An Enhanced Surrogate-Assisted Multi-Objective Differential Evolution Algorithm with Self-Adaptive Strategies for Order Planning in Hot Rolling Process: Surrogate-Assisted MODE with Self-Adaptive Strategies for OPHR Problem

This paper studies a crucial decision-making problem of order planning for a hot rolling process (OPHR). The OPHR problem is to determine the optimal sequence of orders, with the goals of minimizing penalties for earliness/tardiness, maximizing profits from special orders, and minimizing transition penalties of adjacent orders. A multi-objective differential evolution (MODE) method with self-adaptive strategies (SaMODE) is developed to tackle it. Then, a radial basis function neural network is integrated into the SaMODE (RBFNN-SaMODE) as a surrogate model to reduce the computational burden of evaluating objectives throughout evolution. The proposed method is extensively tested on real production data, and the results demonstrate its superiority over the classical MODE, yielding highly effective solutions.