Identification of Gas-Solid Two-Phase Flow Regimes Based on Electrostatic Sensor and CB-ResNext Network

Electrostatic sensors are commonly used to extract parameter information from gas-solid two-phase flow in the energy and power industry, characterized by a simple structure and high sensitivity. Additionally, composite electrostatic sensors can effectively enhance the measurement accuracy of electrostatic signals. Therefore, this paper proposes a gas-solid two-phase flow regimes identification method based on electrostatic sensors and the CB-ResNext network. Firstly, composite electrostatic sensors are employed to collect electrostatic signals. Then, leveraging the characteristics of the MTF (Markov Transition Field) to retain complete signal information, the one-dimensional electrostatic signals are transformed into two-dimensional feature images. Finally, the CB-ResNext network, incorporating the CBAM (Convolutional Block Attention Module) attention module with the ResNext network, is established to extract key features from the images for flow regimes identification. Experimental results indicate that this method achieves identification accuracy of over 95% for laminar flow, annular flow, and central flow, demonstrating promising prospects for industrial applications.