动态环境下基于耗散结构的新型粒子群算法及应用研究

项目名称： 动态环境下基于耗散结构的新型粒子群算法及应用研究

项目编号： No.71461027

项目类型： 地区科学基金项目

立项/批准年度： 2015

项目学科： 管理科学

项目作者： 刘衍民

作者单位： 遵义师范学院

项目金额： 34.5万元

中文摘要： 粒子群算法(PSO)是群智能领域中十分活跃的分支，设计PSO处理动态优化问题（DOP）已成为该领域的一个研究热点和难点。本项目借鉴耗散结构理论，从一个新的视角审视动态优化问题，系统研究面向动态环境的粒子群优化模型，算法设计及应用问题。具体包括：①建立面向动态环境的粒子群统一优化模型，揭示PSO处理DOP的关键机制；②以统一模型为指导，构建具有遗忘机能的自适应动态记忆模块，并设计个体选择机制和个体删除机制；③借鉴生物多样性思想，设计种群多样性提升机制；④建立基于回归理论的预测环境变化时间模型、基于马尔科夫链的预测环境变化方式模型，以提升种群对环境变化的应变能力；⑤提出带有记忆功能的动态PSO及扩展算法, 并将其用于优化一类动态闭环供应链模型。该研究不仅为动态环境下PSO设计与开发提供了新思路，丰富了PSO理论，而且为动态环境下闭环供应链模型求解提供了新方法，具有重要的科学意义和应用价值。

中文关键词： 动态环境；粒子群算法；记忆功能；耗散结构

英文摘要： Particle swarm optimization (PSO) is a very active branch in swarm intelligence field, and it is a hot and difficult topic designing PSO to solve dynamic optimization problem (DOP). The project gleans ideas from dissipative structure theory to scan the DOP from a new perspective, and systematically research PSO model in dynamic environment, algorithm design and application problems. The details of this project include: firstly, proposing a unified PSO model to reveal the key mechanisms while optimizing DOP with PSO; secondly, by the unified model guiding, constructing the dynamic memory module of adaptively adjustable size with forgotten function, and designs the selection and deletion mechanism; thirdly, devising the mechanism of improvement of the swarm diversity based on the biodiversity thought; fourthly, proposing the prediction model of environmental change approach based on Markov chain, and the prediction model of environmental change time based on theory of regression to improve the swarm adaptability on the environmental change; lastly, proposing dynamic PSO with memory function and its extension, and used them to optimize a kind of dynamic closed-loop supply chain model. The study achievements not only provide new ideas for PSO design and development, but also propose a new method for dynamic closed-loop supply chain solution, which has important scientific significance and application prospects.

英文关键词： Dynamic Environments;Particle Swarm Optimization;Memory function;Dissipative Structure