基于Logistic统计的模糊判别法在储层流体识别中的应用

Application of Fuzzy Discrimination Method Based on Logistic Statistics in Reservoir Fluid Identification

在测井储层评价中准确识别储层流体性质是其关键。但当储层测井响应特征并不特别明显时,只采用常规测井识别方法不易准确区分气、水层。为此,综合常规测井和试油资料,选取反映气、水层特点的特征参数建立划分气层、气水同层及水层的模糊综合判别因素,利用Logistic统计法确定隶属函数曲线,即以Logistic函数为激活函数,利用模糊统计法确定隶属函数曲线;然后结合权重集建立了模糊综合判别模型对流体的性质进行模糊判别,经检验该方法能较准确地判别出水层、气水同层和气层。

Accurate identification of reservoir fluid properties is the key in Logging Reservoir Evaluation, but it cannot accurately identify the nature of the fluid only using conventional logging methods when log response characteristics of gas and water layers are not obvious. Therefore, this study selected three po- rosity and deep resistivity logs that reflect gas and water layer characteristics based on conventional logging data to establish Fisher discriminant function used to classify gas layer, gas-water layer and water layer. The study also utilized logistic statistical method to determine the membership function curves, which means using logistic function as activation function to determine the membership function curves by fuzzy statistics. Then, a fuzzy discriminant model was established in combination of the weight set to distinguish the nature of fluid in a fuzzy manner. Test shows that this method method can accurately determine the water, gas and water layer and gas reservoirs.