2018 IEEE 7th Data Driven Control and Learning Systems Conference (DDCLS)

5 / 2017-12-19 21:58:38

Turbulence / Multiphase Flow Intelligence Measurement Based on Coriolis Effect

Liquefied Natural Gas, Coriolis flowmeter, deep learning, Bayesian theory

With the increasing demand for information measurement in the process of industrial production, the monitoring and metering requirements for mass flow of complex fluids are also proposed.During the process of LNG filling, the flow characteristics of the fluid are turbulent / multiphase flow. There are complex phase interfaces with interfacial effects and relative velocities. The flow states are very complex and the flow characteristics depend on the relative Speed, flow properties, pipe structure and flow direction. To ensure the measurement accuracy of Coriolis flowmeter during LNG filling, intelligent selection of characteristic parameters is required.
With the goal of precise measurement of Coriolis signal in LNG filling process, an experimental platform for the intelligent measurement of Coriolis signal and chord signal is established. With the deep learning algorithm and Bayesian verification method, Coriolis signals and chord signals match the process parameters, and establish a flow measurement process database. Through the theoretical analysis, numerical simulation and accuracy comparison experiment under different working conditions, form the new technology of flow intelligence detection based on Coriolis effect.