摘要
针对气力输送管道中测控装置后常见的三种过渡流型,即中心流、环状流和层状流,采用静电传感器作为测量装置获得静电流动噪声信号,借鉴语音信号处理方法,提取静电流动噪声信号的梅尔频率倒谱系数(MFCC)及其一阶差分作为特征参数,用特征参数训练连续高斯混合密度隐马尔科夫模型(CGHMM),建立不同流型的模型库,再用训练好的CGHMM模型对提取的特征参数进行分类,进而实现流型识别.实验结果表明,该方法识别率达到98%,为气固流流型识别及气力输送测控装置提供了新的研究方法.
Aiming at three common transitional flow regimes behind the detection and control devices in the pneumatic conveying pipeline, namely central flow, annular flow and stratified flow, the electrostatic flow noise signals were obtained through adopting an electrostatic sensor as the measuring equipment. With the speech signal processing method, the mel-frequency cepstrum coefficient （MFCC） and its first order difference of electrostatic flow noise signals were extracted as the feature parameters. In addition, the continuous Gaussian mixture hidden Markov model （CGHMM） was trained with the feature parameters, and the model libraries for different flow regimes were established. Then the extracted feature parameters were classified with the trained CGHMM model, and thus, the flow regimes identification was realized. The experimental results show that the identification rate of the proposed method reaches 98%, and a novel research method for the gas-solid flow regime identification as well as the pneumatic conveying detection and control devices is provided.
出处
《沈阳工业大学学报》
EI
CAS
北大核心
2013年第5期555-560,共6页
Journal of Shenyang University of Technology
基金
国家自然科学基金资助项目(51177120)
关键词
气固两相流
测控装置
语音信号处理
流型识别
梅尔频率倒谱系数
静电传感器
流动噪声信号
连续高斯混合密度隐马尔科夫模型
gas-solid two-phase flow
detection and control device
speech signal processing
flow regime identification
mel-frequency cepstrum coefficient （MFCC）
electrostatic sensor
flow noise signal
continuous Gaussian mixture hidden Markov model （CGHMM）
