Model and algorithm for hot rolling batch planning in steel plants

Abstract

In many iron-steel plants, hot rolling batch planning is usually considered the bottleneck of production operation management. Consequently, the hot rolling batch planning problems (HRBPPs) have been attracting considerable attention among researchers and practitioners, and numerous models and algorithms were developed. However, most of the models are single objective focused. While in practice, production managers consider more than one objective. In this paper, a multi-objective prize collecting vehicle routing problem (PCVRP) model is formulated to solve the HRBPPs, and a decomposition-coordination ant colony system (DCACS) is designed. Firstly, the DCACS divides the candidate slabs into multiple groups which have similar profiles, then it solves each group as a prize collecting traveling salesman problem (PCTSP) and merges the solutions of the PCTSPs to construct a complete solution for the PCVRP. After obtaining the complete solution the DCACS applies a local search procedure to improve it. The above processes are iterated until the stop criterion is met. Taking twenty practical HRBPPs as instances, the model and algorithm are tested for performances. Computational results show that the model and algorithm outperform human-machine coordination method.