摘要
运用软计算融合算法识别储层含油性的关键属性,建立了预测这些关键属性的软计算融合模型。具体步骤为:首先采用遗传算法(GA)和模糊C均值嵌套算法(GA-FCM)对含油性的测井属性进行约简,得到能够描述含油性的关键属性;然后再把GA和BP神经网络(GA-BP)进行融合构建预测关键属性的软计算融合模型,即通过GA优化BP的结构(包括网络输入属性的组合和最佳隐含层神经元个数的确定),并且用测试样本的误差作为评判该预测模型的优劣;最后对某油田的oilsk81,oilsk83,oilsk85 3口井进行了实证研究。
The paper proposes the fusion methods of soft computing which can recognize the core attributes of oil-bearing formation and sets up the forecasting model. The steps as follows: Firstly, obtaining the core attributes that can represent the oil-bearing formation by attributes of reduction using GA-FCM. Secondly, setting up the forecasting model to forecast the core attributes. The forecasting model is BP neural network(BPNN) and the optimal inputs of BPNN and the optimal number nodes of hidden layer are obtained by GA which is evaluated by the test data of the sum of error square. At last, make empirical research of oilskS1, oilsk83, oilsk85 well in some oil field in China.
出处
《系统管理学报》
北大核心
2008年第1期87-93,共7页
Journal of Systems & Management
基金
国家自然科学基金资助项目(70573101)
高等学校博士学科点专项科研基金资助项目(20070491011)
关键词
软计算
测井属性
GA—FCM
优化
soft computing
well log attributes
genetic algorithm & fuzzy C-means
optimizing