基于极端梯度提升算法的地震同相轴自动识别

Automatic detection of seismic event based on eXtreme gradient boosting

在常规地震同相轴识别方法基础上,通过引入极端梯度提升算法(XGBoost)智能化策略,并结合地震数据相邻道相似性特征,发展一种基于极端梯度提升算法的地震同相轴自动识别技术方法。在编程实现方法的基础上,通过简单层状模型和复杂Marmousi模型模拟的记录进行测试,验证方法的正确性。对含噪音数据和实际资料中的同相轴进行识别测试,同时进行单道对比定量分析以及不同信噪比情况下算法预测结果精度对比。结果表明:新方法对含噪数据和实际资料均具有较好的适应性;在低信噪比(-6.98 dB)情况下,同相轴的查准率仍可超过90%。

This paper presents an automatic seismic event detection method based on eXtreme gradient boosting(XGBoost) by integrating intelligent strategies and leveraging the similarity characteristics of adjacent traces in seismic data.The proposed method is validated through programming and testing on both simple layered and complex Marmousi models.The detection tests conducted on noisy data and the real data demonstrate the method's robustness and adaptability,even in low signal-to-noise ratio(SNR)conditions(-6.98 dB),achieving a seismic event detection precision of 90%.Additionally,single channel contrast quantitative analysis and comparison of algorithm prediction accuracy under various SNR conditions further confirm the method's feasibility and applicability.