Identification of gas/solid two-phase flow regimes using electrostatic sensors and neural-network techniques

Abstract

In the gas/solid two-phase system, solid particles can accumulate a large number of electrostatic charges because of collision, friction and separation between particles or between particles and the wall. Through the detection and processing of the induced fluctuation charge signal, a measuring system can obtain two-phase flow parameters, such as flow regime, concentration and velocity. A novel methodology via introducing the characteristics of speech emotion recognition into flow regime identification is proposed for improving the recognition rate in gas/solid two-phase flow systems. Three characteristics of electrostatic fluctuation signals detected from an electrostatic sensor are extracted as the input of back propagation (BP) neural networks for flow regime identification. They are short-term average energy, Mel-frequency cepstral coefficients (MFCC) and cepstrum. The results show that the method based on each characteristic of the electrostatic fluctuation signal and BP neural networks can identify the three flow regimes of gas/solid two-phase flow in a horizontal pipe, and the identification rate of the method based on the three characteristics and BP neural networks is up to 97%, much higher than the methods based on a single characteristic.