深水井气侵监测预警智能技术研究  

Research on intelligent technology for monitoring and early warning of gas invasion in deep-water wells

作　　者：刘书杰 罗鸣 李文拓 魏安超 胡益涛 于晓东 徐一龙 杨耀明 LIU Shujie;LUO Ming;LI Wentuo;WEI Anchao;HU Yitao;YU Xiaodong;XU Yilong;YANG Yaoming(Hainan Branch,CNOOC(China)Co.,Ltd.,Haikou 570311,China;Zhanjiang Branch,China France Bohai Geoservices Co.,Ltd.,Zhanjiang 524033,China;Hainan Energy Co.,Ltd.,CNOOC(China)Co.,Ltd.,Haikou 570311,China;Beijing Aerospace Huateng Technology Co.,Ltd.,Beijing 100176,China)

机构地区：[1]中海石油(中国)有限公司海南分公司,海南海口570311 [2]中法渤海地质服务有限公司湛江分公司,广东湛江524033 [3]中海油海南能源有限公司,海南海口570311 [4]北京航天华腾科技有限公司,北京100176

出　　处：《中国测试》2024年第9期83-90,共8页China Measurement & Test

基　　金：中海油重点综合科研项目(YXKY-2021-HN-01)。

摘　　要：针对深水井开采中气侵溢流监测技术的预警滞后、虚警率高和准确率低等问题,提出一种早期监测与录井监测相结合的气侵预警智能模型。比较分析基于超声波的隔水管早期监测方法和基于神经网络的录井监测方法的优缺点,优选有效特征,作为模型训练的基础数据集;通过工况识别规则及标签处理方法,实现两种监测数据之间的气侵事件标签匹配;基于神经网络设计智能模型的训练方法,并研究预警分析软件的架构和工作流程。利用实际案例数据进行验证,试验结果表明,新模型相比于传统录井监测方法,预警时间提前;与早期监测方法相比能显著减少误报概率,且气侵量级的预报准确性也有大幅提升。因此早期监测与录井监测方法相结合的气侵预警智能技术,可以有效提高深水井开采过程中气侵预警的准确性和及时性,为深水钻井安全提供一种创新的技术支持。Aiming at the problems of delayed warnings,high false alarm rates,and low accuracy in gas invasion overflow monitoring techniques during deepwater well drilling,an intelligent gas invasion prediction model that combines early monitoring with logging monitoring is proposed.A comparative analysis of the advantages and disadvantages of the ultrasonic-based riser early monitoring method and the neural networkbased logging monitoring method was conducted to select effective features as the foundational dataset for model training;through operating condition identification rules and label processing methods,gas invasion event label matching between the two types of monitoring data is achieved;a training method for the intelligent model based on neural networks was designed,and the architecture and workflow of the prediction analysis software are studied.The model is validated using actual case data,and experimental results show that compared to traditional logging monitoring methods,the new model advances the warning time;compared to early monitoring methods,it can significantly reduce the probability of false alarms,and the accuracy of predicting gas invasion levels has also been greatly improved.Therefore,the intelligent gas invasion prediction technology that combines early monitoring with logging monitoring methods can effectively enhance the accuracy and timeliness of gas invasion prediction during deepwater well extraction,providing innovative technical support for the safety of deepwater drilling.

关 键 词：深水井 气侵 风险预警 早期监测 录井监测 

分 类 号：TB9[一般工业技术—计量学] X937[机械工程—测试计量技术及仪器] TE249[环境科学与工程—安全科学]