基于案例推理的井漏风险预警方法

发布时间：2018-12-16 13:28

【摘要】：常规智能井漏预警方法存在知识获取成本高、处理规则复杂及系统自我进化难等问题,因此预警效果不理想。案例推理在解决信息不完备、经验丰富、内部机理认识尚不完全清楚等方面的问题时,具有很强优势。为了辅助或代替工程师长时间监测和分析特征参数的异常,避免或降低漏报和误报,文中提出了基于案例推理的井漏风险智能预警方法。该方法主要包括预警案例表示、案例检索及案例学习3部分。预警流程:首先把井漏的特征参数的实时数据流表示成新案例,其次调用预警方法实现函数检索相似案例,最后利用检索结果的评价标准及相似案例的相关信息进行井漏风险智能预警。四川盆地的10个井漏案例的实验结果表明:该方法能够提前进行井漏风险预警,不受地质、钻井操作及主观性等因素的影响,其适用范围更广,智能性更强,具有重要的现场指导意义。
[Abstract]:The conventional intelligent well leakage early warning method has many problems, such as high cost of knowledge acquisition, complex rules and system self-evolution, so the effect of early warning is not ideal. Case-based reasoning (CBR) has a strong advantage in solving the problems of incomplete information, rich experience and incomplete understanding of internal mechanism. In order to assist or replace engineers in monitoring and analyzing the anomalies of characteristic parameters for a long time, and to avoid or reduce false alarms and false alarms, a CBR-based intelligent early warning method for well leakage risk is proposed in this paper. This method mainly includes three parts: early warning case representation, case retrieval and case learning. Early warning process: first, the real-time data flow of the characteristic parameters of the well leakage is expressed as a new case, and then the early warning method is called to retrieve the similar cases. Finally, the evaluation criteria of retrieval results and the relevant information of similar cases are used to carry out intelligent warning of well leakage risk. The experimental results of 10 well leakage cases in Sichuan Basin show that the method can predict the risk of well loss in advance and is not affected by geology, drilling operation and subjectivity, and its application scope is wider and its intelligence is stronger. It has important site guidance significance.
【作者单位】：西南石油大学石油与天然气工程学院;
【基金】：国家科技重大专项课题“钻井工程设计与工艺软件”(2011ZX05021-006)
【分类号】：TE28

【相似文献】