摘要
将二维卷积运算引入智能优化算法的种群位置更新过程,提出一种新的智能优化算法,即卷积优化算法(Convolution Optimization Algorithm,COA)。该算法主要包括卷积搜索和解质量增强2种机制:在卷积搜索过程中,分别定义纵向卷积核、横向卷积核和区域卷积核,依次进行二维卷积运算并更新种群的位置向量,然后将3种卷积核更新后的种群的位置向量进行随机权重或等比例权重相加,进一步更新种群的位置向量;在解质量增强过程中,对最优解的搜索空间逐维进行带非惯性权重的高斯变异,并对最优解进行扰动,从而提高算法的局部搜索能力。结合其他几种算法对12个不同类型的基准函数进行仿真实验,结果表明:COA具有优异的收敛精度、收敛速度、稳定性和寻优能力。
Two-dimensional convolution operation is introduced into the updating process of population position of intelligent optimization algorithm,and a new intelligent optimization algorithm is proposed,which is called Convolution Optimization Algorithm(COA).COA mainly includes two mechanisms:convolution search and quality enhancement.In the process of convolution search,longitudinal convolution kernel,transverse convolution kernel and regional convolution kernel are defined respectively,and the position vector of the population is updated by two-dimensional convolution operation in turn.Then,the position vectors of the population updated by the three convolution kernels are randomly weighted or equal-scale weighted addition,and the position vector of the population is further updated.In the process of solution quality enhancement,gaussian mutation with non-inertial weight is carried out dimensionally in the search space of the optimal solution,and the optimal solution is disturbed,so as to improve the local search ability of the proposed algorithm.Comparing with other algorithms,the performance of COA is verified in 12 sets of benchamark function test,and the results demonstrate that COA has excellent convergence accuracy,convergence speed,stability and optimization ability.
作者
陈克伟
魏曙光
张嘉曦
CHEN Kewei;WEI Shuguang;ZHANG Jiaxi(Army Academy of Armored Forces,Beijing 100072,China)
出处
《装甲兵学报》
2023年第1期102-108,共7页
Journal of Armored Forces
关键词
卷积优化算法
智能优化算法
卷积核
横向卷积
纵向卷积
区域卷积
convolution optimization algorithm
intelligent optimization algorithm
convolution kernel
transverse convolution
longitudinal convolution
regional convolution
作者简介
陈克伟(1983—),男,讲师,硕士。
