Slow and Steady Feature Analysis Based Status Monitoring for Gas-liquid Two-phase Flow in Horizontal Pipe

Abstract

To effectively obtain flow status information and monitor the flow status of gas-liquid two-phase flow in horizontal pipe, a status monitoring method based on slow and steady feature analysis is proposed with concurrent monitoring of steady states and process dynamics. In this method, both the first and higher order temporal constraints are imposed on feature extraction to extract the essential characteristics that reflect the flow law in different flow statuses of two-phase flow. Representation models for different typical flow regimes are established to describe the steady states and process dynamics. On this basis, by monitoring statistics, the current flow status can be monitored online and the dynamic characteristics can be detected with meaningful physical interpretation. Moreover, the dynamic monitoring statistics contain lots of derivative information which captures the overall change trend and dynamic characteristics, so this method is particularly suitable for monitoring the variations of flow status. The applications to monitoring three flow regimes and transition status from wave flow to slug flow demonstrate the feasibility and efficacy.