基于地震属性的煤层厚度预测模型及其应用

Prediction models of coal thickness based on seismic attributions and their applications

地震属性技术在岩性和构造解释等方面得到了越来越广泛的应用,特别是在煤、油气资源勘探中具有重要的作用.基于淮南矿区谢桥1区131煤层地震勘探资料,提取了28种地震属性数据;通过地震属性的分析,优选出平均峰值振幅、振幅的峰态、最大绝对振幅、瞬时频率斜率等4种地震属性作为煤层厚度预测模型基本参数,结合已知钻孔资料,利用多元多项式回归以及BP人工神经网络方法,求出了各属性与煤厚之间的多元多项式回归模型及人工神经网络模型,并对模型进行了误差分析和应用结果对比分析,反映出人工神经网络模型在煤厚预测中具有好的应用效果.

As a technology used widely in lithology and structural interpretation, seismic attribute technology has been playing an important role in coal and oil exploration. Based on 3-D seismic exploration data of coal seam 13-1 in the Xieqiao colliery, Huainan coal field of China, 28 seismic attributes are extracted. Through analysis of seismic attributes, four usable seismic attributes, such as average-peak-amplitude, kurtosis-in- amplitude, maximum-abs01ute-amplitude and slope rate of instantaneous-frequency, are selected as the basic analysis parameters of prediction models of coal thickness. Combined with the real drill data, the prediction models between coal thickness and multi attributes are established by using analytical methods of multivariant polynomial regression and BP neural networks（ BPNN）, and error analysis of predicting coal thickness is carried out. From the comparison of prediction results of coal thickness of coal seam 13-1 in the Xieqiao colliery, Huainan coal field of China by using these two models, it is concluded that the BPNN model has higher accuracy in predicting coal thickness.