分布式计算智能理论及应用

项目名称： 分布式计算智能理论及应用

项目编号： No.61332002

项目类型： 重点项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 张军

作者单位： 华南理工大学

项目金额： 310万元

中文摘要： 面向实际应用中日益增长的复杂优化问题的求解需求，针对大规模环境下传统计算智能方法在计算效率方面存在的不足，开展分布式计算智能方法的理论与应用研究，旨在结合分布式计算的强大计算能力，突破计算智能方法在大规模优化中的性能瓶颈。开展分布式计算智能的基本框架及理论研究，建立并行粒度与算法性能之间的关联模型，为分布式计算智能方法的设计提供可靠的理论基础；开展分布式计算智能的关键技术研究，包括基于功能分割和协同进化的问题解耦优化技术，基于自适应信息迁移和拓扑结构的自适应通讯技术，以及结合机器学习方法的混合分布式计算智能技术，提高算法的鲁棒性、通用性和计算效率；在集成上述关键技术的基础上，进一步引入虚拟资源池及种群规模自适应技术，实现弹性的分布式计算智能系统。最终，将相关的研究成果推广应用于大规模资源分配与调度、物联网部署优化及智能交通物流调度等“智慧城市”相关的应用领域，有效配合“智慧广东”的建设。

中文关键词： 计算智能;分布式计算;进化计算;自适应;高维优化

英文摘要： In terms of the requirements of solving continuously emerging complex optimization problems in practice and the computing deficiency of traditional Computational Intelligence (CI) methods in large-scale environments, this project conducts research into the theory and application of Distributed Computational Intelligence (DCI). The primary goal is to utilize the powerful computation capability of distributed computing devices so as to overcome the performance bottleneck of CI in large-scale optimization. The basic framework and theory of DCI will be developed, and the correlation model between the granularity of parallelism and the performance of DCI algorithm will be established, which will lay a foundation for the design of DCI algorithms. Based on this, the key techniques for improving the robustness, universality, and efficiency of DCI algorithms will be developed, including problem decoupling techniques based on function partition and cooperate coevolution, adaptive communication techniques such as adaptive information mitigation and adaptive topology, and hybrid DCI techniques embedded with Machine Learning (ML). After integrating the above key techniques of DCI, we will realize highly scalable DCI systems by introducing techniques of virtualization and population size adaptation. Finally, the corresponding achievements will be promoted and applied in areas related to “Smart City”, such as large-scale resource allocation and scheduling, optimal deployment of the Internet of Things (IoT), and intelligent transportation and logistics scheduling, so as to conform with the construction of “smart Guangdong” effectively.

英文关键词： Computational Intelligence;Distributed Computing;Evolutionary Computation;Adaptation;High-Dimensions Optimization