粒子群算法在克里金三维地质建模中的应用

Application of particle swarm optimization to Kriging three-dimensional geological modeling

变差函数是克里金法中反映区域化变量空间变化特征的有效数学模型,由其确定的拟合模型参数直接影响插值精度.采用基于线性递减权值的粒子群算法估计变差函数参数,实验对比发现:所提方法比传统的最小二乘法和手工拟合法更精确,采用某油田小层数据进行地质建模,插值得到的数据点与实际离散数据拟合更好,能够体现该地区地质特征,取得较好应用效果.

Variogram is an effective model for reflecting the variation of regionalized variable space,and the parameters of its fitting model affect the accuracy of interpolation directly.LDIW-PSO was used to estimate the parameters of variogram,and the experiments showed that the proposed method was more accurate than the traditional least squares and hand-fitting method.In the experiment,a small layer of oil field was used as geological modeling data,and the interpolated data fitted better with the actual discrete data,reflecting the region's geological features.Experiments proved that the proposed method of this paper achieved good practical results.