基于自适应遗传算法的MRS-TEM联合反演方法研究

地面磁共振法(MRS)因具有定性、定量分析地下水能力,而备受关注.传统磁共振地层含水量反演多采用均匀半空间模型,忽略电阻率分布信息对结果的影响.针对这一问题,本文基于多层电介质中磁共振响应理论,提出MRS与瞬变电磁(TEM)联合反演方法,通过电阻率分布信息对含水量反演过程的实时修正,提高了解释结果的准确度.反演算法采用自适应遗传算法(AGA)进行,基于繁殖规则,动态调整交叉概率和变异概率,解决了标准遗传算法易未成熟收敛而难以得到全局最优解问题.模型数据表明,含噪10%情况下,联合反演仍能较准确地反映地下含水单元模型结构,对比MRS单独反演优势明显.同时,内蒙古白旗野外观测数据联合反演结果与钻井资料基本一致,充分验证了AGA反演算法的实用性及MRS-TEM联合反演的实际意义.

Joint inversion of MRS and TEM data based on adaptive genetic algorithm

The method of Magnetic Resonance Sounding (MRS) provides a new technology to determine subsurface water qualitatively and quantitatively, which has received much concern of geophysicists. The traditional inversion of MRS uses homogeneous half-space models and ignores effects of resistivity distribution on inversion results. Based on the MRS theory in the multi-layer earth model, we developed a new scheme for joint inversion of MRS data and Transient Electromagnetic (TEM) data. By correcting MRS inversion with resistivity in real time, the accuracy of inversion result was improved greatly. The Adaptive Genetic Algorithm (AGA) was applied. By adaptively adjusting the probability of crossover and mutation, AGA has not only avoided the premature convergence of GA but also has obtained the global optimal solution. Studies with synthetic data show that the MRS-TEM joint inversion scheme can obtain the aquifer structure accurately even 10% noise applied. Compared with the traditional MRS inversion scheme, the advantage of this new approach is obvious. By applying this method to filed data processing, and with the good inversion result, we have verified the the applicability of AGA and the practical significance of MRS-TEM joint inversion scheme.