基于U形多层感知机网络的地震波初至拾取与反演  

First-arrival picking and inversion of seismic waveforms based on U-shaped multilayer perceptron network

作　　者：孙明皓 余瀚[1] 陈雨青 陆恺 SUN Minghao;YU Han;CHEN Yuqing;LU Kai(School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing Jiangsu 210046,China;School of Geophysics and Information Technology,China University of Geosciences,Beijing 100029,China;Key Laboratory of Polar Science of Ministry of Natural Resources(Polar Research Institute of China),Shanghai 200136,China)

机构地区：[1]南京邮电大学计算机学院,南京210046 [2]中国地质大学(北京)地球物理与信息技术学院,北京100029 [3]自然资源部极地科学重点实验室(中国极地研究中心),上海200136

出　　处：《计算机应用》2024年第7期2301-2309,共9页journal of Computer Applications

基　　金：国家自然科学基金资助项目(12371440);中国博士后科学基金资助项目(2021M692367);南京邮电大学校级自然科学基金资助项目(NY222140)。

摘　　要：针对传统勘探地震波初至拾取工作量大、抗噪性差和精度低所导致的低质量速度反演影响生产安全的问题,提出一种基于U形多层感知机(U-MLP)网络的地震波初至拾取与反演方法。首先,为解决传统U形网络(U-Net)中的交叉熵损失函数在数据类别不平衡时导致的性能变差问题,设计一种基于加权交叉熵Lovász归一化指数(WLS)的损失函数;然后,在特征融合阶段引入残差连接,缩小低级特征与高级特征间的差距,还原更多细节信息;最后,为使U-MLP网络更好学习图像局部特征,为高级语义引入标记化的多层感知机(MLP)模块,此模块降低了参数量和计算复杂度。实验结果表明,与U-Net相比,U-MLP网络在训练中收敛性更强,初至拾取最大误差降低了20%以上,交并比(IoU)值提升了约2%。可见,U-MLP网络在提取勘探地震波初至时不仅提高了拾取精度,而且拾取的初至在仿真数据和实际数据中的速度分布反演均达到了理想效果,具有更好的性能且适应性更强。A method for first-arrival picking and inversion of seismic waveforms based on U-shaped MultiLayer Perceptron(U-MLP)network was proposed to solve the safety problem in production because of the low quality of inverted velocity,caused by heavy workload,poor noise immunity and low accuracy in traditional first-arrival picking methods.Firstly,a Weighted cross-entropy Lovász-Softmax(WLS)loss function was designed to solve the problem of poor performance of U-shaped Network(U-Net)caused by the traditional cross-entropy loss function in processing unbalanced data categories.Then,residual connections were introduced in the feature fusion stage to reduce the discrepancies between low-level and high-level features,restoring more detailed information.Finally,a MultiLayer Perceptron(MLP)module was introduced for high-level features to learn local image features better,and reduce the number of parameters and computational complexity.Experimental results show that compared with U-Net,U-MLP network converges faster in training,and its maximum error of first-arrival picking decreases by at least 20%while its value of Intersection over Union(IoU)increases by about 2%.It can be seen that the proposed network model not only improves the accuracy of first-arrival picking significantly,but also produces first arrivals for inverting ideal velocity distributions in both the synthetic and the real datasets,hence demonstrating its better performance and stronger adaptability.

关 键 词：U形网络 多层感知机 初至拾取 反演 成像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]