基于LSTM深度神经网络的TGS地震数据去噪重构在滇中引水工程中的应用  

Denoising Reconstruction of TGS Seismic Data Based on LSTM Deep Neural Network:A Case Study from the Central Yunnan Water Diversion Project

作　　者：石贵飞 王俊 刘杰 刘宇锋 姚富宝 肖长安 SHI Gui-fei;WANG Jun;LIU Jie;LIU Yu-feng;YAO Fu-bao;XIAO Chang-an(PowerChina Kunming Engineering Corporation Limited,Kunming 650032,China)

机构地区：[1]中国电建集团昆明勘测设计研究院有限公司,云南昆明650032

出　　处：《水电能源科学》2025年第3期142-146,100,共6页Water Resources and Power

摘　　要：提高地震原始记录的信噪比对隧洞超前地质预报至关重要,滇中引水工程隧洞中常用TGS法开展超前地质预报,针对隧道TGS地震数据采集时容易受到大量干扰信号的影响进而产生较大的噪音。为了提高地震数据的质量,首先模拟TGS采集模式,并结合滇中引水工程隧洞实际设计资料有针对性地建立多个地质模型,通过正演模拟得到大量高质量的隧洞地震数据,并加入噪声;其次引入LSTM深度学习网络寻找加噪地震数据与原始地震数据之间的复杂非线性关系;最后将在滇中引水项目隧洞中采集到的真实数据输入到训练好的LSTM网络进行去噪重构。实际应用效果表明,采用基于LSTM深度神经网络的去噪重构方法可有效提升地震数据的质量,并在一定程度上提升在滇中引水工程开展超前地质预报的预测效果。为地震数据去噪领域的研究提供了一种新思路。The enhancement of the signal-to-noise ratio of seismic primary records is crucial for the advanced geological forecasting of tunnels.The TGS method is commonly used to carry out advanced geological prediction in the tunnel of the Central Yunnan Water Diversion Project.Due to the influence of a large number of interference signals during the TGS seismic data acquisition in tunnels,a considerable amount of noise is generated.In order to improve the quality of seismic data,the simulation of the TGS data acquisition mode is first performed.Several geological models are established according to the actual design data of the tunnel of the Central Yunnan Water Diversion Project.A large number of high quality seismic data of tunnel are obtained through forward modeling,and noise is then added.Secondly,an LSTM deep learning network is introduced to find the complex nonlinear relationship between the noisy seismic data and the original seismic data.Finally,the real data collected in the tunnel of the Central Yunnan Water Diversion Project is input into the trained LSTM network for noise removal and reconstruction.The results show that the noise reduction and reconstruction method based on LSTM deep neural network can effectively improve the quality of seismic data and to some extent improve the prediction effect of the geological forecast in the Central Yunnan Water Diversion Project.Thus,it provides new ideas for research on noise reduction of seismic data.

关 键 词：滇中引水工程 超前地质预报 地震数据 去噪重构:LSTM 

分 类 号：TV672.1[水利工程—水利水电工程]