基于拍卖理论与多代理系统的分散式资源受限多项目调度优化研究

发布时间：2018-10-14 15:07

【摘要】：项目管理是第二次世界大战后期发展起来的新管理技术之一,随着世界经济的发展,组织规模的扩大,在一系列项目管理大师的推动下,项目管理已经成为现代管理学中的重要组成部分。项目调度作为项目管理下的一个分支,自20世纪60年代(关键路径法的诞生)至今,对单项目调度的研究已经逐渐丰富完善,在现实中也有较多丰富的应用,而多项目调度在现实中常有应用,在研究领域则研究较少。考虑到现实中多项目环境的实际情况(项目各有调度细节并共享部分资源),本文以分散式资源受限多项目调度为研究对象,基于拍卖理论建立多代理系统模型研究多项目并行调度问题。本文的研究内容可分三个部分。第一部分研究考虑多模式单目标的多项目调度问题,在该问题中,本文将考虑六种属性的资源,包括本地可更新资源、本地不可更新资源、全局共享可更新资源、全局共享不可更新资源、全局专属可更新资源以及全局专属不可更新资源,并使用简单案例对所建立的算法体系进行验证。第二部分研究单模式多目标的多项目调度问题,以本地/全局可更新资源为例,将关键链方法与多代理系统模型相结合,使用拍卖理论解决资源冲突问题,并使用复杂案例对所建立算法体系进行验证。第三部分研究将进一步扩展本文所提方法的应用范围,讨论多代理系统与拍卖理论相结合,在更现实和广泛的多项目调度环境中的应用,并使用简单案例进行阐述。本文主要研究分散式资源受限多项目调度问题,主张在分散式环境下,建立和利用内部竞争机制,以达到兼顾各方利益的目的,从而实现资源在多项目调度中的优化配置。本文通过多代理系统建立基本架构,通过拍卖理论建立和解决竞争机制,描述了在不同多项目环境下如何具体建立求解模型,对多项目调度领域的研究和实践有积极的推动意义。
[Abstract]:Project management is one of the new management techniques developed in the late World War II. With the development of the world economy and the expansion of the scale of the organization, it is promoted by a series of project management masters. Project management has become an important part of modern management. Project scheduling as a branch of project management, since the 1960s (the birth of the critical path method), the study of single project scheduling has been gradually enriched and perfected, and there are also a lot of rich applications in reality. However, multi-project scheduling is often applied in reality, but it is less studied in the field of research. Considering the actual situation of multi-project environment (each project has scheduling details and sharing part of resources), this paper takes decentralized resource-constrained multi-project scheduling as the research object. A multi-agent system model based on auction theory is established to study the multi-project parallel scheduling problem. The research content of this paper can be divided into three parts. In the first part, we consider the multi-project scheduling problem with multi-mode, single-objective and multi-mode. In this problem, we will consider six kinds of attribute resources, including local renewable resources, local non-renewable resources, global shared renewable resources. The global shared non-renewable resources, global exclusive renewable resources and global exclusive non-renewable resources are used to verify the proposed algorithm. In the second part, the multi-project scheduling problem with single mode and multi-objective is studied. Taking the local / global renewable resources as an example, the key chain method is combined with the multi-agent system model, and the auction theory is used to solve the resource conflict problem. A complex case is used to verify the proposed algorithm. The third part of the study will further expand the scope of application of the method proposed in this paper, discuss the combination of multi-agent system and auction theory, in a more realistic and extensive multi-project scheduling environment, and use a simple case to illustrate. In this paper, we mainly study the decentralized resource constrained multi-project scheduling problem, and advocate the establishment and utilization of the internal competition mechanism in the decentralized environment, so as to achieve the purpose of taking into account the interests of all parties, and thus realize the optimal allocation of resources in the multi-project scheduling. In this paper, the basic framework of multi-agent system is established, and the mechanism of competition is established and resolved by auction theory, and how to build the solution model in different multi-project environment is described. It has positive significance to the research and practice of multi-project scheduling.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F713.359

【参考文献】