基于PSO-GA的Kriging插值法建立透地通信分层地层媒质模型

Establishment of a layered stratum model for though-the-earth communication based on the Kriging interpolation method with PSO-GA

地层层系的不同分布对透地通信电磁波的传播特性有重要影响。Kriging插值法利用地层媒质的有限采样点可有效建立分层地层媒质模型,其中的插值计算过程中变异函数模型参数的优化是准确构造地层分层模型的关键。为更好地解决由于采样点分布不匀造成的插值点与采样点之间的误差,使求出的构建地层分界面的插值点更加逼近采样点,从而更精确地建立地层分层模型,提出了一种基于粒子群-遗传(PSO-GA)的Kriging插值法,对变异函数模型的参数进行优化。相比于基于PSO的Kriging插值法与基于GA的Kriging插值法,所提出的算法在粒子寻优迭代整个过程中通过粒子间的两次信息交换,不仅解决了陷入局部最优的问题,实现了全局最优样本,还加快了收敛的速度。仿真与分析结果表明采用所提出的基于PSO-GA的Kriging插值法建立透地通信分层地层媒质模型,相比于基于PSO的Kriging插值法与基于GA的Kriging插值法可获得更好的寻优性能及寻优稳定性。

The different distributions of stratum layer have an important influence on the propagation characteristics of electromagnetic wave through the earth communication. Kriging interpolation method utilizes the finite sampling points of the stratum to establish the layered medium model effectively,in which the parameter optimization of variation function model is the key to the accuracy of the layered medium model. To solve the error problem between the interpolation points and the sampling points which is caused by the unevenly distributed sampling points,and make the interpolation points more close to the sampling points for establishing a layered stratum model more accurately,a Kriging interpolation method based on particle swarm optimization-genetic algorithm（ PSO-GA） jointly is proposed to optimize the parameters of variation function model. Compared to Kriging interpolation method based on the PSO or the GA,the proposed algorithm not only solves the problem of falling into the local optimum and achieves a global optimal samplebut also speeds up the convergence through the information exchange among particles twice during the iteration searching. Simulation and analysis results show that a better searching performance and optimizing stability are achieved with the proposed algorithm for establishing a layered stratum model for though-the-earth communication than that with Kriging interpolation method based on the PSO or the GA.