基于案例推理的井漏风险预警方法

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。

Risk pre-warning method of well leakage based on case reasoning

Conventional intelligent pre-warning methods of lost circulation have the following problems:high cost of acquiring knowledge,complicated processing rules and difficulty of system self-evolving.Therefore,they are not ideal.The intelligent casebased reasoning technology has an advantage in dealing with the problems when the information is not complete,solving experience is vast and internal mechanism is not fully clear.In order to replace or assist the long time monitoring and analysis of the abnormal feature parameters,and reduce or even avoid misinterpretation and omission,this paper puts forward the well leakage risk prewarning method based on case reasoning.The method mainly includes three parts:the representations,retrieval and study of prewarning case.The early warning process includes three steps:firstly,turn the real-time data of lost circulation parameters into a new case;secondly,invoke application function of the pre-warning method to retrieve similar cases;finally,warn the risk of lost circulation intelligently by evaluating criteria of similar retrieval results and the relevant information of some similar cases.The results of simulation experiment show that this method can warn risk of lost circulation in advance intelligently through 10 warning cases in the Sichuan Basin and is not affected by factors like geology,drilling operation and subjectivity,so it is used much widely and intelligently and has great significance in field guidance.