摘要:
面波多道分析方法(MASW)是获取垂向剪切波速度剖面的一种有效方法。频散曲线反演是MASW中关键的一步。由于瑞雷波频散曲线反演具有非线性、多参数和多极值的特征,这对于常规的局部线性化反演方法是极大的挑战。为此,本文采取确定性的全局优化算法,广义模式识别算法(GPS)对瑞雷波频散曲线进行反演。其原理可以简述为:算法首先通过模式以确定性的方式对目标函数进行采样来搜索一个点序列;然后使序列中每一个点到下一个点的目标函数值逐渐减少,从而使点序列逐渐逼近全局最优解,最后的解便为待求的最优模型参数。为验证GPS的有效性,首先利用设计的3种典型的6层地质模型通过快速矢量传递算法正演模拟产生基模式频散曲线(频率范围为5~101Hz,频率间隔为2Hz,频点数为49),并对理论频散曲线进行反演。反演结果表明,模型的真实值已经被高度精确地重建。说明GPS可以用于实际勘探中的基模式频散曲线反演。为进一步验证GPS的有效性,在吉林大学校园采集瑞雷波实测数据,并提取基模式频散曲线,应用GPS进行反演。反演重建的横波速度剖面与先验的地质信息吻合得很好。理论模型和真实数据的反演结果表明,GPS可以应用在瑞雷波频散曲线非线性反演中。
Abstract:
Multichannel Analysis of Surface Waves (MASV0 is an efficient method to obtain the vertical shear-wave profile. Inversion of dispersion curve is a key step of Multichannel Analysis of Surface Waves(MASW).Inversion of Rayleigh-wave dispersion curves is challenging for most local-search methods due to its high nonlinearity, its multiparameters and its multimodality. In this paper, we impIement a deterministic global optimization algorithm named generalized pattern search (GPS). The algorithm is adopted to inverse Rayleigh-wave dispersion curve. The basic principle can be summarized as follows: the objective function is sampled by the algorithm through a pattern in a deterministic way in order to search a point sequence;then the objective function value of each point in the sequence to the next point should be reduced, so that the point sequence gradually approximates the global optimal solution, and the final solution is the optimal model parameter. First of all, the base mode dispersion curve (frequency range is 5-101 Hz, frequency interval is 2 Hz, the number of frequency point is 49) of three typical six-layer geological model is forwarded by the fast vector-transfer algorithm. And the theoretical dispersion curve is inversed. The inversion results show that the true value of the model has been reconstructed with high precision. It is shown that the GPS algorithm can be used for the inversion of the basic mode dispersion curves in actuat exploration. In this paper, we gather the real data of Rayleigh-wave in Jilin University campus and base mode dispersion curve is extracted. The inversion is proceeded in order to verify the effectiveness of GPS algorithms, furtherly. The reconstructed S-wave velocity profile through inversion is in good agreement with prior geological information. The inversion results of theoretical model and real data show that GPS algorithm can be applied to nonlinear inversion of Rayleigh-wave dispersion curves.
百度学术
下载:
