基于ACO和RVM的两相流流型特征选择方法  

Selection method of two-phase flow regime features based on ant colony optimization and relevance vector machine

作　　者：孙斌[1] 杨晓明[1] 

机构地区：[1]东北电力大学能源与动力工程学院,吉林132012

出　　处：《仪器仪表学报》2012年第10期2181-2186,共6页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(50706006)资助项目

摘　　要：为提高流型识别的准确率,提出了基于蚁群算法(ant colony optimization,ACO)和相关向量机(relevance vector machine,RVM)封装模式的流型特征选择方法。首先采用小波包变换(wavelet packet transform,WPT)、经验模式分解方法(empirical mode decomposition,EMD)对原始压差波动信号进行分解,分别提取压差波动信号的时域无量纲指标和各分解信号的能量和熵组成融合特征。然后采用ACO和RVM进行特征选择和识别,选出有利于流型识别的特征优化组合。空气-水两相流型识别的实验结果表明:该方法能实现流型特征的有效缩减,经优化组合的最优特征子集识别率达95%以上,与其他方法相比具有更高的识别率。A selection method of flow regime features of wrapper mode based on ant colony optimization and relevance vector machines （ACO-RVM） is proposed in order to improve the accuracy of flow regime identification. Firstly, the original differential pressure fluctuation signals are decomposed with wavelet packet transform （WPT） and empirical mode decomposition （EMD）, respectively. The dimensionless indicators of original differential pressure fluctuation signals in time domain, as well as the energy and entropy of each decomposed signal are extracted respectively, which constitute the fusion features. Then ACO and RVM classifiers are used to carry out feature selection and flow regime identification so as to select more superior feature optimization combination in favor of flow regime identification. The experiment results of air-water two phase flow regime identification demonstrate that this method can reduce the number of flow regime features effectively and the identification accuracy reaches above 95% using optimized feature subset, which indicates that the proposed method has higher identification rate than other methods.

关 键 词：特征选择 蚁群优化算法 相关向量机 流型识别 

分 类 号：O359.1[理学—流体力学] TH73[理学—力学]