基于自适应遗传算法的Kriging曲面拟合及应用  

Kriging Surface Interpolation and Its Application Based on Self-adaptive Genetic Algorithm

作　　者：魏振华[1] 刘志锋[1] 邵玉祥[1] 吴冲龙[1] 

机构地区：[1]中国地质大学(武汉),武汉430074

出　　处：《微计算机应用》2010年第7期17-21,共5页Microcomputer Applications

基　　金：国家重大专项<渤海湾盆地东营凹陷勘探成熟区精细评价示范工程>(2008ZX05051)

摘　　要：半变异函数是Kriging中的重要数学模型,也是描述矿床区域化变量特征的有效数学模型,其拟合模型参数的确定直接影响Kriging曲面拟合的精度。本文首先通过自适应调整遗传算法的变异概率,避免早熟的同时保证了算法的效率;其次利用该遗传算法改进Kriging中半变异函数模型实现曲面拟合;最后将其应用到油藏模拟中烃源岩表面的生成。该方法与距离平方反比法的拟合效果进行比较,得出结论为在实际工程应用中,采用改进Kriging插值得到的曲面与实际提供散点数据拟合更好,充分地体现了工程勘探数据的作用,符合工程需求。Semi-variant function as an important mathematical model of Kriging spatial analysis can effectively describe the features of the variants in some districts of ore deposit. Semi-variant function parameter estimation affects the Kriging surface interpolation precision directly. Firstly,this paper adjusts the mutation probability of genetic algorithm to avoid premature convergence and to guarantee the algorithm efficiency. Secondly,the paper improves the Kriging semi-variant function for surface interpolation using the self-adaptive genetic algorithm. At last,the improved Kriging is applied in creating hydrocarbon source rock surface of reservoir simulation. The simulation effect through Kriging is compared to the effect of inverse distance weighted method. The comparative result shows that the surface interpolated by improved Kriging fits better with the practical discrete points in the practical engineering application. The improved Kriging embodies the effect of engineering exploration data sufficiently and is more suitable for engineering requirement.

关 键 词：遗传算法 半变异函数 KRIGING 曲面拟合 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] P628.3[自动化与计算机技术—控制科学与工程]