| 基于萤火虫和蝙蝠群智能算法的瑞雷波频散曲线反演 |
| |
| 引用本文： | 蔡伟,宋先海,袁士川,胡莹.基于萤火虫和蝙蝠群智能算法的瑞雷波频散曲线反演[J].地球物理学报,2018,61(6):2409-2420. |
| |
| 作者姓名： | 蔡伟 宋先海 袁士川 胡莹 |
| |
| 作者单位： | 1. 中国地质大学地球物理与空间信息学院, 武汉 430074;2. 中国地质大学(武汉)湖北省地球内部多尺度成像重点实验室, 武汉 430074 |
| |
| 基金项目： | 国家自然科学基金（41574114，41174113）资助. |
| |
| 摘 要： | 反演瑞雷波频散曲线能有效获取地层横波速度和厚度.但由于其高度的非线性、多参数、多极值等特点，传统的全局搜索方法易出现收敛速度慢、早熟收敛及搜索精度低的问题.鉴于此，本文提出并测试了基于萤火虫优化算法（FA）和带惯性权重的蝙蝠优化算法（WBA）的新的瑞雷波频散曲线反演策略.在瑞雷波频散曲线反演中，FA全局搜索能力强，但后期搜索精度低，而WBA局部搜索能力强，搜索精度高，但易出现早熟收敛.故本文将二者结合，提出了一种新的优化策略，称其为WFBA，即在反演前期使用FA，后期使用WBA，很好地解决了FA后期搜索精度低及WBA早熟收敛的问题.本文首先反演了三个典型理论模型的无噪声、含噪声的数据，验证了WFBA对瑞雷波数据反演的有效性与稳定性.然后将WFBA与WBA、FA单独反演以及不含惯性权重的FBA和粒子群优化算法（PSO）反演的结果进行了对比，说明了WFBA相对于WBA、FA、FBA和PSO具有更稳定、收敛速度更快、求解精度更高等优点.最后，反演了来自美国怀俄明地区的实测资料，检验了WFBA对瑞雷波数据反演的实用性.理论模型试算和实测资料分析表明，WFBA很适用于瑞雷波频散曲线的定量解释，具有很高的实用性价值.
|
| 关 键 词： | 瑞雷波 频散曲线 萤火虫优化算法 蝙蝠优化算法 粒子群优化算法 |
| 收稿时间： | 2017-05-27 |
Inversion of Rayleigh wave dispersion curves based on firefly and bat algorithms |
| |
| CAI Wei,SONG XianHai,YUAN ShiChuan,HU Ying.Inversion of Rayleigh wave dispersion curves based on firefly and bat algorithms[J].Chinese Journal of Geophysics,2018,61(6):2409-2420. |
| |
| Authors: | CAI Wei SONG XianHai YUAN ShiChuan HU Ying |
| |
| Institution: | 1. Institute of Geophysics and Geomatics, China University of Geosciences, Wuhan 430074, China;2. Hubei Subsurface Multi-scale Imaging Lab(SMIL), China University of Geosciences, Wuhan 430074, China |
| |
| Abstract: | Inversion of Rayleigh wave dispersion curves can effectively obtain shear wave velocity and thickness of formation. However, due to its high degree of non-linearity, multi-parameter, multi-extremes and so on, the traditional global search method is prone to slow convergence, premature convergence and low precision. To solve this problem, this paper presents and tests a new inversion strategy for Rayleigh wave dispersion curves based on the firefly optimization algorithm (FA) and bat optimization algorithm with inertia weight (WBA). In the inversion of Rayleigh wave dispersion curves, the global search ability of FA is strong, but the search precision is low, while the local search ability of WBA is strong, and the search precision is high, but it is prone to premature convergence. Therefore, this paper combines the two, and proposes a new optimization strategy, named WFBA. That is, using the FA in the early stage, and the WBA in the late stage, a good solution to the low search accuracy of FA and premature convergence problem of WBA can be achieved. In this approach, we first invert the noise-free and noise-contaminated data of three typical theoretical models, and verify the validity and stability of WFBA. Then, the inversion results of WFBA, WBA, FA, particle swarm optimization (PSO) and FBA without inertia weight are compared. It shows that WFBA has higher stability, faster convergence, and better accuracy. Finally, the data from the Wyoming area in the United States are inverted, and the applicability of WFBA to Rayleigh wave data inversion is examined. Calculations on the theoretical model and measured data show that WFBA is suitable for the quantitative interpretation of Rayleigh wave dispersion curves, demonstrating a high practical value. |
| |
| Keywords: | Rayleigh wave Dispersion curves Firefly algorithm Bat algorithm Particle swarm optimization |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《地球物理学报》浏览原始摘要信息 |
| 点击此处可从《地球物理学报》下载免费的PDF全文 |
|
