基于数据挖掘与机器学习技术的低渗储层产量预测

Production prediction of low permeability reservoir basedon data mining and machine learning technology

低渗储层的区域延展性大、油气生产效率低,以及钻井和储层改造作业频繁等特点使得多年在常规油气藏生产作业中所总结出的知识体系与开发方式不能很好地应用于此。借助AI技术快速掌握储层数据信息、提升认知水平、高效制定工程措施,是实现致密储层降本增效开发的新方法。从数据挖掘的角度出发,收集了来自加拿大Cardium致密储层1286口井的50余种影响参数,利用敏感性测试深入探索了该低渗储层历史产量与地理/物性/工程等多维参数的相关性,通过建立和综合对比多种机器学习技术优选出最佳的产量预测模型;为了增加数据利用效用,应用了k-Fold交叉验证方法可以显著改善模型预测精度。结果表明,在所有机器学习算法中随机森林方法表现突出,优化模型可以将产量预测的准确率提高到85%以上。这项研究为海外低渗油气田的储层快速评价、项目综合决策和开发中的降本增效提供了有力的保障。

The characteristics of low permeability reservoir,such as large regional ductility,low oil and gas production efficiency,and frequent drilling and reservoir reconstruction operations,make it impossible to apply the knowledge system and development methods summarized in conventional oil and gas reservoir production operations for many years.Using AI technology to quickly grasp reservoir data information,improve cognition level,and efficiently formulate engineering measures is a new way to achieve cost reduction and efficiency development of tight reservoirs.From the perspective of data mining,more than 50 influencing parameters from 1286 wells in Cardium tight reservoir in Canada were collected.Sensitivity testing was used to deeply explore the correlation between historical production and multi-dimensional parameters such as geography/physical property/engineering,and the best production prediction model was selected by establishing and comprehensively comparing multiple machine learning techniques.In order to increase the utility of data utilization,the k-Fold cross-validation method was applied to significantly improve the prediction accuracy of the model.The results show that the random forest method is outstanding among all machine learning algorithms,and the optimization model can improve the accuracy of yield prediction to more than 85%.This study provides a strong guarantee for rapid reservoir evaluation,comprehensive project decision-making and cost reduction and efficiency increase in overseas low permeability oil and gas fields.