地质雷达数据中克里金插值采样数据选择算法

Sampling data selection algorithm for Kriging interpolation based on ground penetrating radar data

采样数据选择的合理性是克里金算法插值结果是否准确的前提。针对如何从地质雷达数据中选择恰当的克里金插值采样数据的问题,本文提出了两种克里金插值采样数据选择算法,即动态三维球体覆盖(D3DGC)选点算法和双次反距离选点(DRD)算法。两种算法在基于粒子群优化的普通克里金算法下验证了采样数据选择的合理性,实验结果分析表明,D3DGC算法在最佳选点个数下误差率比传统选点方法降低了6.7%,在其他选点个数下误差率(<10%)也明显低于传统选点方法,DRD算法在参数e和c为4以及最佳选点个数的条件下误差率也达到了6.2%,在其他情况下误差率也低于传统选点方法。

The rationality of the data sampling selection method is a prerequisite for judging whether the Kriging interpolation results are accurate. Considering the problem of how to properly choose sampling data from the probing data of ground penetrating radar,in this paper,we propose two sampling selection algorithms for Kriging interpolation: the dynamic three-dimensional globe cover( D3DGC) algorithm and the double reverse distance( DRD) algorithm. We verified the rationality of these two methods with respect to the ordinary Kriging algorithm based on particle swarm optimization. The experimental results show the D3 DGC algorithm to reduce the error rate to 6. 7% compared with the traditional method for an optimal number of sampling points. With respect to other sampling numbers,D3 DGC also demonstrated obvious advantages( error rate < 10%). The DRD algorithm error rate was 6. 2%for the optimal number of sampling points when the parameter e and c values were 4,respectively. For other sampling numbers,the DRD algorithm also obtained better results than the traditional method.