基于神经网络的沉积微相自动识别方法

Neural Network-Based Automatic Recognition Method of Sedimentary Microfacies

沉积微相划分是油田产能建设、储层三维地质建模过程中必不可少的工作之一。面对大规模产建和精细地质研究的需求，庞大的井数据、分小层的沉积微相数据难以用手工完成，采用神经网络技术可以快速识别出各小层的微相类型、在沉积学规律的指导下结合测井曲线的平面分布特征、砂地比、砂体厚度等进行修正，实现沉积相的快速准确识别。油田常用的GPTLog软件中自带神经网络算法，能应用到沉积相分类中，文中多种修正方法对其他区块沉积相研究具有借鉴作用。

The partition of sedimentary microfacies is one of the essential works during the process of oilfield production capacity con- struction and three-dimensional reservoir geological modelling. Facing the requirement of large-scale production capacity construction and fine geological research, gigantic wells data and sedimentary mirco-facies data in small layers are hardly finidhed manually. Sed- imentary microfacies of each small layer can be quickly identified by using the neural network technology. Under the guidance of sedimentology, modified by combining with the plane distribution characteristics of logging curves, sand-strata thickness ratio and sandstone reservoir thickness, the sedimentary microfacies can be identified rapidly and accurately. The neural networks algorithm that is brought with the oilfield often-used GPLog software can be applied to the classification of sedimentary microfacies. The various correction methods can provide a certain reference for sedimentary microfacies research of similar blocks.