Session: 06-17-02 AI Technology for Ocean Engineering - II (On-Demand Video Session)

Submission Number: 180154

Intelligent Spatiotemporal Reconstruction of Pipeline Solid-Liquid Two-Phase Flow From Sparse Observations 

Real-time perception of global flow field evolution is critical for ensuring the operational safety of deep-sea mining hydraulic lift systems. However, the complex hydrodynamics within ultra-long risers, characterized by violent turbulent pulsations and non-uniform particle agglomeration, pose significant monitoring challenges. Existing measurement techniques are often constrained by sparse, local observations and lack the predictive capabilities necessary to anticipate potential hazards like pipeline blockages. To address these limitations, this paper proposes a novel spatiotemporal prediction and reconstruction framework tailored for solid-liquid two-phase flows. We introduce a cascaded deep learning architecture that sequentially integrates a Long Short-Term Memory (LSTM) network with a pre-trained generative spatial reconstruction model. This framework establishes a non-linear mapping mechanism from sparse historical cross-sectional observations to future three-dimensional (3D) global particle volume fraction (PVF) fields.Validated against high-fidelity CFD-DEM simulation datasets, the proposed model demonstrates exceptional performance. The LSTM module effectively captures latent temporal dynamics, achieving a prediction accuracy with a coefficient of determination R² exceeding 0.94 and a Mean Squared Error (MSE) as low as 7.27E-4 using an look-back window of 5 frames. Furthermore, systematic evaluations reveal the model's robustness: it retains high predictive credibility (R² ≈ 0.8) even under significant interference (SNR ≈ 9.4 dB). By enabling the high-fidelity, forward-looking visualization of flow instabilities from limited data, this method provides a foundational digital twin technology for intelligent early warning and proactive safety control in deep-sea engineering operations.

Presenting Author: Shengpeng Xiao shanghai jiao tong university

Presenting Author Biography: The presenter is from Shanghai Jiao Tong University, specializing in two-phase flow transport in pipelines for deep-sea mining.

Authors:

Shengpeng Xiao shanghai jiao tong university
Haoxi Li Shanghai jiao tong university
Hongbo Zhu shanghai jiao tong university
Dai Zhou shanghai jiao tong university
Yan Bao shanghai jiao tong university
Zhaolong Han shanghai jiao tong university