K-均值聚类在CCERT系统流型辨识中的应用

流型是气水两相流的重要参数之一,对气水两相流的流动影响很大。基于电容耦合电阻层析成像(CCERT)系统,以水平管道气水两相流流型为研究对象,利用主成分分析(PCA)方法提取不同流型下采集到的电导率信息的主成分,消除不同电极对之间信号的冗余,进而结合K-均值聚类算法实现流型辨识。实验结果表明,该方法具有较高准确度,对于水平管道泡状流、层状流和环状流3种典型的气水两相流流型的静态辨识准确率可以分别达到97%、96%和99%,动态辨识准确率可以分别达到92%、90%和87%。

Application of K-means clustering in flow pattern identification of CCERT system

Flow pattern is one of the most important parameters of gas-liquid two-phase flow and has great influence on the flow of gas-liquid two-phase flow. In this paper,gas-liquid two-phase flow patterns in a hori-zontal pipe are analyzed with capacitively coupled electrical resistance tomography (CCERT)system. Princi-pal component analysis (PCA)method is used to extract the principal components of the electrical conductivi-ty information under different flow patterns,and thus the redundancy of signal between different electrode pairs can be eliminated. The three flow patterns are recognized by using K-means clustering algorithm. The experi-mental results show that this method has high accuracy. The static identification accuracy for bubble flow, stratified flow and annular flow is 97%,96% and 99%,respectively,and the dynamic identification accuracy is 92%,90% and 87%,respectively.