Multi-sensor Signal Statistical Feature Processing Method for Status Monitoring of Gas-Water Two-Phase Flow in Horizontal Pipe

Abstract

As a complex and time-varying nonlinear process, gas-water two-phase flow widely exists in various industries. It has a variety of steady flow statuses and uncertain transition flow statuses. In order to realize accurate monitoring of gas-water two-phase flow status, a method combining Independent Component Analysis (ICA) and Canonical Variable Analysis (CVA) is proposed. Multi-sensor data of gas-water two-phase flow are comprehensively measured. ICA is used to obtain independent components of data. CVA is used to extract flow status canonical features and establish statistical monitoring indicators, which effectively solves the cross-correlation and temporal correlation of multi-sensor data and realizes flow status monitoring. The monitoring indicator limits of flow status are calculated by Kernel Density Estimation (KDE) method. Comparing Wasserstein Distance (WD) of monitoring indicator probability distributions, various flow statuses are analyzed in details to reflect the changing trend of flow status. The method is verified by using the measured data of the horizontal loop of gas-water two-phase flow experimental facility, and its effectiveness is proved.