利用数据融合实现流型的在线识别

REALIZING ON-LINE FLOW PATTERN IDENTIFICATION BY DATA COALESCENCE

两相流的流动是一个复杂的多变量随机过程。认识流型的形成机理和特征以及流动规律一直是两相流领域的重要研究方向。传统的流型识别方法 (流型图和流型转换判别 )能够预测流型及其转换 ,但是要应用它来获取流体的实时流动参数还是有较大的难度。而在石油天然气工程 (钻井、固井、集输管道等 )应用实际中 ,往往需要实时了解流动参数以确定系统的运行状况。针对这个问题 ,为掌握实时流动参数的特性而进行了较深入的研究。文章提出了一种基于数据融合理论两相流流型识别新方法 ,并设计了流型的在线识别系统。研究结果表明 ,该识别方法具有置信度高 ,容错性好 ,性能稳定 ,减小了传感器的不确定误差 ,降低了单个传感器的性能要求等优点。这种方法对两相流流型具有实时识别的能力 。

Two-phase flow is a complex multivariable random process. It, is always an important research project to identify the formation mechanism and characteristics of flow pattern and the flow law. The conventional flow pattern identification methods (flow pattern map and flow pattern transform identification) can predict flow pattern and its transform, but it is difficult to obtain the real-time flow parameters of fluids by applying these methods. In oil and gas engineering as drilling, cementing and pipeline transportation, etc., it is often necessary to know the flow parameters in time to determine the operating circumstances in the system. For this reason, a deepgoing research was carried out for grasping the characteristics of real-time flow parameters. Through investigation, a new method of identifying the two-phase flow pattern on the basis of data coalescence theory was put forward and an on-line flow pattern identification system was designed. The results indicated that such a identification method was of these advantages as high, confidence level, good fault-tolerance and stable performance, thus decreasing the uncertain error of sensor and reducing the demand for the performance of single sensor. This method is of real-time two-phase flow pattern identification capability and a vast range of prospects for applying it to oil and gas engineering.