人工智能地震弱信号处理技术数据

基于压缩感知的盐下超深层微弱信号重建与增强：研究压缩感知框架下的弱信号表征，构建基于重建误差稀疏约束的最优化问题；引入伪地震数据迭代求解稀疏优化问题实现弱信号重建与增强；探究稀疏域弱信号与噪声分布特征，基于深层模型构建弱信号与噪声数据标签，优选深度神经网络智能区分信号与噪声，构建智能去噪算法流程；引入深度学习对信号频谱特征进行提取和分割；基于地震、测井等多数据构建深层地震弱信号样本，开展拓频深度学习训练，构建频谱自适应动态拓宽方法，有效提升地震资料分辨率Reconstruction and Enhancement of Ultra-deep Subsalt Weak Signals Based on Compressed Sensing: This research focuses on the characterization of weak signals within the compressed sensing framework, and constructs an optimization problem with sparse constraints based on reconstruction errors. Pseudo-seismic data is introduced to iteratively solve the sparse optimization problem to realize weak signal reconstruction and enhancement. The distribution characteristics of weak signals and noise in the sparse domain are explored; labels of weak signal and noise datasets are generated based on deep models, optimal deep neural networks are selected to intelligently distinguish signals from noise, and an intelligent denoising algorithm workflow is established. Deep learning is adopted to extract and segment the spectral features of signals. Deep seismic weak signal samples are constructed using multi-source data including seismic and logging data, frequency-broadening deep learning training is carried out, a spectrum-adaptive dynamic broadening method is developed, which effectively improves the resolution of seismic data.

搜集汇总

背景与挑战

背景概述

该数据集聚焦于人工智能在地震弱信号处理中的应用，涵盖基于压缩感知的盐下超深层微弱信号重建与增强、智能去噪算法以及频谱自适应动态拓宽方法。数据量达9.1GB，包含110个文件，由中国石油大学（华东）在塔里木盆地盐下超深层油气田研究项目中创建。