【作者】 王浩；

【导师】 刘士新；

【作者基本信息】 东北大学 ， 系统工程， 2018， 硕士

【摘要】 随着德国“工业4.0”的提出和“中国制造2025”战略的推进,下游企业对钢铁原材料的需求更加个性化,钢铁企业大批量生产方式与客户的多品种、小批量的个性需求之间的矛盾更加突出,给钢铁企业的生产设计和生产组织带来了很大的难题。如何解决上述矛盾成为钢铁企业面临的关键难题之一。炼钢-连铸工序位于钢铁企业生产过程的上游,是钢铁产品生产过程的关键环节。在该工序上大批量生产方式与用户个性化需求的矛盾主要体现在钢种设计上。本文基于钢种设计优化问题的实际特点建立了钢种设计两阶段鲁棒优化模型。为提高模型的求解效率,首先根据模型特点,采用了行列生成的求解框架;然后利用对偶模型将子问题简化,针对存在非线性的情况,采用了 KKT条件、强对偶等算法解决。实验结果表明这些算法可以求解中小规模问题,但是难以获得大规模问题的精确解。为此,进一步根据钢种设计中存在的钢种组的现象,利用拉格朗日松弛算法将问题分解成在钢种组上的模型,基于次梯度法的思想设计了新的迭代步长、方向和终止条件。最后总结出了基于拉格朗日松弛分解策略的行列生成算法,并根据企业的实际生产数据设计实验,实验结果表明该算法可以获得大规模问题的精确解。本文针对钢铁企业亟待解决的钢种优化设计问题,建立了与实际生产过程契合度很高的优化模型,设计了一系列算法获得了模型的精确解。无论对于指导企业生产或是理论方法的研究都有一定的意义。更多还原

【Abstract】 With the proposal of "Industy 4.0" in Germany and the "Made in China 2025 strategy"strategy,the demand of the downstream enterprises is more individualized for steel raw materials.The contradiction between mass production mode and individual demand of multiple varieties and small batch in iron and steel enterprises is more prominent,which brings great difficulties to the production design and production organization of iron and steel enterprises.How to solve the above contradiction has become one of the key problems faced by the iron and steel enterprises.Steelmaking and continuous casting process is located in the upstream of the iron and steel enterprise production process,and it is the key link in the production process of iron and steel products.In this process,the contradiction between the mass production mode and the personalized demand of the customer is mainly reflected in the steel grade design.Based on the actual characteristics of steel grade optimization and design problem,we established the two stage robust optimization model of steel grade design.In order to improve the efficiency of the model,firstly,according to the characteristics of the model,the column-and-constraint generation algorithm is used to solve the problem.Then,the sub-problem is simplified by using dual model.And the KKT condition and strong duality algorithm are used to solve the nonlinear situation.Experiments show that these algorithms can be used to solve small and medium scale problems,but it is difficult to obtain the exact solution of large-scale problems.For this reason,we decompose the problem into a diagonal model on the steel grade group by using the Lagrangian relaxation algorithm because of the phenomenon of steel group in the steel grade design.We design a new iterative step size,direction and termination condition according to the idea of sub-gradient method.Finally,we summarized column-and-constraint generation algorithm based on Lagrangian relaxation decomposition strategy,and the experiment is designed according to the actual production data of the enterprise.The experiment shows that the algorithm can find the exact solution of large-scale problem.In order to solve the urgent steel grade design and optimization problem in iron and steel enterprises,we established an optimization model with high degree of agreement with the actual production process.And we design a series of algorithms to obtain the exact solution of the model.It has certain significance for the guidance of enterprise production or the research of theory and method.更多还原