节点文献
罩式退火炉优化调度方法及其应用
Optimal Scheduling Method and Its Application for Bell-type Batch Annealing Process
【作者】 刘全利;
【导师】 王伟;
【作者基本信息】 大连理工大学 , 控制理论与控制工程, 2005, 博士
【摘要】 罩式退火炉退火工序是冶金工业冷轧薄板生产中的重要工序之一。罩式退火炉车间的生产调度过程是一个典型的多工序、有重入的多机并行调度问题,且伴随着复杂的资源约束、能源约束和工艺约束,难于解析建模和实现优化调度。另外,退火过程中钢卷冷却时间事先未知,如何预报钢卷冷却时间对优化调度和节省能源则显得尤为重要。针对这些生产实际问题,在查阅大量文献资料和对现场工艺进行充分调研的基础上,本文完成了以下主要工作。 提出了基于仿真优化框架的罩式退火炉生产优化调度的新思路。与相关文献中提出的混合整数线性规划(MILP)和贪婪排序方法相比,该思路直接面向罩式退火炉生产的自然结构,能够实现罩式退火炉生产的动态调度。 将罩式退火炉冷却时间预报的多入单出建模问题化为一组单入单出建模问题,采用模糊C均值聚类方法对退火生产数据进行处理,并结合最小二乘回归提出罩式退火炉冷却时间预报方法。该方法在不增加硬件投资的情况下,实现了罩式退火炉冷却时间预报,为罩式退火炉工艺的全流程模拟和优化调度打下了基础。 利用离散事件仿真技术和面向对象建模技术建立了罩式退火炉离散事件仿真模型。建立的模型具备如下特点:模型的输入是用户给定的装炉顺序,这一顺序易于用自然数编码,便于用遗传算法中的染色体表达。模型考虑了仿真过程中的各种约束,因而对于用户给定的任一装炉顺序都能给出一个可行调度。在此基础上,将离散事件仿真模型和改进的遗传算法相结合,提出了罩式退火炉生产优化调度方法。该方法中,离散事件仿真作为隐式模型出现,遗传算法根据离散事件仿真的结果对调度解可行集进行搜索,最后求得优化的调度方案。 采用面向对象程序设计方法,在宝钢益昌薄板有限公司实现了整个罩式退火炉车间生产的动态优化调度系统,目前系统已在该厂稳定运行。 本文选题来自生产中存在的实际问题,提出的罩式退火炉优化调度方法已成功应用于生产现场,提高了生产设备的利用率,为企业带来了很好的经济效益。
【Abstract】 Bell-type batch annealing is one of the important procedures in cold-roll thin sheet production in iron and steel industry. The production scheduling of bell-type batch annealing furnaces is a typical multi-working procedure, re-entry, multi-machine parallel scheduling problem with complicated resource constraints, energy constraints and working procedure constraints. It is very difficult to build the mathematic model of the optimal scheduling of bell-type batch annealing production. The process time of cooling procedure in bell-type batch annealing is unknown when the scheduling begins. It is very important to predict the cooling time of the annealing furnaces, which is the first step to realize the optimal scheduling of the annealing production. Realizing the dynamic scheduling of bell-type batch annealing is also urgent in the cold-roll thin sheet production factory. Aiming at solving these problems and based on review of the related research reference, this dissertation has carried on the following research.A new production scheduling method of bell-type batch annealing based on the simulation-optimization frame is proposed in this dissertation. It is better than the MILP method and Greedily Sorting Method proposed in the correlative references. It accords with the natural structure of bell-type batch annealing production, and can realize the dynamic scheduling of bell-type batch annealing production.The cooling time prediction of the annealing furnaces is a multi-input and single-output process, which is simplified to a group of single-input and single-output prediction problems. The fuzzy C-Means algorithm is used to cluster the production data in batch annealing process, then the exponent least square algorithm is used to form a cooling time prediction method for annealing furnaces. It realizes the cooling time prediction of batch annealing production without new facility investment needed, which paves the way for optimal scheduling and the whole flow simulation of annealing process.A discrete event simulation model of bell-type batch annealing shop is proposed in this dissertation, in which the discrete event simulation technique and object-oriented modeling technique are used. The input of this model is the charging order of the annealing plans being annealed, which can be coded in natural number and expressed as the gene in genetic algorithm conveniently. This model considers all kinds of constraints in the optimal scheduling of bell-type batch annealing production and can give a feasible scheduling corresponding to each charging order the dispatcher gives. An optimal scheduling method of bell-type batch annealing production is proposed by combining an improved genetic algorithm and discrete event simulation technique. The discrete event simulation model of
【Key words】 Iron and steel industry; bell-type batch annealing; optimal scheduling; genetic algorithm; discrete event simulation; fuzzy C-means clustering;