带有专家部分知识的模糊学习算法及在储层识别中的应用

A Fuzzy Learning Algorithm with Expert's Knowledge and Its Application to Reservoir Identification

已有的示例学习算法大多没有处理分类模糊的情况。针对辽河油田冷东地区关于储层评价出现的模糊性，提出了一种模糊学习模型。由于评价过程中人为经验占有主要成分，故这种学习模型中融入了部分专家经验，将现有的模糊分类问题转化为一个求类比值的数学规化问题，它以极大化类信息熵为目标函数。以分类的模糊性极小作为决策树学习算法的启发式，学习结果可转化为一组带可信度的模糊产生式规则。这种产生式规则的使用为该地区新解释标准的建立以及油层的综合评价提供了更直观、更可靠的依据。

The author suggests a fuzzy learning model coping with the fuzziness of classification for identifying reservoirs in Liaohe Oilfield .Sine humans experiences play a key role in the process of identifying reseruoirs this algorithm incorporates the expert's knowledge.The fuzzy classification is converted into a programming problem with maximum information entropy.The minimum clssification fuzziness is selected as the heuristic and the learnig results are converted to some fuzzy if-then rules with some degree of truth.These fuzzy if-then rules provide more intuitive and more reliable baisi for establishing new explanation standard and for overall identifying reservoir in that area.