A Flowrate Estimation Method for Gas–Water Two-Phase Flow Based on Multimodal Sensors and Hybrid LSTM-CNN Model

Accurate flowrate estimation in gas-water two-phase flow is pervasive and important in the energy industry for monitoring and optimizing the production process. How to accurately estimate the flowrate in the gas-water two-phase flow remains a challenging problem. This work proposes a flowrate estimation method based on multimodal sensors and hybrid LSTM-CNN model. The multimodal sensors including the Venturi tube and the electrical resistance tomography (ERT) sensor, which provide pressure, temperature, and electrical voltage information. The hybrid LSTM-CNN model is composed of the bidirectional long short-term memory network (Bi-LSTM), and the multiscale convolutional neural network (MSCNN). This method can estimate gas and water phase flowrates simultaneously. Experimental results suggest that the proposed method has good performance of flowrate estimation, compared with using only Venturi tube. Additionally, we demonstrate the rationality of the design model through ablation experiments and assess the robustness of the model with different flow patterns. Finally, the method is compared qualitatively and quantitatively with other flowrate estimation methods, and the results show its superiority in different aspects and various indicators.