微扰动和混合变异的差分生物地理学优化算法  

Differential biogeography optimization algorithm based on micro-perturbation and mixed variation

作　　者：葛强[1,2] 李玉晶 乔保军[2] 左宪禹[1,2] 王更科 GE Qiang;LI Yu-jing;QIAO Bao-jun;ZUO Xian-yu;WANG Geng-ke(Institute of Data and Knowledge Engineering,Henan University,Kaifeng 475004,China;Henan Key Laboratory of Big Data Analysis and Processing,Henan University,Kaifeng 475004,China;Aerospace Information Research Institute,Chinese Academy of Sciences,Beijing 100101,China;University of Chinese Academy of Sciences,Chinese Academy of Sciences,Beijing 100101,China)

机构地区：[1]河南大学数据与知识工程研究所,河南开封475004 [2]河南大学河南省大数据分析与处理重点实验室,河南开封475004 [3]中国科学院空天信息创新研究院,北京100101 [4]中国科学院中国科学院大学,北京100101

出　　处：《计算机工程与设计》2021年第2期432-441,共10页Computer Engineering and Design

基　　金：国家自然科学基金项目(U1704122);国家重点研发计划课题基金项目(2017YFD0301105);河南大学研究生教育创新与质量提升计划基金项目(SYL18020103)。

摘　　要：为增强生物地理学优化算法(biogeography-based optimization,BBO)的优化能力并克服其不能很好平衡开发能力与避免陷入局部最优解之间的矛盾,提出基于微扰动和混合变异的差分生物地理学优化算法(differential biogeography optimization algorithm based on micro-perturbation and mixed variation,MDEBBO)。引入差分变异算子和自适应的微扰动因子来改进迁移算子,使算法朝着最优解快速移动,提高算法的查找精度。采用混合变异算子代替原变异算子,在迭代前期算法具有良好的全局探索能力,在后期具有较优的局部开发性。基准测试函数的仿真结果表明了MDEBBO算法的有效性。通过MDEBBO算法对Richards模型进行参数估计预测谷氨酸菌体生长浓度,实验结果表明,MDEBBO算法较对比算法更适用于Richards模型的参数估计。To enhance the optimization ability of biogeography-based optimization algorithm(BBO)and to overcome the contradiction between balancing the development ability and avoiding falling into a local optimal solution,a differential biogeography optimization algorithm based on micro-perturbation and mixed mutation(MDEBBO)was presented.The differential mutation operator and adaptive perturbation factor were introduced to improve the migration operator,which made the algorithm move toward the optimal solution quickly and improved the search accuracy of the algorithm.The hybrid mutation operator was used to replace the original mutation operator,which made the algorithm have good global exploration ability at the early stage of iteration and better local development at the later stage.Results of simulation experiments based on benchmark functions show the effectiveness of the MDEBBO algorithm.The parameters of Richards model were estimated using MDEBBO algorithm to predict the growth concentration of glutamic acid bacteria.Experimental results show that the MDEBBO algorithm is more suitable for the parameters estimation of the Richards model than the comparison algorithm.

关 键 词：生物地理学优化算法 差分变异算子 微扰动因子 混合变异算子 参数估计 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]