储层参数自组织分类模型研究  

On Self-organizing Classification Model of Reservoir Parameters

作　　者：杜环虹 屈乐[1,3] 李新[1,2,3] 

机构地区：[1]中国石油集团测井有限公司技术中心,陕西西安710077 [2]中国石油测井重点实验室,陕西西安710077 [3]西北大学地质学系,陕西西安710069

出　　处：《测井技术》2011年第B12期649-651,共3页Well Logging Technology

摘　　要：应用人工神经网络技术，实现数据的自组织分类，针对不同的类，建立相应的解释模型，实际计算时，采用统计模式识别方法选择解释模型。以渗透率为例，比较了经验公式、人工神经网络方法和人工神经网络自组织分类方法三者的计算结果，结果表明，由经验模型到神经网络单一混合模型再到神经网络自组织分类模型，其平均相关系数逐步提高，平均绝对误差、平均相对误差依次降低。据此，自组织分类方法具有更大的优越性。Artificial neural network method is applied to realize self-organizing classification of data and establish the appropriate interpretation model for different classes. In actual calculation, statistic pattern identification method is applied to select interpretation model. Compared are the permeability calculation results of experiential formula, artificial neural network method and self- organization classification method. The above results show that, from experiential model to sin- gle neural network model to neural network self-organization classification model, the average correlation coefficient gradually increases; the average absolute error and average relative error reduce in order. Therefore, self-organization classification method has greater advantages.

关 键 词：测井解释 储层参数 人工神经网络 自组织 分类 模型建立 

分 类 号：P631.84[天文地球—地质矿产勘探]