO) approach required design. However, computational cost MDO increases number disciplines increases. Recent studies on inverse use deep learning (DL) established possibility instantly generating an optimal can satisfy without implementing iterative process. This study proposes DL-based (MID) simultaneously satisfies multiple targets, such APT system. Results show proposed find more efficiently compared conventional methods, backpropagation sequential quadratic programming. MID achieved similar to single-disciplinary in terms accuracy cost. A novel was derived basis results, same satisfied existing