简要介绍了地面核磁共振找水的正演理论方法,采用高斯求积并结合连分式展开的方法对核磁共振核函数中包含双重贝塞尔函数的积分核进行了数值积分,进而计算出精度较高的核函数值,在此基础之上,讨论了导电性对地面核磁共振信号的影响.基于奇异值分解算法,对核磁共振找水理论模型进行了反演研究,在不导电的情况下,重构出了比较理想的含水率分布.总结并改进了模拟退火算法,用改进的模拟退火算法分别对覆盖层高阻和导电条件下的核磁共振人工合成数据进行了反演,试验结果表明,利用改进的模拟退火算法反演地面核磁共振是可行的,而且反演结果较稳定,收敛速度较快.
The paper simply introduced the basic principle of surface nuclear magnetic resonance and computed the numerical integral of nuclear magnetic resonance kernel function which contained double Bessel functions using the method of Gauss quadrature and continued fraction expansion, and then discussed influence of conductivity to nuclear magnetic resonance′s signal. Based on singular value decomposition algorithm, we studied the theoretical model of surface nuclear magnetic resonance and gained the ideal inversion result. We also made inversion experimentation respectively on insulative and electric model using the improved simulated annealing algorithm, and showed that it was desirable to this method. The inversion result was relatively stable and the convergence was very fast.