基于粗糙集与遗传算法的储层识别技术

Reservoir recognition technology based on rough set theory and genetic algorithm

储层的含油气性识别是储层综合评价的难点和关键,文中以粗糙集理论为基础,利用布尔逻辑和粗糙集理论相结合的离散化算法对每个条件属性进行离散化处理;利用基于遗传算法的粗糙集理论提取具有一定决策概率的不精确判别规则;利用规则的支持度、置信度和覆盖度挑选有效规则,进行储层类型的识别。实例应用结果表明,该方法提高了储层识别的正确率,提取的判别规则具有可解释性且较易理解,能够有效挖掘勘探数据中的潜在关键信息,对储层开发方案的制定具有重要的指导意义。

The reservoir recognition for oil and gas is a difficulty and key in reservoir comprehensive evaluation. Based on the rough set theory., each condition attribute is discret with discretization mothod combining Boolean logic with rough set theory in this paper. The inaccurate recognition rules with a certain decision probability are extracted with the rough set theory on the basis of genetic algorithm. And the effective rules are selected to identify the type of reservoir according to the support, confidence and coverage. The example results show that this method improves the accuracy of reservoir recognition. The discriminant rules which have been chosen have good interpretability and comprehensibility. The method can effectively mining the potential key information from exploration data and has important guiding significance for making reservoir development scheme.