【摘要】 本文利用地震多属性对地震数据进行分析,并基于概率神经网络的学习方法,对测井数据与地震多属性进行训练,从而利用测井信息和井间地震资料反演出全区密度等储层参数的分布图,为煤层气富集区预测提供依据。通过寺河矿区实例分析,该方法与常规反演方法相比具有更多的优点,它不但克服了常规反演只能计算声波、密度、波阻抗的限制,可以反演电阻率、孔隙度等储层参数,还融入了地震道的多种地震属性,信息更为全面。更多还原

【Abstract】 This article analyzes seismic data using seismic multi-attribute and trains the logging data and seismic multi-attribute basing on probabilistic neural network.The distribution map of reservoir parameter such as density is gained after inversion by log data and seismic data to provide related basis for prediction of CBM-rich area.Through the analysis of Sihe mining area,it is found that this method has more advantages than other conventional inversion which can only calculates acoustic wave,density and wave impedance.More reservoir parameter can be calculated,such as resistivity and porosity with the new method.In addition,the data volume is more comprehensive by integrating seismic multi-attribute into the inversion.更多还原