基于整体编码遗传算法的探地雷达全波形反演

探地雷达(GPR)的全波形反演(FWI)能精确刻画出地下介电常数模型,得到了广泛研究。但全波形 反演受初始模型、反演算法的影响,容易使结果陷入局部最优解,难以准确地反演出地下真实情况。为反演隧道衬 砌空洞的病害情况,文中提出了一种基于整体编码遗传算法(OCGA)的全波形反演方法。该方法在遗传算法的基础 上改进了编码策略,对个体的整体特征进行编码。实验表明,对于预设的圆形空洞(或钢筋)物理模型,整体编码策 略能使结果更接近真实解。该算法能在不依赖初始模型的前提下大大改善结果收敛于局部最优解的问题,并进一 步量化反演结果空洞的位置与大小。

GPR Full Waveform Inversion Based on Overall Coding Genetic Algorithm

The full waveform inversion (FWI) of the ground penetrating radar (GPR) can accurately depict the underground dielectric constant model and has been widely studied. However, due to the influence of initial model and inversion algorithm, the result of full waveform inversion is easy to fall into local optimal solution, and it is difficult to accurately reflect the real underground situation. A full waveform inversion method based on overall coding genetic algorithm (OCGA) is proposed to solve the problem of tunnel lining cavity disease inversion. In this method, the coding strategy is improved based on the genetic algorithm to encode the overall characteristics of individuals. Experimental results show that the overall coding strategy can make the result closer to the real solution for the preset circular void (or rebar) physical model. The algorithm can greatly improve the convergence of results to the local optimal solutions without relying on the initial model, and further quantify the location and size of the inversion results.