薄层厚度预测一直是公认的难题之一,其难度就在于如何准确地识别和提取薄层的地震属性.常规方法是利用时间域或频率域地震属性与薄层厚度的线性关系计算.但是理论与实际资料表明,不同的薄层和地层组合对地震波的动力学的信息影响很大,各种参数与薄层厚度成非线性关系,使用单一的信息不可能准确预测薄层厚度.本文利用三种线性预测原理(模型),经数学变换为属性参数,采用非线性BP网络预测薄层厚度,取得了令人满意的效果.
The prediction of thin bed thickness is one of the difficult problems generally Acknowledged. Its difficulty lies in how to exactly extract the seismic attributes of thin bed. The traditional method is to use the linear relations between seismic attributes parameter of time or frequency and the thin bed thickness. But both theoretical and practical data indicate that different thin bed and strata combinations have a great influence on the dynamic information of seismic wave. All parameters depend nonlinearly on the thin bed thickness. The thin bed thickness can not be predicted exactly by using single seismic information. This essay converts the three linearity prediction theories into attributive parameters through mathematic ways. we adopt the several parameters, and use the neural network to predict the thin bed thickness. And with application of theoretical model and practical data,it is high-precision to predict thin bed thickness, we achieve a satisfactory result.