Abstract:

Aiming at the gas-liquid two phase flow in horizontal pipeline, a comprehensive method for flow pattern identification and flow velocity and phase fraction measurement was proposed, using only a single conductance sensor. Firstly, a conductance sensor composed of six ring-shaped electrodes was adopted to acquire flow information about the water holdup and cross-correlation velocity of gas-liquid two phase flow under different flow conditions. Secondly, based on the signal fluctuation analysis, the flow characteristics of different flow patterns were revealed. The mean and variance of the time series of normalized voltage and the cross-correlation velocity were calculated as a feature vector for flow pattern representation. Using support vector machine (SVM) method which was suitable for small samples, 15 binary classifiers with radial basis function as kernel function were constructed by “one-to-one”strategy. The parameters of the identification model were optimized by cross validation method, and 6 flow patterns (stratified flow, wavy flow, bubbly flow, plug flow, slug flow and annular flow) were identified with the identification rate of 93.1%. Thirdly, based on the identification results, the measurement models for calculations of phase fraction and mean velocity through normalized voltage and cross-correlation velocity were established for every specific flow patterns. Compared with reference values at inlet, dynamic experiments showed that the root mean square errors (RMSE) of water fraction and gas fraction were 2.56% and 2.73%, and the RMSE of mean velocity was 0.69m/s. The proposed method provides a simple, efficient, low-cost, and non-invasive flow pattern identification and process parameter measurement strategy for gas-liquid two-phase flow, which has important scientific and engineering significance.

Key words: gas-liquid flow, instrumentation, velocity, fraction, measurement, flow pattern identification

摘要：

针对水平管道气液两相流，提出一种仅采用单模态电导传感器实现流型准确辨识和流速、含率准确测量的综合性测试方法。首先，采用由6个电极环组成的电导传感器在线获取不同流动状态下含率和速度的相关信息。其次，在信号波动分析基础上，揭示了不同流型下的流动特性，进而提取了量纲为1电压序列的均值和方差及互相关速度作为流型表征的特征向量；采用适用于小样本的支持向量机方法，以“一对一”策略构建了15个以径向基为核函数的二分类器用于分层流、波状流、泡状流、塞状流、弹状流和环状流的划分，采用交叉验证方法对模型参数进行了优化选取，实现了6种流型的准确识别，平均识别率为93.1%。最后，在流型划分基础上，针对性地提出利用量纲为1电压和互相关速度获取不同流型下分相含率和平均速度的计算模型，并与入口参考含率和参考流速进行了对比。动态实验表明，含水率和含气率的均方根误差分别为2.56%和2.73%，平均流速的均方根误差为0.69m/s。

关键词: 气液两相流, 仪器仪表, 流速, 含率, 测量, 流型识别