Volume 93
您当前的位置:首页 > 期刊文章 > 当期目录 > Volume 93
Zhang, D., Huang, H., Zhou, W., Feng, M., Zhang, D., & Gao, L. (2024). Reconstruction of particle distribution for tomographic particle image velocimetry based on unsupervised learning method. Particuology, 93, 349-363. https://doi.org/10.1016/j.partic.2024.06.016
Reconstruction of particle distribution for tomographic particle image velocimetry based on unsupervised learning method (Open Access)
Duanyu Zhang a b, Haoqin Huang a b, Wu Zhou a b, Mingjun Feng a b, Dapeng Zhang a b, Limin Gao c
a School of Energy and Power Engineering, University of Shanghai for Science and Technology, Shanghai, 200093, China
b Shanghai Key Laboratory of Multiphase Flow and Heat Transfer for Power Engineering, University of Shanghai for Science and Technology, Shanghai, 200093, China
c National Key Laboratory of Aerodynamic Design and Research, Northwestern Polytechnical University, Xi'an, 710072, China
10.1016/j.partic.2024.06.016
Volume 93, October 2024, Pages 349-363
Received 18 January 2024, Revised 11 June 2024, Accepted 17 June 2024, Available online 18 July 2024, Version of Record 6 August 2024.
E-mail: zhouwu@usst.edu.cn

Highlights

• Reconstruction of particle distribution in Tomo-PIV via 3D CNN are explored.

• An unsupervised reconstruction technique based on U-net is proposed.

• The proposed approach outperforms traditional algebraic reconstruction techniques.


Abstract

The development of deep learning has inspired some new methods to solve the 3D reconstruction problem for Tomographic Particle Image Velocimetry (Tomo-PIV). However, the supervised learning method requires a large number of data with ground truth as training information, which is very difficult to gather from experiments. Although synthetic datasets can be used as alternatives, they are still not exactly the same with the real-world experimental data. In this paper, an Unsupervised Reconstruction Technique based on U-net (UnRTU) is proposed to reconstruct volume particle distribution explicitly. Instead of using ground truth data, a projection function is used as an unsupervised loss function for network training to reconstruct particle distribution. The UnRTU was compared with some traditional algebraic reconstruction algorithms and supervised learning method using synthetic data under different particle density and noise level. The results indicate that UnRTU outperforms these traditional approaches in both reconstruction quality and noise robustness, and is comparable to the supervised learning methods AI-PR. For experimental tests, particles dispersed in cured epoxy resin are moved by an electric rail with a certain speed to obtain the ground truth data of particle velocity. Compared with other algorithms, the reconstructed particle distribution by UnRTU has the best reconstruction fidelity. And the accuracy of the 3D velocity field estimated by UnRTU is 12.9% higher than that from the traditional MLOS-MART algorithm. It demonstrates significant potential and advantages for UnRTU in 3D reconstruction of particle distribution. Finally, UnRTU was successfully applied to the high-speed planar cascade airflow field, demonstrating its applicability for measuring complex fluid flow fields at higher particle density.

Graphical abstract
Keywords
Tomographic particle image velocimetry; 3D reconstruction; Unsupervised learning; Convolutional neural network