【作者】 刘丽珏；

【导师】 蔡自兴；

【作者基本信息】 中南大学 ， 计算机应用技术， 2008， 博士

【摘要】 人工免疫系统是一新的模拟自然免疫系统的人工智能方法,是基于生物免疫系统的功能、原理、基本特征以及相关理论而建立的,用于解决各种复杂问题的计算系统。其研究旨在通过深入探索生物免疫系统中蕴含的信息处理机制,建立相应的工程模型和算法,开拓新型智能信息处理系统。本文基于免疫克隆选择机制,提出了几种免疫进化算法,其中包括亲和度引导的粒群进化免疫算法,多样度引导的变异协同进化免疫算法和多种群免疫协同进化算法。通过相应算法在函数优化问题中的应用,验证了研究的结果,肯定了其具有解决复杂问题的潜力。同时将免疫进化策略应用于多移动机器人的协作路径规划,论述了采用免疫进化算法进行机器人路径规划的方法和实验结果。设计了带有进化决策和协同进化机制的多机器人体系结构。论文的主要工作如下:1.为了克服传统免疫进化算法所存在的收敛速度较慢等问题,通过理论分析和实验,提出了一种将改进的粒群进化方程引入免疫进化过程的新算法,并证明了其收敛性。2.探讨了变异算子对种群多样性的影响,并提出了一种多样度引导的,利用变异体种群和抗体种群协同进化的免疫算法,通过理论分析和仿真实验证实了该方法明显提高了解的多样性和收敛速度,并证明了其收敛性。3.在协同进化一般框架的指导下构造了免疫协同进化的一般框架,并提出一种多种群共享记忆集的免疫协同进化算法,该算法的特点是加入一个所有种群共享的全局记忆集以记录最优的协作行为,在各种群独立进化的过程中加入种群之间的协作信息交互,提高了协同进化算法的收敛速度。4.在进化计算框架下,融合全局并行搜索的克隆选择和启发式局部搜索的免疫疫苗接种以及免疫网络的行为控制,构造了一类用于单机器人和多机器人路径规划的免疫路径规划算法,并通过仿真实验和机器人平台实验验证了算法的有效性。总之本文对免疫克隆进化技术进行了研究和改进,提出了几种具有不同针对性的改进算法,并设计了能够进行进化计算的多机器人体系结构,将免疫进化运用到移动机器人的路径规划问题中,通过理论分析、仿真实验和移动机器人路径规划实验对提出方法的有效性可靠性和实用性进行了验证。更多还原

【Abstract】 Artificial Immune System (AIS) is a new intelligent method simulating natural immune system. It is a kind of computing system to solving many kinds of complex problems based on the functionalities, disciplines, characteristics and other related immune theories of biological immune system. The purpose of the AIS research is to extract the special information processing mechanisms contained in biological immune system, to build the corresponding models and algorithms, and to implement novel intelligent information processing systems.In this paper, some novel immune evolutionary algorithms based on the immune clonal selection are presented, including the affinity guided clone selection algorithm based on particle swarm optimization, the diversity guided immunity algorithm with mutation coevolution, and the multi population immunity coevolution algorithm. The applications of these algorithms to some numerical optimization tasks validate their potential of solving complex problem. Mean while, the immune algorithms are applied into cooperative path planning of multi mobile robots, the methods and experiment results of this kind of application are presented. And architecture of multi-robots system with evolutionary decision and coevolutional machnism is designed. The main work can be summarized as follows:1. In order to overcome the low convergence speed of ordinary immune algorithms, a noval immune algorithm with modified particle swarm evolutional equation is presented by analysis of theory and experiment, and its convergence is proved.2. The impact on the diversity of population of the mutation is discussed in great detail, and a diversity guided immune algorithm with mutation coevolution is presented. The theory analysis and simulation experiments prove that the algorithm improve the diversity of the population, and the convergence speed as well. The convergence of the algorithm is also proved. 3. Under the guidance of the general framework of coevolution, the framework of immune coevolution is established, and a multi population immune coevolution algorithm with sharing memory is presented. The character of this algorithm is recording the successful cooperative action in a sharing memory, so the coevolutional populations can exchange information in it and get a quicker convergence speed.4. Under the framework of evolutionary computation, immune path planning algorithms of robots are constructed, which integrate the vaccination with heuristic local search, clonal selection with parallel global search and immune network action control. The simulational experiments and expetiments of robots show that the method is effective.In a word, by studying the immune clone evolutionary algorithm, there are several modified immune evolutionary algorithms presented in this paper, architecture with evolutional capability of multi robots is designed, and the immune algorithms are applied into robots path planning. Theory analysis, simulational experiments and robots experiments all show that these algorithms and methods are effective.更多还原