概率神经网络技术在地震岩性反演中应用

针对研究区目的储层单层厚度小和储层岩性横向变化较大情况,为了解决利用地震资料进行常规储层预测反演较困难问题,采用概率神经网络方法,讨论了网络模型的构造和预测识别等步骤。根据该区储层的测井响应特征、地震属性特征与地质岩性特征的相关性,利用概率神经网络方法对地层特征进行预测识别,研究结果表明:在实际资料处理中取得了好的应用效果。研究成果对油田勘探开发具有一定的指导意义。

Application of probabilistic neural network technique in lithology inversion of seismic data

In view of the fact that the target reservoir thickness is small and the lateral change of lithology is big, a specific method of probabilistic neural network is used so as to overcome the difficulty of lithology inversion using seismic data in conventional way.The construction of the network model and the steps of prediction are discussed.The probabilistic neural network method is used based on the correlation among the features of logging, seismic attributes and geological lithology characteristics,and it has a good effect in processing the actual data to identify the reservoir lithologic information.The result is has certain guiding significance to the petroleum exploration and development in the survey area.