A non-radioactive flow meter using a new hierarchical Neural network

In this paper, a Multilayer neural network has been developed to carry out the fusion of multi-sensor information for a new multiphase flow meter (MPFM) device. The velocity and density of each phase are determined using the fluid electrical and acoustic property signals which are combined with the physical models of multiphase fluids, in addition to the venturi, differential pressure, and absolute pressure sensors. Two rings of high and low frequency ultrasonic sensors are used to overcome the uncertainties of the electrical sensors in the range of 40-60% water-cut for low and high gas fractions respectively. Experimental results on a multiphase flow loop show that real-time classification of phase flow rates for up to 90% gas fraction can be achieved with less than 10% relative error.