紧凑热带生产系统的智能优化组批方法研究

In recent years, a number of advanced compact strip production (CSP) lines have been set up in China. However, it is hard to produce the effective batch planning methods by the production experience based heuristics which are commonly used in the iron and steel companies. This leads to low productivity, high energy consumption, and prone to accidents of shutdown of the CSP. To address the above problems, this project is presented to solve typical batch planning problems existed in the CSP processes in the iron and steel companies in our country. We consider three representative production patterns which are the single caster and single steel grade, double casters and single steel grade, and double casters and double steel grades, and focus on the hot roiling batch planning problems, charge and cast batch planning problems, integrative batch planning problems based on decomposition and iteration strategy, and batch re-planning problems with disturbance classification mechanism. We will systematively study the complexity, feasible solution, neighborhood structures of the problems, and propose effective discrete swarm intellignece optimization methods for solving the problems. We try to provide a serials of research achivements on the the design procedures of batch planning algorithms, the batch planning strategies for dealing with dynamic production process, the handing methods for the multiple objectives and constraints, the convergence and complexity analysis of the prsented algorithms. The achievements can enrich and deepen the existed optimization and batch planning theory, and it is also expected to be applied to the CSP enterprises in China, and to enhance their management standards and their competitiveness as well.

近年来，我国多家钢铁企业组建了先进的CSP（Compact Strip Production，紧凑式热带生产）系统，但现有以人工经验为主的组批方法难以获得高质量的批量计划，造成CSP系统的效率低、能耗高、容易发生停产事故。本项目拟从我国CSP企业的生产需求出发，针对单铸机同钢种、双铸机同钢种和双铸机异钢种三种典型生产模式，系统研究轧制单元组批、炉次-浇次组批、基于分解迭代策略的一体化组批以及基于扰动分类机制的重组批等问题，探索问题的复杂性、可行解特性和邻域特性，提出基于离散群智能的高性能组批方法。旨在CSP组批问题的优化算法设计、约束和多目标处理、动态重组批策略、算法收敛性和复杂性分析等方面给出一批具有工程应用价值的理论成果。研究成果将丰富和深化已有的批量计划理论，可望直接服务于我国CSP企业，促进其生产管理水平和市场竞争力的提高。

紧凑式热带生产工艺是一种先进的薄板钢连铸连轧生产技术，在我国钢铁企业中有着广泛的应用。轧制单元组批问题和炼钢连铸调度问题是紧凑式热带生产过程中的两个关键优化问题。合理的轧制单元组批与炼钢连铸调度对保持生产过程高效稳定的运行具有重要意义。本项目深入研究了轧制单元组批和炼钢连铸调度的优化理论与方法，主要内容包括：①在轧制单元组批方面，按照生产工艺要求，兼顾炼钢和连铸生产阶段，考虑相邻板坯的厚度和表面质量等约束条件，针对典型的轧制单元组批问题，建立了数学模型；探讨了问题的复杂性、可分解性、邻域结构及其生成方式；研究了果蝇优化算法、变邻域搜索算法和多目标进化算法等的优化机理、搜索模式、算法框架、操作算子和参数设定；提出了解决典型轧制单元组批问题的构造式组批算法、变邻域搜索组批算法与多目标智能组批算法。②在炼钢连铸调度方面，考虑轧制单元对炉次浇注次序的要求、炉次的连浇要求、加工时间的可控性等，针对静态和动态两种情况，建立了典型炼钢连铸调度问题的数学模型；研究了问题的复杂性、解的表达方式、可行解的邻域结构、邻域变换技术、邻域搜索技术等；探讨了人工蜂群算法、果蝇优化算法、迁徙鸟群算法、迭代局部搜索算法等求解炼钢连铸调度问题的关键理论与技术；提出了协同蜂群优化调度算法、改进人工蜂群优化调度算法、Lagrangian松弛优化调度算法和人工蜂群重调度算法等。本项目共完成研究论文18篇，其中SCI检索12 篇，EI检索4篇，ISTP检索1篇。申请发明专利1项。培养博士后2名，博士生3名，硕士生2名。本项目的研究成果受到国内外学者的关注。经Web of Science检索，相关论文总被引225次。