Multi-Objective Differential Evolution Algorithm And Its Application On Rolling Schedule Optimization For Tandem Cold Rolling

Rolling schedule is an important aspect in the steel production process.The various factors need to be considered in making rolling schedule such as rolling energy consumption,balance of power distribution,shape and gauge and so on.Compared with traditional rolling schedule methods,it is able to obtain reasonable rolling schedule by single-objective optimization.Obviously,the limitations of single-objective optimization are unable to consider many objectives at the same time of optimization.Multi-objective optimization can consider many factors and have many unmatched advantages compared with other optimization methods.The research works of this article aim at differential evolution algorithm and its application in rolling schedule optimization,the main contents are as follows:(1)A brief review of rolling schedule optimization and building methods of its optimization model are presented.The basic principle of differential evolution algorithm and its strategy of mutation operation and crossover operation are introduced.(2)At the beginning of adaptive cross-generation differential evolution algorithm,it is unable to utilize information across generations from the same populations.In order to obtain population which is closer to the Pareto Front,opposition learning is added in the beginning of the algorithm.It makes the algorithm use the information across generations efficiently from beginning to end,and convergence speed is ensured.According to actual rolling schedule of thin and narrow slab,multi-objective model of rolling schedule is established.Opposition learning adaptive cross-generation differential evolution algorithm is applied to the optimization model.Results show that values of the three objectives are reduced compared with the actual rolling schedule.(3)For the random mutation operation in differential evolution algorithm,a differential evolution algorithm based on evolutionary direction is proposed.The algorithm calculates the horizontal angle of the difference vector,which is used to choose mutation vector.The chosen vectors contain converging direction and they change the random mutation operation in differential evolution algorithm.Efficiency of the proposed algorithm is verified by two benchmarks.(4)For the mechanical wear in the rolling process,diameters of working roll have uncert-ainty factor which will cause poor robustness of rolling schedule.In order to solve this problem,a robust multi-objective optimization model of rolling schedule which considers the uncertainty of work roll is established.Differential evolution based on evolutionary direction is applied to optimize robust multi-objective model,and the robustness is better than the solutions of rolling schedule optimization with the same objectives.