地震属性融合技术在煤层厚度预测中的研究

Seismic attributes fusion and its research in predicting thickness of coal

煤层与岩层的顶、底板间形成的地震反射强度的变化将引起煤层厚度预测结果的变化.首先提取了目的层的反射波的地震属性,然后深入研究、分析了这些属性特性,最后应用BP人工神经网络方法以及多项式回归分析的方法进行研究,并在实际地震资料中应用这些方法,结果表明:在对矿区的煤厚预测中,BP人工神经网络模型误差最小,多元二次回归次之,多元一次线性回归模型误差最大,证明了用多属联合分析技术进行煤厚预测是一种卓有成效的方法.

The change of seismic reflection intensity formatted between the coal and the strata floor causes a change in the thickness of the coal seam. According to the characteristics of the coal seam,select the relevant seismic attributes,study and analysis the characteristics of them. Using polynomial regression analysis and BP artificial neural network to forecasting coal seam thickness and apply them to practical seismic data,the results show that： in the thickness prediction of mining area,BP artificial Neural network model causes the smallest error,followed by multivariate quadratic regression,multivariate linear regression model causes the largest error. And these proved that using the multi-attribute comprehensive analysis to predict the thickness of coal is an effective method.