A study on soft measurement method of water holdup for gas/liquid two phase flow

For the conductance fluctuating signals measured from gas/liquid two phase flow in vertical upward pipe, the 10 feature quantities, which reflect flow characteristics of gas/liquid two phase flow, were extracted from time domain and frequency domain. The features in frequency domain were derived by using linear prediction method of speech signal processing and the features in time domain were derived by time series statistical analysis. The all extracted features were taken as the inputs of RBF neural network. At the flow conditions of water flow rate ranging from 1 (m(3)/hour) to 10(m(3)/hour) and gas flow rate ranging from 1 (m(3)/hour) to 130(m(3)/hour), the soft measurement model of the RBF neural network gave a good water holdup prediction result of gas/liquid two phase flow. This study provides a new way to measure the phase volume fraction of two phase flow by soft sensor.