一种基于随机粒子群的变差函数优化方法  

Based on the Variogram of the Random Particle Swarm Optimization Method

作　　者：成洪甲[1] 杨雨[1] 孙寅萍[1] 

机构地区：[1]东北石油大学,大庆163318

出　　处：《科学技术与工程》2012年第31期8373-8378,共6页Science Technology and Engineering

摘　　要：变差函数是克里格法中反映区域化变量空间变化特征的有效数学模型,其在克里格插值算法中是重要的环节,由其确定的拟合模型参数直接影响插值精度。由于变差函数采用非线性模型,导致拟合误差曲面存在很多局部极值点。因此传统的最小二乘法不易达到满意的结果。基于这一问题,提出了基于随机粒子群优化(RPSO)算法的变差函数设计方法。实验结果验证了该方法的有效性。The variogram is reflected in regionalized the kriging. It is an important part of Kriging algorithm. variable spatial variation of effective mathematical model in By its determined by fitting the model parameters directly affect the interpolation accuracy. Because the nonlinear model is used in the variogram, there are many local ex- treme points that exist in fitting error surface, so the traditional least square method is not easy to achieve satisfacto- ry results. Based on this problem, variogram design method is put forward that based on random particle swarm （RPSO） algorithm, and the experimental results verify the validity of the method.

关 键 词：克里格法 随机粒子群优化 变差函数 插值 非线性模型 

分 类 号：TE19[石油与天然气工程—油气勘探]