人群搜索算法拟合Kriging参数的空间数据插值  

Spatial data interpolation of seeker optimization algorithm fitting Kriging model parameters

作　　者：胡芸 侯明勋[1] HU Yun;HOU Mingxun(School of Naval Architecture Ocean and Civil Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海交通大学船舶海洋与建筑工程学院,上海200240

出　　处：《湖北大学学报（自然科学版）》2023年第6期865-871,共7页Journal of Hubei University：Natural Science

基　　金：国家自然科学基金(41572255)资助。

摘　　要：在Kriging理论模型参数拟合过程中,为减小变异函数权重分配带来的相关误差,全面考虑空间相关性,得到更准确的变异函数模型.采用人群搜索算法(seeker optimization algorithm, SOA),以变异函数拟合值与观测值的加权残差平方和最小为目标建立适应度函数,考虑样本数据的各向异性,迭代搜索模型最优参数进行Kriging插值.实验结果表明,相比于加权最小二乘法拟合,采用SOA拟合Kriging模型参数时,变异函数的加权残差和降低5%~40%,在球状模型、指数模型和高斯模型下Kriging插值的绝对误差分别降低31.02%、24.02%和25.13%.In the process of parameter fitting of Kriging theoretical model,in order to reduce the correlation error caused by weight distribution of variogram and obtain a more accurate model of variogram by comprehensively considering spatial correlation,we adopted seeker optimization algorithm(SOA),established the fitness function with the objective of minimizing the square sum of weighted residuals between the fitting value of the variogram and the observed value.Considering the anisotropy of the sample data,the optimal parameters of the model were searched iteratively for Kriging interpolation.The experimental results show that the weighted residual sum of variogram is reduced by 5%~40%and the absolute error of Kriging interpolation is reduced by 20%~30%when the parameters of Kriging model are fitted by SOA compared with the weighted regression method.

关 键 词：KRIGING 人群搜索算法 空间插值 优化算法 变异函数 加权回归法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]