A Comprehensive Framework and Solution Procedure for Dynamic Scheduling in Manufacturing Systems

制造业是加拿大经济中最重要的创造财富的部门。然而，这个非常重要的经济部门正面临激烈的全球竞争。为了在这种竞争环境中取得成功，制造商需要显著提高能力。技术，以设备和自动化的形式，一直是竞争力的手段。然而，一个同样重要和决定性的竞争力途径是我们组织设施、计划生产和安排运营的方式。生产计划和调度已经成为生存的必需品，其重要性怎么强调都不为过。随着时间的推移，将有限的资源分配给任务，同时满足活动和资源之间的大量时间和关系约束，这是一个关键的决策制定过程。事实上，它最终决定了制造系统的运行性能。它也是科学界所知的非常困难的优化问题之一，一直是一个活跃的研究领域。在这个领域发表了无数的文章。然而，关于如何将理论研究付诸实践的文章却很少。也许研究界过分强调了数学的严谨性，而忽略了实际的实现。将调度算法转换为实际系统的一个主要问题是，大多数算法都是特定于框架的，例如作业车间或流程车间，很少与实际系统匹配。在这项研究中，我们将尝试开发一个框架和调度方法，可以应用于更广泛的制造业。这将使行业采用自动化调度工具，而无需开发昂贵的专门算法。文献中调度算法的另一个主要限制是，大多数调度算法都是为了解决假设静态条件的数学模型而设计的。然而，制造系统在动态/随机环境中运行，频繁发生不可预测的实时事件，使得先前可行的计划变得不可行和过时。这种实时事件的例子包括机器故障、紧急工作的到达、订单的取消、截止日期的更改、操作员缺席等。为了解决这一问题，本研究将开发一个统一的监控和动态调度工具。在开发通用调度方法（这是本建议的主要内容）的同时，还将与当地工业合作开发许多专门的算法（不同于标准研究问题），以满足具体的调度要求。因此，这一建议和研究项目为许多研究生和本科生在复杂工业系统的高级规划、调度和控制方面提供了充足的训练机会。

Manufacturing is the most important wealth-generating sector of the Canadian economy. However, this very important economic sector is facing intense global competitions. To be successful in this competitive climate, manufacturers require significantly improved capabilities. Technology, in the form of equipment and automation, has always been a means to competitiveness. However, an equally important and a determining avenue to competitiveness is the way we organize facilities, plan production and schedule operations. Production planning and scheduling has become a necessity for survival and its importance cannot be overstated. It is a critical decision making process of allocating limited resources to tasks over time, while a large number of time and relationship constraints among the activities and resources are being satisfied. Indeed, itultimately determines the operational performance of a manufacturing system.It is also one of the very difficult optimization problems known to the research community and has been an active field of research. Countless number of articles are published in the area. However, very little has been written on how to bring theoretical research into practice. Perhaps the research community has over-emphasized the mathematical rigor at the cost of under-emphasizing real implementations. One major problem in translating scheduling algorithms to real systems is that most algorithms are framework specific such as job-shop or flow-shop which rarely match real systems. In this research, we will attempt to develop a framework and a scheduling methodology that can be applied to a wider range of manufacturing industries. This will enable industries adopt automated scheduling tools without the need to develop costly specialized algorithms. Another major limitation of scheduling algorithms in literature is that most of them are designed to solve mathematical models that assume static conditions. However, manufacturing systems operate in dynamic/stochastic environments where frequent unpredictable real-time events occur making a previously feasible schedule infeasible and obsolete. Examples of such real-time events include machine failures, arrival of urgent jobs, cancellation of orders, due date changes, operator absentee etc. To address this problem, in this research, a unified supervisory control and dynamic schedule tool will be developed. While working in developing generalized scheduling methodologies which are the main thrust of this proposal, many specialized algorithms (different from standard research problems) will also be developed for specific scheduling requirements in collaboration with local industries. Consequently, this proposal and the research program brings ample training opportunities for many graduate and undergraduate students in advanced planning, scheduling and controlling of complex industrial systems.