论文部分内容阅读
航空电磁探测数据量大,二维、三维反演算法复杂、计算速度慢,通常采用一维反演,利用层状模型拼接描述地下复杂结构,但航空电磁数据信噪比低,容易引起一维反演结果横向连续性差等问题.本文针对上述问题,基于一维反演算法,通过整合测线观测数据,建立了测线数据整体的目标函数,并根据Tikhonov正则化反演理论,引入包含空间粗糙度和先验信息的模型参数约束项,确定了拟二维整体反演的目标函数,推导了反演迭代方程组,利用超松弛共轭梯度算法,求得由于整条测线整体反演所致的大型稀疏矩阵的极小化解,实现了对整条测线数据同时反演的固定翼航空电磁数据的拟二维整体反演算法.在反演迭代过程中,正则化因子采用线性搜索自适应迭代的方法自动选取,提高了反演结果的稳定性.对比分析了仿真数据的一维反演与拟二维整体反演结果,得出拟二维整体反演算法横向连续性较好,对高导覆盖层下的导体分辨率优于一维反演,同时受高斯噪声的影响较小.最后,将直升机飞行实测噪声加入仿真数据中,拟二维整体反演结果平均相对误差较一维反演结果降低了31.6%,进一步验证了拟二维整体反演算法的有效性.
Aeronautical electromagnetic detection of large amount of data, two-dimensional, three-dimensional inversion algorithm is complex, the calculation speed is slow, usually using one-dimensional inversion, the use of layered model stitching to describe the underground complex structure, but the electromagnetic signal electromagnetic aeromagnetic data is low, In this paper, based on the one-dimensional inversion algorithm, the objective function of the survey data is established based on the one-dimensional inversion algorithm by integrating the survey observation data. Based on the Tikhonov regularization inversion theory, Roughness and a priori information model parameter constraints to determine the objective function of the quasi two-dimensional global inversion, derived iterative equations, the use of over-relaxation conjugate gradient algorithm, obtained as a whole line inversion of the entire survey , The quasi-two-dimensional global inversion algorithm of fixed-wing aeronautic electromagnetic data for the simultaneous inversion of the entire survey data is realized. In the iterative process of inversion, the regularization factor is linearly searched The method of adaptive iteration is automatically selected to improve the stability of the inversion results.Compared with the one-dimensional inversion and the quasi-two-dimensional global inversion of simulation data, The algorithm has good lateral continuity and better resolution than the one-dimensional inversion of the conductor under the high-conductivity overburden, and is less affected by the Gaussian noise.Finally, the measured helicopter flight noise is added to the simulation data, The average relative error of the results was reduced by 31.6% compared with the one-dimensional inversion results, which further validated the validity of the quasi-two-dimensional global inversion algorithm.