Two-component Flow Identification Based on Neural Network and 8-Electrode Capacitance Sensor

摘要

A new method of two-component flow pattern identification based on neural network and an 8-electrode ECT sensor is proposed in this paper. The general idea relies on the finding of flow pattern information hidden in ECT sensor outputs by means of a trained neural network. To obtain good identification effect, the neural network inputs are not simple ECT sensor outputs, but the feature parameters extracted from the ECT sensor outputs. 8 feature parameters are extracted and a two-rank competitive neural network using the feature parameters as inputs is set up, trained and used to identify flow pattern on-line. Simulation results show that the proposed method has good identification precision and fast identification speed.