基于智能体多目标进化的应急资源调度模型与算法

项目名称： 基于智能体多目标进化的应急资源调度模型与算法

项目编号： No.61271301

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 无线电电子学、电信技术

项目作者： 刘静

作者单位： 西安电子科技大学

项目金额： 65万元

中文摘要： 应急物流随着一系列自然灾害和突发公共卫生事件的发生应运而生，其重要性不言而喻。应急物流的核心是应急资源调度，而应急资源调度可以建模为资源受限项目调度问题。本项目将通过解决资源受限项目调度问题来为应急资源调度提供新的解决方案。进化算法在求解复杂优化问题上已取得了丰硕的成果，而且我们前期项目中设计的单目标多智能体进化算法取得了优越的性能。因此，本项目将进一步挖掘多智能体系统与进化计算相结合的潜力，致力于开发不确定环境下的智能体多目标进化模型，并通过对资源受限项目调度问题的研究，提出基于新模型的资源受限项目调度算法、资源投资项目调度算法以及综合考虑资源受限与资源投资两方面的调度算法。最后，利用我们前期项目开发的数据挖掘和复杂网络技术对突发事件的特性进行研究，基于不确定环境下资源受限项目调度问题与应急物流的密切关系，研究与开发基于新模型的应急资源调度算法。

中文关键词： 多智能体系统；进化算法；资源受限项目调度；鲁棒性；应急资源调度

英文摘要： Since lots of natural disasters and emergent events happened in the last couple of years, the supply chain under emergent environment is becoming more and more important. We think the core problem in the supply chain under emergent environment is resource scheduling, and such a kind of resource scheduling can be modelled as resource constraint project scheduling problems. Thus, this project tries to provide new scheme for resource scheduling under emergent environment by studying resource constraint project scheduling problems. Evolutionary algorithms have achieved fruitful ourcomes in solving complex optimization problems, and the multi-agent evolutionary algorithms for single objective problems designed in our previous project showed excellent performance. Therefore, this project will further mine the potential of combining multiagent systems and evolutionary algorithms, and devote to developing multi-agent multi-objective evolutionary model for uncertainty environment, and proposing various new algorithms for resource constaint project scheduling problems based on the new model. Finally, by making use of data mining and complex network techniques developed in our previous project, the characteristics of emergent events will be studied, so that new model and algorithms can be proposed for resource scheduling u

英文关键词： Multi-agent systems；evolutionary computation；resource-constrained project scheduling；robustness；emergency resource scheduling