石油地球物理勘探2026，Vol.61Issue(1)：86-97,12.DOI:10.13810/j.cnki.issn.1000-7210.20240511
阈值约束的加权K均值聚类自动速度拾取方法
Automatic velocity picking with threshold-constrained weighted K-means clustering
摘要
Abstract
In conventional K-means clustering,the number of clusters and the initial values need to be predeter-mined,and the picking result is the geometric center of the energy cluster and is greatly influenced by the initial values.This paper proposes an automatic velocity picking method based on weighted K-means clustering with threshold constraints.Multiple rectangles of appropriate length are obtained by applying a variable velocity point threshold.The number of clustering centers,initial time,and initial velocity are obtained with the rectangles and prior velocity.At the same time,the prior velocity is used to limit the velocity picking range,and then the con-stant threshold and adaptive threshold are used to eliminate the velocity points with small amplitudes,reduce the number of velocity points involved in the calculation,and improve calculation efficiency.The weighted K-means clustering algorithm uses the amplitude of velocity points to calculate the weights,and removes the points far away from the center step by step through the distance threshold,so that the cluster center overlaps with the energy cluster center.Finally,multiples are eliminated by comparing with the slope of the prior velocity to make the picking result more accurate.The processing of model and actual data shows that the method proposed in this paper can intelligently pick up seismic velocity under the premise of ensuring accuracy and has high effi-ciency.关键词
速度拾取/聚类中心/加权K均值聚类/先验速度/无监督Key words
velocity picking/cluster center/weighted K-means clustering/prior velocity/unsupervised分类
天文与地球科学引用本文复制引用
谢俊法,刘文卿,盛萍,吴杰,伍敦仕,黄紫晨..阈值约束的加权K均值聚类自动速度拾取方法[J].石油地球物理勘探,2026,61(1):86-97,12.基金项目
本项研究受中国石油集团公司关键核心技术攻关项目"三维VSP井地联合成像关键技术与软件研发"(2025ZG53)资助. （2025ZG53）
