基于布谷鸟搜索的地震属性聚类方法及其在塔中地区碳酸盐岩礁滩储层的应用

Seismic attribute clustering based on Cuckoo Search and its application in carbonate reef-shoal reservoirs in Middle Tarim Basin

地震属性聚类是提取隐藏在地震数据中地质特征的重要途径,K均值是最常见的聚类方法之一,方法简单且高效,但是该算法存在局部收敛、结果依赖于初值等问题。为了解决该问题,将具有全局寻优能力和更高搜索效率的布谷鸟搜索算法引入到地震属性聚类中,通过扩大搜索范围,增加种群数量,更容易跳出局部极值。结果表明,通过2个理论数据集试验证明基于布谷鸟搜索的聚类算法能较好地发现非线性数据结构中低维特征。通过实际地震数据应用可以看出基于布谷鸟搜索的地震属性聚类算法能比较准确地刻画塔里木盆地塔中地区碳酸盐岩礁滩储层和油气的分布。

Seismic attribute clustering is an important approach to extract geological features from seismic data.Kmeans,one of the most popular clustering method,is very simple and efficient,but has problems of local convergence and its results highly depending on initial value.In order to solve this problem,Cuckoo Search algorithm characterized by global optimizing capability and higher search efficiency is introduced into seismic attribute clustering.Through widening search scope and increasing population quantity,it is easy to jump out local extremum.Proved by 2 theoretical data set tests,the clustering method based on Cuckoo Search can better find middle and low dimensional features from non-linear data structure.Actual application of seismic data shows that seismic attribute clustering method based on Cuckoo Search can correctly characterize distributions of carbonate reef-shoal reservoir and hydrocarbon in Middle Tarim Basin.