基于SA算法反演层状介质介电常数的算法

Inversion Algorithm of Permittivity of Layered Media Based on SA Algorithm

为解决现有方法在反演层状介质介电常数时耗时长、效率低和对薄层介质不敏感的问题,提出了一种基于改进的模拟退火算法反演层状介质介电常数信息的新方法。该方法利用电磁波在各介质层中的双程时延的先验信息来减少反演参数的个数,降低算法复杂度,提高反演效率;利用非均匀变异思想改进传统模拟退火算法的扰动模型来加快收敛速度。同时,通过模拟回波与实际回波的差值来提取被强回波覆盖的时延信息,增强算法对薄层介质的识别能力,提高算法的实用性。实验结果证明了该方法对层状介质介电常数的反演具有很高的准确性和稳定性。

To solve problems of the existing algorithm that inversion of permittivity of layered media is time-consuming, inefficiency and not sensitive to the thin-layer media, a new inversion method is proposed based on improved simulated annealing algorithm. The method used prior knowledge of two-way travel time of electromagnetic wave in each dielectric layer to lower the number of parameters of inversion, reduce the algorithm complexity and improve the efficiency of the inversion. And the disturbance model of traditional simulated annealing algorithm is improved, based on the ideology of non-uniform mutation, to boost the convergence speed. Meanwhile, the time delay signature covered by the stronger reflections is extracted , by using the difference between measured signal and modeled signal, to enhance the recognition capacity of thin-layer media to increase practicability. The experiments testify that the method has high accuracy and stability for inversion of permittivity of layered media.