基于大规模时间序列的井漏事故预警方法
|
摘要
为了解决人工监测井漏事故在及时性与准确性上的不足,以录井观测日志与自动录井数据为数据支持,通过对工程录井参数的研究,选择并构造合理的数据特征,基于XGBoost算法建立井漏事故预警模型。基于油田真实数据的实验表明该预警方法不仅能够对井漏事故进行准确预警,而且在及时性上比人工监测更为优秀,有助于油田管理人员对井漏事故的防范与应对。
As the manual warning of loss circulation facing the deficiency in timeliness and accuracy,this paper proposes a novel intelligent method of warning. By analyzing the logging parameters of mud logging data and manual logs,constructing effective data features and leveraging XGBoost algorithm as the classifier,an effective warning model of loss circulation is established. Testing over the mud logging data in a real scenario demonstrates that the presented model performs better than traditional manual warning in precision and timeliness.
As the manual warning of loss circulation facing the deficiency in timeliness and accuracy,this paper proposes a novel intelligent method of warning. By analyzing the logging parameters of mud logging data and manual logs,constructing effective data features and leveraging XGBoost algorithm as the classifier,an effective warning model of loss circulation is established. Testing over the mud logging data in a real scenario demonstrates that the presented model performs better than traditional manual warning in precision and timeliness.
引文
[1]蒋希文.钻井事故与复杂问题[M].北京:石油工业出版社,2006.
[2]李凤霞,崔茂荣,王丽华,等.综合录井技术在实时监测钻井事故中的应用[J].断块油气田,2007,14(3):66-68.
[3]杨金华,邱茂鑫,郝宏娜,等.智能化——油气工业发展大趋势[J].石油科技论坛,2016(6):36-42.
[4]朱文鉴,李砚藻.井涌井漏实时预测策略研究[J].国外地质勘探技术,1996(2):29-33.
[5] Friedman J H. Greedy Function Approximation:A Gradient Boosting Machine[J]. Annals of Statistics,2001,29(5):1189-1232.
[6] Chen T,Tong H,Benesty M,et al. xgboost:Extreme Gradient Boosting[J]. 2016,5(9):222-208.
[2]李凤霞,崔茂荣,王丽华,等.综合录井技术在实时监测钻井事故中的应用[J].断块油气田,2007,14(3):66-68.
[3]杨金华,邱茂鑫,郝宏娜,等.智能化——油气工业发展大趋势[J].石油科技论坛,2016(6):36-42.
[4]朱文鉴,李砚藻.井涌井漏实时预测策略研究[J].国外地质勘探技术,1996(2):29-33.
[5] Friedman J H. Greedy Function Approximation:A Gradient Boosting Machine[J]. Annals of Statistics,2001,29(5):1189-1232.
[6] Chen T,Tong H,Benesty M,et al. xgboost:Extreme Gradient Boosting[J]. 2016,5(9):222-208.