协同教学优化算法及其应用研究

项目名称： 协同教学优化算法及其应用研究

项目编号： No.61304082

项目类型： 青年科学基金项目

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： 邹锋

作者单位： 淮北师范大学

项目金额： 24万元

中文摘要： 教学优化算法是一种群智能优化算法，其结构简单、控制参数少，寻优能力在某些方面优于其他群智能优化算法，已在组合优化、数据分析等方面发挥了一定作用。然而，在处理复杂多模优化和转移优化问题时，迭代过程中种群多样性易于丢失，迭代进化后期收敛速度慢，优化精度不高。本项目首先对现有的教学式优化方法的优缺点进行理论分析和仿真，从群体的构成出发，分析种群个体间邻域拓扑结构对种群进化的影响，构建基于拓扑结构的教学优化模型；在此基础上，分析子群体间的协同对种群进化的影响,设计合适的多子群协同进化算法，开展维持种群的多样性，平衡算法的探测性与开发性，加快算法的搜索效率方面等方面的研究，提升其解决复杂优化问题能力；针对设计的模型和算法，分析各方法的复杂度和收敛性，并将其应用于组合优化、无线传感网络等不同领域。本项目的研究成果，将为教学优化算法性能提升及工程领域应用拓展提供理论与方法支撑，具有重要的理论和实践意义。

中文关键词： 教学优化；群体多样性；协同优化；邻域搜素；混合优化

英文摘要： Teaching-learning-based optimization algorithm (TLBO) is a class of swarm intelligence optimization algorithm. It has simple structure, less control parameters, better optimization ability than other swarm intelligence optimization algorithms in some ways. Hence it has been used in combinatorial optimization, and data analysis. However, for complex multimodal optimization problems and shifted optimization problems, population diversity is easy to lose in the iterative process, late the convergence speed slow and the final accuracy is not high. In this project, firstly, the impact of neighborhood topology structure on the population evolution is analyzed and teaching-learning-based optimization algorithm based on neighborhood topology structure is built. Secondly, on this basis, the impact of the cooperation of sub populations on the population evolution is analyzed and the suitable multi-population co-evolution mechanism is built, and thus to maintain the population diversity, balance the exploration and exploitation properties of the proposed algorithm, speed up the search efficiency, enhance computational performance on complex optimization problems. Then, the complexities and convergences of these models and algorithms are analyzed. Finally, these models and algorithms are used for different application field

英文关键词： Teaching–learning-based optimi；Population diversity；Cooperative optimization；Neighborhood search；Hybrid optimization