一种基于随机粒子群的变差函数优化方法

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。

Based on the variogram of the random particle swarm optimization method

The variogram is reflected in regionalized variable spatial variation of effective mathematical model in the kriging.It is an important part of Kriging algorithm. 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 extreme points that exist in fitting error surface, so the traditional least square method is not easy to achieve satisfactory results. Based on this problem, this paper puts forward variogram design method that based on random particle swarm(RPSO) algorithm, and the experimental results verify the validity of the method.