Yu Wu

Yu Wu

Auto-generated profile, based on publicly available information

Yu Wu

National Key Laboratory of Marine Engine Science and Technology, Shanghai 201108, People’s Republic of China

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 April 2021 in Measurement Science and Technology
Verified
Journal Article
Closed access
PDF file
YWYu Wu
Yu Wu
QWQian Wang
Qian Wang
CZC. Y. Zhao
C. Y. Zhao
In the present study, a novel temporal three-dimensional particle tracking velocimetry (3D PTV) algorithm for flow measurements with only two views is developed and validated with synthetic particles. The spatial information in image and object spaces, as well as the temporal predictions, are strongly coupled to improve the particle tracking accuracy. A well-designed cost function, simultaneously penalizing the reconstruction and tracking processes, is minimized to determine the most reliable traces. The algorithm shows a correctness over 98% up to 0.0273 ppp (particles per pixel) when using ideal synthetic particle positions, which is superior to several artificial intelligence methods. Moreover, an improved particle identification algorithm is proposed to handle overlapped particles and reduce the error introduced into the 3D PTV scheme. The algorithm adopts a particle position shifting process to tackle the correct particle numbers iteratively, which shows better performance than some other methods. A comparative study indicates that particle identification accuracy has a significant effect on the subsequent 3D reconstruction and tracking processes. The 3D PTV and particle identification algorithms show good consistencies under two types of flow conditions: a homogeneous isotropic turbulent flow and a vortex ring flow. Comparing with multiple-view setups, two-view systems are more compact and cost-effective, especially in conditions requiring high-speed cameras. With the newly established algorithms, a two-view system is now able to handle higher particle-seeding densities and thus can resolve higher spatial resolutions, which is significant for applications in turbulent flow and particle motion measurements.

18 March 2022
Verified
Book Chapter
Closed access
PDF file
Tao WangTao Wang
Tao Wang
YWYu Wu
Yu Wu
WJWenkai Jiang
Wenkai Jiang
YZYongwei Zhang
Yongwei Zhang
WLWen Lv
Wen Lv
ZYZhi Yang
Zhi Yang

3 May 2022 in Measurement Science and Technology
Verified
Journal Article
Closed access
YWY Wu
Y Wu
CZC. Y. Zhao
C. Y. Zhao
QWQian Wang
Qian Wang
Determining the time-resolved three-dimensional (3D) three-component (3C) velocity is essential for complex turbulent flow measurements. The current study is an extension of a recently developed temporal-spatial three-dimensional particle tracking velocimetry (TS 3D-PTV) technique established for two-view imaging systems. Two improvements have been embedded in TS 3D-PTV algorithm to improve the accuracy at high particle image densities (up to 0.03 ppp). One is using the neighboring particle information to correct the predicted positions and select the temporal particles with higher probability; the other is to iteratively optimize the 2D particle positions during the tracking process using the temporal and image information. The synthetic particle tests indicate that the correctness can be increased by 4.7%–5.8%, to reach a value about 92% with the improved algorithm around 0.03 ppp. The comparative results also indicate that using an advanced particle identification algorithm can improve the correctness over 20%. Two experiments, including a buoyancy jet in water and a transient droplet splashing process, have been conducted with a two-view shadowgraph imaging system. Different tracking algorithms have been conducted to determine the 3D trajectories of seeding particles or secondary droplets comparatively. The new algorithm has shown the best performance with much longer and more reliable trajectories, which indicates the tracking interruption caused by particle overlapping is reduced. The newly developed algorithms have further improved the performance under high seeding density conditions, which makes the two-view shadowgraph 3D PTV system adaptable to more experimental conditions.

1 October 2025 in Physics of Fluids
Verified
Journal Article
Closed access
DZD. L. Zhuang
D. L. Zhuang
YWY. Wu
Y. Wu
QWQian Wang
Qian Wang
Particle tracking velocimetry (PTV) is a flow velocity measurement technique with great potential to achieve high spatial resolution. It measures the velocity through the Lagrangian trajectories of tracer particles. When the tracer concentration is high, one of the most challenging steps lies in how to track the particles at different times correctly. Errors in particle matching not only interrupt the tracking process but also reduce the accuracy of the obtained velocity field. In the current study, recurrent neural network architectures, which are proficient in dealing with sequential data, are employed to address this issue. Various combinations of long short-term memory (LSTM) and gated recurrent unit (GRU) networks are proposed and compared. The networks are trained and tested using a dataset containing several typical turbulent flow fields, including flow around a cylinder, wake of a hydrofoil, channel flow, and thermal plumes. Three-dimensional trajectories are generated based on computational fluid dynamics simulation results, covering a wide range of particle densities in different complex flow fields. The long trajectories are segmented into short path snippets, with a total number of 150 000, while 80% is used for network training and 20% for validation. The networks are trained to predict the next frame particle positions based on the trajectories in the past 6–10 frames. Based on the parametric comparison, a double-layer LSTM-GRU network with 128 neurons for each has shown the best performance. The results indicate that the accurate prediction rate from the LSTM-GRU network has increased by over 15% compared with two traditional PTV algorithms, including the four-frame best estimation and the relaxation algorithm. In channel flow characterized by twisted particle trajectories, when the particle density is ∼0.03 ppp, the extraction rate of the LSTM-GRU algorithm exceeds 95%. Even in complex turbulent regions of a thermal plume flow field, where the performance of traditional algorithms drops significantly as the particle concentration increases, the extraction rate of the LSTM-GRU algorithm remains higher than 90%. Finally, the prediction algorithm is applied to practical experimental data of a buoyant jet. The results reveal that the proposed network can resolve satisfactory long trajectories with varying loop segmentation lengths between 6 and 10. The proposed LSTM-GRU network has been proven to be robust and computationally efficient, which has significant potential for large-scale particle tracking tasks in a wide range of experimental applications.

30 March 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
CLChunfeng Li
Chunfeng Li
WPWeihua Pu
Weihua Pu
Yanzhao AnYanzhao An
Yanzhao An
GCGuoliang Chu
Guoliang Chu
Yiqiang PeiYiqiang Pei
Yiqiang Pei
MMM. S. Mansour
M. S. Mansour
Diesel-methanol partially premixed combustion (DM-PPC), characterized by prior mixing and controlled auto-ignition, has attracted increasing attention as an effective strategy to improve combustion controllability and extend methanol utilization in compression-ignition engines. However, the spatiotemporal evolution of ignition and flame development under different injection intervals remains insufficiently quantified. In this work, diesel-methanol dual fuels direct-injection jet flame tests are conducted in a constant-volume chamber (750K and 40bar) with different injection intervals. We proposed a CNN-based denoising and recognition method for jet flame Schlieren images, centered on FFDNet with noise-level map generation-estimation and robust segmentation. Three DM-PPC modes were classified based on two jet flames characteristics with different spray intervals: 1) In methanol early injection premixing leading, low-temperature DM-PPC mode, blue flame lasts about 0.7–0.8 ms, yellow flame is extremely weak and short-lived about 0.4 ms, the flame-contour area growth rate peak is low about 2.0–3.0m²/s, and the total flame lift-off length (FLOL) is short about 40–80 mm toward the methanol side. 2) In diesel-methanol co-injection, premixed-diffusion-coupled DM-PPC, multi-point ignition kernels within the diesel-methanol mixing zone leads to middle flame contour-growth peaks about 7.0–9.0 m²/s and average flame-speed spikes of 170–220 m/s, with the global flame layer located between the diesel and methanol jets, reflecting a balanced coupling of premixed and diffusion flames. 3) In diesel early injection leading, diffusion-flame-dominated DM-PPC, blue-flame stage is shorter but followed by strong, long-lived yellow-flame clusters, the flame contour growth rate peak reaches 1.5–1.8m²/s, the initial average flame-speed spike rises to 200–230 m/s, and the early total FLOL is the largest about 100 mm with a pronounced shift toward methanol side. The novelty of this study is the integration of a spatially adaptive FFDNet-based denoising framework with high-speed Schlieren diagnostics to enable quantitative identification of injection-interval-governed DM-PPC flame modes under high-temperature, high-pressure conditions.

23 September 2026 in Fuel
Verified
Journal Article
Closed access
CLChunfeng Li
Chunfeng Li
WPWeihua Pu
Weihua Pu
Yanzhao AnYanzhao An
Yanzhao An
GCGuoliang Chu
Guoliang Chu
YPYiqiang Pei
Yiqiang Pei
MMMohy Saad Mansour
Mohy Saad Mansour