【作者】 孙涛；

【导师】 张宏建；

【作者基本信息】 浙江大学 ， 控制科学与工程， 2002， 博士

【摘要】 两相流广泛存在于动力、化工、石油、冶金、管道运输、医药、制冷等领域。研究两相流体力学，探讨两相流动的机理，准确检测两相流的各种参数对于现代工业设备的研制和开发具有非常重要的意义。本文在总结前人工作的基础上，将数据融合技术应用于两相流参数的检测研究。本文的主要工作与创新点在于：1) 将分位数方法与一阶差分法相结合，提出了一种基于一阶差分的粗差剔除方法，该方法与低通滤波器的结合可以有效去除两相流压力波动信号中的粗大误差以及信号频带以外的频率成分。该方法已经在两相流压力波动信号预处理中得到成功的应用。2) 熵是信号序列信息量的表征，作者在对差压信号的分析中提出了两种基于熵概念的特征——香农熵和阈值熵。实验表明两种特征可以有效的区分不同流型的差压信号。3) 将数据融合技术应用于流型辨识研究，对其中的层次、结构等问题进行了深入的探讨。采用差压波动信号作为流型辨识的参量，提出了基于多个特征的模糊积分流型辨识方法。实验在DN20和DN25的水平管气液两相流上进行，差压信号的取压距离分为155mm和1000mm。对泡状流、弹状流和环状流三种典型流型的辨识实验表明，模糊积分融合方法可以有效地融合各个特征的信息，其流型辨识结果优于参与融合的任一特征量的辨识效果。155mm和1000mm两个取压距离的流型辨识实验表明，两个取压距离上的差压信号都可以有效地辨识流型，两者之间不存在显著的差别。4) 研究了差压波动信号与两相流系统参数之间的关系。讨论了影响两相流压降的因素，用多项式拟合的方法研究了不同流型下差压信号方差与Lockhart—Martinelli参数X之间的函数关系。在此基础上提出利用单一变量进行两相流双参数测量的方案。DN20和DN25水平管气液两相流实验数据表明，利用差压波动信号进行两相流分相流量的测量是可行的。更多还原

【Abstract】 Two-phase flow widely exists in processes such as power, chemical, petrol, metallurgy, pipeline transport, medical and refrigeration industries. Researching of the hydrodynamics mechanisms and precisely detection of parameters in two-phase flow are very important for the development of modern industrial equipment. This dissertation focuses on the application of data fusion in two-phase flow regime identification. Following is the main contribution of the dissertation.1)Based on quartile and first order differential, a new outlier detection algorithm is presented, Experiments show that the method combining with low pass filter can remove gross error and unwanted frequency components. The method has been applied successfully in two phase flow pressure signal processing.2)Entropy is a measure of information in signals. Based on the conception of entropy, two new features-Shannon entropy and Threshold entropy are proposed in flow regime identification. Experiments show that the features can effectively discriminate flow regimes.3)Data fusion is proposed in flow regime identification researches. Some problems such as fusion level and architecture of applying data fusion in flow regime research are studied. A Fl (Fuzzy Integral) fusion regime identification algorithm is implemented on horizontal gas-liquid two-phase flow. Inputs of the Fl algorithm are three features extracted from differential pressure signal. The experiment is done on pipes of diameter 20mm and 25mm, and pressure sample distance of 155mm and 1000mm. Three typical flow regime, bubble, slug and annularIIIABSTRACTflow are considered. Results proved that the Fl fusion method outperforms any individual sensors. The results also show that no significant different exist between pressure sample distance of 155mm and 1000mm. Pressure signal from the two distances can all identify flow regime effectively.4)Relations between differential pressure signal and two-phase flow parameters are studied in the dissertation. Factors that affect two-phase flow pressure drop are discussed in details at first. Polynomial is used to fit the relation between property of differential pressure signal and Lockhart- Martinelli parameter X. Based on the relation, a dual- parameter identification scheme using only differential pressure signal is given. The experiment data from horizontal gas-liquid flow on 20mm and 25mm diameter pipes proved that separate phase flow-rate can be measured from differential pressure signal.更多还原