基于生物行为机制的粒子群算法改进及应用

发布时间：2018-02-04 03:55

本文关键词： 粒子群优化算法生物行为机制群体感应寄生免疫细菌觅食　出处：《华南理工大学》2014年博士论文　论文类型：学位论文

【摘要】：粒子群优化算法（Particle Swarm Optimization, PSO）是一种典型的群体智能优化算法，其基本思想来源于对社会型群居生物的某些行为的模拟，算法的搜索过程基本不利用外部信息，仅以适应度函数值作为进化的依据，是一种以“生成+检验”为特征的自适应智能算法。粒子群优化算法在许多工程优化的实际问题中得到成功的应用，得益于其原理简单、容易实现、全局搜索能力强等特点。鉴于PSO算法的起源来自于模拟社会型群居生物的某些行为，算法本身存在一些难以克服的缺陷，因此将存在于自然界中的各种生物行为机制融入PSO是一种潜在的改进途径。本文在深入分析PSO研究现状的基础上，基于生物行为机制的视角提出了几种改进的PSO算法，使算法的寻优性能更加优越，并将改进算法应用于管理实践中的优化和预测问题，拓展了粒子群优化算法的应用领域。本文采用规范研究与应用研究相结合的方法，主要研究内容及成果包括： (1)粒子群优化算法存在的问题及改进途径。系统分析仿生优化算法的基本思想、分类及特性，研究群体智能算法的基本原理、主要方法和发展趋势，重点分析粒子群优化算法的基本原理、存在问题及原因，为探索粒子群优化算法的改进途径理清思路。 (2)基于细菌群体感应机制的粒子群优化算法。分析了将生物行为嵌入到粒子群优化算法实现改进算法寻优性能的相关研究，结合细菌群体感应机制提出了基于细菌群体感应的粒子群优化算法（PSOQS），采用智能计算中常用的6个标准测试函数进行仿真实验，验证了算法对SPSO的改进程度；考虑到细菌群体感应发生时机对细菌群体感应粒子群优化算法的影响，针对不同群体感应频率下的细菌群体感应粒子群优化算法进行测试分析，得到最佳感应频率的PSOQS；群体智能算法的种群规模会影响算法的成功率，在最佳感应频率PSOQS的基础上采用不同的种群规模对改进算法进行测试，并对实验结果进行了分析。 (3)基于寄生免疫机制的粒子群优化算法。在分析细菌寄生行为和生物免疫机制的基础上，提出了将寄生免疫机制嵌入粒子群优化算法的基本思想及实现途径，构造了基于寄生免疫机制的粒子群优化算法（PSOPI），具体思路是将具有很强寄生能力的寄生群采用精英学习机制以提高算法跳出局部极值的能力，宿主则针对寄生群的寄生行为产生获得性免疫，以增强宿主种群粒子的多样性，当迭代次数进行到既定的代数时发生寄生行为；将所得的改进算法PSOPI与CPSO和PSOPB的寻优性能进行比较分析，针对标准测试函数进行仿真实验。 (4)基于免疫逃避机制的粒子群优化算法。针对存在于自然界中寄生虫逃避宿主免疫效应的现象，深入探索生物协同进化过程中免疫逃避的策略和机制，将细菌免疫逃避机制嵌入粒子群优化算法，分别采用高斯变异（Gaussian）和柯西变异（Cauchy）来模拟寄生物的免疫逃避行为，形成基于免疫逃避机制的粒子群优化算法模型，设置相应的实验参数，针对标准测试函数进行仿真实验。 (5)基于生物觅食机制的粒子群优化算法。系统总结了生物觅食行为的规律，，深入探寻自然界中生物的觅食机制，将细菌觅食机制嵌入粒子群优化算法构造细菌觅食粒子群优化算法（PSOBF），将蜜蜂采蜜机制嵌入粒子群优化算法构造蜜蜂觅食粒子群优化算法（PSOHF），采用标准测试函数进行仿真，实验结果证实了改进算法的有效性。 (6)改进粒子群优化算法的实际应用。针对管理领域的配送中心选择问题和股指预测问题构建模型，采用基于生物行为机制的改进粒子群优化算法求解配送中心选址问题，改进算法对单个配送中心的案例求解效果表现良好，采用基于生物行为机制的改进粒子群算法训练神经网络对上海股票市场价格指数的样本数据进行拟合预测，验证了改进算法在预测方面的良好性能。
[Abstract]:Particle Swarm Optimization ( PSO ) is a typical swarm intelligence optimization algorithm . The basic idea of PSO is derived from the simulation of some behaviors of community living beings . The main contents and achievements of this paper are as follows : ( 1 ) The problems and ways to improve the particle swarm optimization algorithm are analyzed . The basic idea , classification and characteristics of bionic optimization algorithm are analyzed , the basic principle , main methods and development trend of swarm intelligence algorithm are studied , and the basic principle , problems and reasons of particle swarm optimization algorithm are analyzed . ( 2 ) Particle swarm optimization ( PSO ) algorithm based on bacterial population sensing mechanism is studied . Based on the study of the optimization algorithm of bacterial population , a particle swarm optimization algorithm ( PSOQS ) based on bacterial population induction is proposed . ( 3 ) A particle swarm optimization algorithm based on the parasitic immune mechanism is proposed . A particle swarm optimization algorithm ( PSOPI ) based on the parasitic immune mechanism is proposed . A particle swarm optimization algorithm ( PSOPI ) based on the parasitic immune mechanism is proposed . ( 4 ) A particle swarm optimization algorithm based on immune evasion mechanism is used to study the strategies and mechanisms of immune evasion in the process of biological co - evolution , and the immune evasion mechanism is embedded into particle swarm optimization algorithm . Gaussian and Cauchy mutation ( Cauchy ' s ) are used to simulate the immune evasion behavior of the parasite , and the particle swarm optimization algorithm model based on immune evasion mechanism is formed . The corresponding experimental parameters are set up , and the simulation experiment is carried out for the standard test function . ( 5 ) Based on the particle swarm optimization algorithm based on the biological feeding mechanism , the rules of the biological feeding behavior are summarized , and the feeding mechanism of the organisms in the natural world is deeply explored . The colony optimization algorithm ( PSOBF ) is constructed by embedding the feeding mechanism of the bacteria into the particle swarm optimization algorithm , and the honeybees ' swarm optimization algorithm ( PSOHF ) is constructed by embedding the honeybees into the particle swarm optimization algorithm , and the simulation is carried out by using the standard test function , and the experimental results confirm the effectiveness of the improved algorithm . ( 6 ) To improve the practical application of particle swarm optimization algorithm . Based on the improved particle swarm optimization algorithm based on the biological behavior mechanism to solve the distribution center location problem , the improved particle swarm optimization algorithm based on the biological behavior mechanism is adopted to solve the location problem of the distribution center , and the improved particle swarm optimization algorithm based on the biological behavior mechanism is adopted to fit the sample data of the Shanghai stock market price index , and the good performance of the improved algorithm in prediction is verified .

【学位授予单位】：华南理工大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP18;F830

【参考文献】