Rolling Plan Algorithm For Hot Rolling Processes

Scheduling of hot rolling is the key task of production management of steel rolling. The scheduling of coils in the strip mill has a direct influence on roller wear and production efficiency. Parallel algorithm and serial algorithm are two kinds of strategies to solve the scheduling problems. Parallel algorithm is to make M unit schedules for rolling according to the N orders simultaneously, which is commonly formulated by Vehicle Routing Problem (VRP) model or improved VRP model. The drawbacks of parallel algorithm are the computational complexity and the reduced availability of global optimization under dynamic production environment. Serial algorithm is to make M unit schedules for rolling according to the N orders sequencely, which is usually described by Traveling Salesman Problem (TSP) model or improved TSP model. Compared to parallel algorithm, the advantages of serial algorithm are simpler and stronger adaptability to the dynamic production environment.The scheduling problem of the hot strip mill in this thesis had been formulated using TSP model. The model meet constraint that the total length of coils assigned to a turn was, at most, the maximum length permitted by roll wear. The objective function was the penalty of delayed delivery and the penalty of gauges and widths jump. The greedy algorithm was used to generate initial solution, and Tabu Search (TS) algorithm was used to optimize the objective function. Scheduling program of hot rolling was developed using MATLAB. The program contained six main modules, and the solving process of initial solution and TS algorithm had been completed using this program.Based on the real-world data of strip mill of BaoTou Iron and Steel enterprise, the schedule programmed by TS algorithm in this thesis is not only that width changes smoothly, but also that gauges jump times is less than the schedule programmed manually. Compared to the schedule programmed manually, the total length of coils assigned to a turn increases averagely of 12 percent using the method in this thesis, and the times of roll changing are reduced, so the production efficiency is increased.