基于粒子群算法的GPR与电测深共同反演及等值现象消除

Combined inversion of resistivity sounding and GPR to eliminate equivalent phenomena based on multiple group particle swarm optimization algorithm

电测深曲线的同层等值现象普遍存在于观测误差范围内,地电断面的模型参数不同,而对应的电测深曲线却完全相同,给层状介质的电测深反演带来极大困扰。为消除电测深中等值现象,如实反映层状介质的各层电性参数、厚度等信息,这里提出一种改进的多种群粒子群反演算法,增加同等模型下包含探地雷达几何参数信息的新全局极值点,实现电性参数与几何参数的共同反演,消除单一电测深曲线的S等值现象。通过H型和A型三层电测深曲线的S等值现象数值算例,并将改进的多种群粒子群反演结果与标准粒子群进行了分析和对比。结果表明,加载了探地雷达几何参数信息的改进型多种群粒子群算法,能有效地消除等值现象,更准确地反演出地电模型的各层电性参数和几何参数。

The same layer equivalence phenomenon of electric sounding curve is common in the range of observation error, the model parameters of the ground electric section are different, and the corresponding electric sounding curves are identical.These bring great trouble to the inversion of the layered medium. In order to eliminate the equivalence phenomenon in electrical sounding, and to reflect the information of electrical parameters and thickness of layered media, this paper presents an improved algorithm of multiple group particle swarm inversion. This increases the new global extremum of the geometrical parameter information of ground penetrating radar under the same model and realizes the joint inversion of electrical parameters and geometrical parameters. To eliminate the S equivalence phenomenon of single electrical sounding curve. A numerical example of S equivalent phenomenon in the H-type and A-type three-layer electrical sounding curve compared with the standard particle swarm, the improved multiple particle swarm inversion results show that the improved multi population particle swarm optimization algorithm, which is loaded with the geometrical parameter information of GPR, can effectively eliminate the equivalence phenomenon, more accurate inversion of electrical parameters and geometric parameters of the Earth-electric model.