用于约束优化的人工免疫响应进化策略

A Novel Evolutionary Strategy Based on Artificial Immune Response for Constrained Optimizations

公茂果[1] 焦李成[1] 杜海峰[1,2] 马文萍[1]

基于克隆选择学说及生物免疫响应过程的相关机理，探讨一种新的人工免疫系统模型——人工免疫响应，提出用于解决约束优化问题的人工免疫响应进化策略；基于算法网络拓扑结构的分析表明，新算法比传统的进化策略（μ，λ）-ES具有更大的收敛概率．对10个标准测试问题的测试结果表明，与采用随机排序的进化策略和采用动态惩罚函数的进化策略相比，新算法在收敛速度和求解精度上均具有一定的优势．[著者文摘]

Based on the clonal selection theory and mechanisms of biological immune response, a novel artificial immune systems model, Artificial Immune Response （AIR）, is discussed. And based on Artificial Immune Response a novel evolutionary strategy for constrained optimizations is put forward. Analysis of its network framework shows that the new algorithm is convergent with a higher probability than （μ,λ） evolutionary strategy. The experiments on 10 benchmark problems show that when compared with the （μ,λ） evolutionary strategies adopting stochastic ranking method and dynamic penalty function method, the new evolutionary strategy is capable of improving the search performance significantly no matter in convergent speed or precision.[著者文摘]

改进人工免疫算法的分析研究

郑日荣[1] 毛宗源[1] 罗欣贤[2]

详细分析了人工免疫算法(AIA)基于信息熵计算抗体浓度存在的缺点，提出了三种新的浓度计算方法。分析了基于新的浓度计算方法的各种改进的人工免疫算法的收敛性。时两种变形大海捞针类函数的优化结果表明，改进的人工免疫算法较传统的人工免疫算法有较大的优越性。

人工免疫算法在函数优化中的应用

罗印升 李人厚 张雷 刘芳

基于自然免疫系统中体液免疫响应的机制提出了一种函数优化算法，该算法模拟了抗体的产生、抗体与抗原的黏合、激励、克隆、超突变及未受激励细胞的消亡等自然过程，其主要步骤包括：抗原、B细胞的算法定义，B细胞与抗原之间的亲和度计算与选择，B细胞的克隆、变异和记忆细胞的产生等．算法的主要特点是模拟了不同的自然机制，具有并行性，产生了高亲和度、长寿命的记忆细胞并不断对其更新．用两种不同类型的测试函数进行了仿真实验，并与基本遗传算法进行了比较，结果表明所提算法能以较快的速度完成给定范围的搜索和全局优化任务．

一类用于函数优化的基于混沌搜索的免疫算法

Chaotic-search-based immune algorithm for function optimization

左兴权[1] 范玉顺[2]

将混沌优化算法与克隆选择算法相结合，提出了一类基于混沌搜索的免疫算法．首先利用解空间变换将优化变量表示为混沌变量，并将混沌变量编码为抗体．然后，利用混沌变量的遍历性和随机性特点，通过在高亲和力抗体的邻域内进行混沌搜索以实现局部寻优，通过在整个解空间内的混沌搜索来避免陷入局部最优解．数值仿真结果表明该算法具不易陷入局部最优、解的精度高和操作简单等优点．[著者文摘]

A chaotic-search-based immune algorithm（CSIA） is proposed by integrating chaos optimization algorithm and clonal selection algorithm. Firstly, optimization variables are expressed by chaotic variables through solution space transformation, and the chaotic variables are coded into an antibody. Secondly, by making use of the characteristics of ergodicity and randomness of chaotic variables, the local search is performed by chaotic search in the neighborhoods of high affinity antibodies, and the chaotic search is carded out in the whole solution space to avoid sticking at local optima. The results of numerical simulation show that the algorithm has advantages, such as avoiding local optima, high precision solution and simple operations.[著者文摘]

一种用于函数优化的免疫算法

刘丽珏[1] 唐琎[1,2] 蔡自兴[1]

人工免疫系统是基于生物免疫系统特性而发展的新兴智能系统.利用免疫系统的克隆选择机制,提出一种用于函数优化的改进免疫算法.其主要特点是采用克隆和自适应变异等操作,提高收敛速度和种群的多样性.仿真程序表明,该算法能以较快速度完成给定范围的搜索和全局优化任务.[著者文摘]

一种用于函数优化的免疫算法

An Immune Algorithm Applied to Function Optimization

于瀛 侯朝桢

遗传算法是目前最为广泛使用的可以用于函数优化的寻优方法之一。针对其容易陷入局部极值点等弱点，该文基于生物免疫系统中的学习机理及与其相关的免疫学理论中的克隆选择学说，提出了一种新的用于函数优化的免疫算法。新算法包括选择、克隆扩展、超变异和免疫记忆操作，定义了体现算法学习机制的学习参数和用于保存最优解的免疫记忆集合。提出了根据算法亲合度自适应调节学习参数的方法，以提高算法的全局寻优能力。用不同类型的测试函数进行仿真实验，结果表明该算法是有效的。[著者文摘]

Genetic algorithm is one of the most widely used optimization methods applied to function optimization so far. However, it has such weaknesses as easy to get trapped into local optimal. This paper proposes an immune algorithm applied to function optimization based on the learning mechanism in natural immune system and the clonal selection theory in immunology. This algorithm includes selection, clone, hyper-mutation and re-selection operations. It defines a learning parameter to embody the learning mechanism and an immune memory set to keep the optimal results. The learning parameter adjusts adaptively with the affinity to promote the global search ability. Different testing functions are utilized to this method and the simulation results show this algorithm has good performance.[著者文摘]

用于高维函数优化的免疫记忆克隆规划算法

杜海峰 公茂果 焦李成 刘若辰

基于抗体克隆选择学说及免疫记忆理论，系统阐述了一种新的人工免疫系统算法，免疫记忆克隆规划，并基于Markov链的有关性质，证明了该算法的收敛性．高维函数优化的仿真试验表明，与相应的进化算法(如BGA)相比，算法有效克服了早熟问题，保持了抗体的多样性，而且收敛速度快。

面向多模态函数优化的混沌免疫网络算法研究

Study on chaos immune network algorithm for multimodal function optimization

吕佳[1,2] 熊忠阳[2]

针对人工免疫网络解决多模态函数优化时可能出现的早熟收敛现象和搜索精度不甚满意的问题，提出改进的混沌免疫网络算法。改进算法终止务件及采取相应措施以避免早熟，利用混沌变量来模拟免疫细胞的增殖方式以提高算法的搜索精度。通过对一些典型测试函数进行仿真实验．结果表明该算法能够快速优化抗体，搜索能力强，搜索精度高，是一种效果优良的解决多模态函数优化问题的极值寻优方法。[著者文摘]

Muhimodal function optimization problem has important applications in the fields of engineering, which requires finding the global optimum and as many as possible local optima. Artificial immune network solving multimodal function optimization may generate premature convergence phenomena. Its searching precision can＇t satisfy us yet. So improved chaos immune network algorithm was presented in this paper. Termination conditions as well as relevant measures were improved to avoid possibly generating premature convergence, and chaos variable was used to simulate proliferation mode of immune cells to enhance searching precision, Some classical functions were applied to test the performance of the algorithm. The simulation results illustrate that this algorithm can fast optimize antibodies and take advantage in searching ability and searching precision.[著者文摘]

用于函数优化的改进免疫克隆多样性算法

莫宏伟 金鸿章

免疫系统具有许多可以用于解决工程和科学问题的特性，人工免疫系统是基于免疫系统特性而发展的新兴智能系统．针对免疫算法的多样性，利用免疫系统克隆选择和克隆抑制机制，提出一种用于函数优化的改进免疫克隆多样性算法，以达到简化复杂系统函数算法的目的．多样性算法是一种基于免疫系统抗体多样性机制而开发的免疫算法，这里给出了改进的算法步骤，指出改进算法与原算法的主要区别以及所依据的免疫系统原理；文章还对算法的复杂性进行了分析，证明了改进免疫系统克隆多样性算法可以利用相对小的候选解群体解决复杂函数优化问题．

免疫克隆算法参数的均匀设计

Uniform Design of the Parameters of Immune Clonal Algorithms

高尚[1,2] 陈建忠[3]

人工免疫系统是基于生物免疫系统特性而发展的新兴智能系统,本文介绍了人工免疫系统中的克隆选择原理。免疫克隆算法的参数设置通常是依靠经验和试验来确定,造成试验工作量大且难以得到最优的参数组合,影响了算法的使用。通过将免疫算法基本模型的参数设定问题描述成均匀设计中多因素、多水平的试验设计,从而能够用较少的试验很快设定算法参数的取值。仿真试验表明了该方法的可行性和有效性。[著者文摘]

An artificial immune system is a novel intelligent system based on the characteristics of biological immune systems. The clonal selection principle in artificial immune systems is introduced. In general the parameters of an immune clonal algorithm are determined by experience and experiments. This leads to heavy work load and makes the optimal combination of the parameters difficult to obtain. A uniform design method is used to convert the problem of parameter establishment into the experimental design of multiple factors and multiple levels, and reduces the work load of determining the parameter values based on less experiments. The method applied to the benchmark problems shows good performance feasibility and effectiveness.[著者文摘]