基于过程神经网络的储层微观孔隙结构类型预测

摘要

针对储层微观孔隙结构识别常用的室内实验法价格昂贵和研究成果具有区域性的问题，提出一种利用测井曲线预测储层微观孔隙结构类型的方法。首先，利用过程神经元建立双层过程神经网络模型；其次，结合文化算法和混合蛙跳算法的优势训练网络模型；最后优选8条测井曲线作为模型输入来预测孔隙结构类型。实验仿真结果表明所提方法具有很好的识别效果。%Aiming at the expensive laboratary experiment method for the reservoir’s micro-pore structure and the shortcomings of regional research results,making use of well logging curve to predict the reservoir micro-pore structure types was proposed,which has process neural network adopted to establish a double-layer process neural network model,and then has the cultural algorithm and shuffled frog leaping algorithm com-bined to train the network model and as well as the eight appropriate logging curves selected as the model input to predict the pore structure type.Simulation results show that this proposed method has good recognition effect.