基于离散入侵性杂草优化和问题结构特性的批量流调度方法研究

In recent years, the lot-streaming flow shop scheduling problem, which has important applications in tile, transformer core, solar cell, and many other modern industries, has become a hot area of research. Intelligent optimization algorithms have become the main solution techniques. The existing scheduling methods, however, make little use of the problem-specific properties to improve their performance. And almost all the literature focuses on a single objective. The scheduling method for the multi-objective lot-streaming scheduling problems is still in its infancy. To our best knowledge, there is no literature on the lot-streaming scheduling problems in dynamic environments. Therefore, in this proposal, based on our previous work, we intend to study the lot-streaming scheduling methods based on the discrete invasive weed optimization (DIWO) and the problem-specific characteristics. For the DIWO, we will study the novel operators, co-evolution mechanism, selection of control parameters based on learning mechanism, convergence, computing complexity, and finite time performance. For the problem-specific properties, we will research neighborhood structures, neighborhood connectivity, fitness landscape and etc. we will explore the effective combination of the DIWO and the problem-specific characteristics of the lot-streaming problems. We will propose a series of high-performance algorithms for single objective, multi-objective, and dynamic lot-streaming problems. The research production is expected to be applied in the production process of tile, transformer core, solar cell and many others.

近年来，广泛存在于瓷砖、变压器铁芯和太阳能电池等生产过程的批量流调度问题成为国际上的研究热点之一。智能优化方法成为最主要的求解方法。但现有调度方法较少利用问题特性来提高算法性能，且主要针对单目标的调度问题。多目标批量流调度问题的研究还刚刚起步，缺乏针对动态不确定批量流调度问题的研究。本项目在前期探索的基础上，将深入研究基于离散入侵性杂草优化（DIWO）和问题特性的批量流调度方法，包括研究新颖的DIWO操作算子、协同进化的算法框架、基于学习机制的控制参数设定方法以及算法收敛性、计算复杂性、有限时间性等理论，研究批量流调度问题的邻域结构、邻域连通性和适应度地貌等问题特性，探索DIWO与问题特性的有机结合方法，提出针对单目标、多目标和动态批量流调度问题的高性能优化算法，给出一批具有工程应用价值的批量流调度理论成果, 可望直接服务于我国瓷砖、变压器铁芯和太阳能电池等企业的生产过程。

近年来，批量流调度问题在学术和工业界获得了广泛关注和研究。本项目基于离散入侵性杂草优化等新型群智能优化算法，围绕典型的批量流调度、复杂约束的批量流调度和不确定条件下的多目标批量流调度等问题展开研究。针对典型的批量流调度问题研究了离散入侵性杂草优化算法初始种群的产生方法、操作算子、杂草种子扩散方法、插入邻域结构、局部精确搜索算法，设计了协同进化的算法框架。针对无等待流水车间和阻塞流水车间等复杂约束的批量流调度问题，研究了问题的特征、目标特点、约束条件等，设计了性能良好的问题编码、操作算子和邻域结构，提出了迁徙鸟群和果蝇优化算法。研究了不确定条件下的多目标批量流调度问题，提出了基于离散入侵性杂草优化和迁徙鸟群优化等新型算法的混合调度算法等。研究成果不仅有助于丰富和深化批量流调度理论，而且将促进离散优化调度算法的应用和推广，为企业带来极大的经济效益和社会效益。本项目完成了研究计划，取得了一系列有特色的创新性成果。三年来，项目组共完成相关学术论文36篇，其中SCI收录论文18篇，EI收录论文16篇，中文核心期刊论文1篇。参加国际会议并作分组报告8次。获得山东高等学校优秀科研成果奖一等奖和三等奖各1项。申请专利3项，获批软件著作权3项，培养硕士研究生4名。