面向进制转换和克隆进化的帝国竞争改进算法  

Decimal-Binary Conversion and Clonal Evolution Oriented Improved Imperialist Competitive Algorithm

作　　者：李斌 黄起彬 LI Bin;HUANG Qibin(School of Mechanical&Automotive Engineering,Fujian University of Technology,Fuzhou 350118,China;School of Transportation,Fujian University of Technology,Fuzhou 350118,China)

机构地区：[1]福建工程学院机械与汽车工程学院,福州350118 [2]福建工程学院交通运输学院,福州350118

出　　处：《计算机工程与应用》2022年第5期208-224,共17页Computer Engineering and Applications

基　　金：教育部人文社会科学研究规划基金(19YJA630031)。

摘　　要：帝国竞争算法(imperialist competitive algorithm,ICA)是一种被广泛应用于求解各类理论与实践问题的随机搜索智能优化算法,但它收敛过快的特性令其容易在求解复杂问题时陷入局部最优,故对ICA进行有针对性的改进十分必要。引入二进制转换和克隆进化机制,为算法的进化种群提供新的上升通道和进化模式,帮助进化种群跳出局部最优,从而提出了一种改进的帝国竞争算法(decimal-binary conversion and clonal evolution oriented improved imperialist competitive algorithm,DCCE-IICA)。此外,为修正经典ICA早熟导致的算法过早结束和群体多样性快速降低的缺陷,DCCE-IICA还辅以帝国分裂和出界点替换策略,以确保进制转化和克隆进化机制在改进算法执行中充分发挥区域深度探索和平衡资源分配的初衷。随后,经典函数测试集、CEC2017测试集及CEC2020测试集被用于检验DCCE-IICA在多个维度下对不同类型复杂问题的寻优能力。选取分别在经典函数测试集、CEC2017测试集和CEC2020测试集中表现优异的共14种典型算法,与DCCE-IICA进行实验结果比较。实验结果显示DCCE-IICA引入的改进机制在大多数情况下能够稳定且高效地提升算法性能,使得算法同时具备较好的收敛速度、收敛精度和求解鲁棒性。The imperialist competitive algorithm(ICA)is a random search intelligent optimization algorithm that is widely used to solve various theoretical and practical problems.However,it is liable to fall into a local optimum while solving complex problems because of its characteristics of fast convergence.Consequently,it is necessary to make targeted improvements on ICA.This paper introduces a clonal evolution mechanism in the context of decimal-binary conversion to provide a new ascending channel and evolutionary pattern for the algorithm’s evolutionary population,and it aims at helping the population jump out of the local optimum.Accordingly,the decimal-binary conversion and clonal evolution oriented improved imperialist competitive algorithm(DCCE-IICA)is proposed tentatively.In addition,to cope with the premature and the rapid reduction of population diversity caused by the excessively fast convergence,the DCCE-IICA is also supplemented by the empire split and the new out-of-bounds replacement strategy to help the decimal-binary conversion and clone evolution mechanism to fully play their roles of regional in-depth exploration and balanced resource allocation in the execution of the improved algorithm.Subsequently,a total of 14 kinds of intelligent optimization algorithms that are performed well in the classic function test set,CEC2017 test set and CEC2020 test set are selected to compare the experimental results with DCCE-IICA.The computational experiments and contrastive analysis show that the improved mechanisms introduced in DCCE-IICA can improve the algorithm performance stably and efficiently in most instances,and make the algorithm have good convergence speed,convergence accuracy and solution stability simultaneously.

关 键 词：进制转换 克隆进化 帝国竞争算法 帝国分裂 出界点替换 CEC2017 CEC2020 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]