Songan Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Songan Zhang has a total of 100 co-authors. Here are the top co-authors by number of shared publications:
- Mingmin Zhu
School of Aeronautics and Astronautics, Shanghai Jiao Tong University, Shanghai 200240, China
12 shared publications - Jinfang Teng
School of Aeronautics and Astronautics, Shanghai Jiao Tong University, Shanghai 200240, China
8 shared publications - Xiaoqing Qiang
School of Aeronautics and Astronautics, Shanghai Jiao Tong University 1 , Shanghai
7 shared publications - Hefang Deng
Global Institute of Future Technology, Shanghai Jiao Tong University 2 , Shanghai
6 shared publications Jinfang Teng
5 shared publications- Ming-Hsuan Yang
University of California, Merced, USA
5 shared publications
Latest Publications
- 1 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (6) , 6552-6559
ROAD : R esponsibility- O riented Reward Design for Reinforcement Learning in A utonomous D riving
Reinforcementlearning (RL) in autonomous driving employs a trial-and-error mechanism, enhancing robustness in unpredictable environments. However, crafting effective reward functions remains challenging, as conventional approaches rely heavily on manual design and demonstrate limited efficacy in complex scenarios. To address this issue, this study introduces a responsibility-oriented reward function that explicitly incorporates traffic regulations into the RL framework. Specifically, we first use a vision language model (VLM) to determine the liability in traffic collisions and propose reward signals. To mitigate VLM hallucination and ensure regulatory grounding, we propose a Traffic Regulation Knowledge Graph (TRKG) that structures unstructured traffic laws into a queryable ontology of driving scenarios and liability standards. This mechanism retrieves the precise regulatory context required to quantify accident responsibility (primary/shared/secondary), which is then used to modulate the crash penalty in the reward function. Building on these liability-informed signals, we construct a reward function and employ it to train the agent, thereby encouraging behavior that more faithfully adheres to traffic regulations. Experimental validations show that our approach achieves task success rates of 73.2% (intersection) and 54.0% (roundabout). Compared with the original policy, the success rate improves by +8.2 pp/+11.2 pp, while the ego vehicle's primary-liability share decreases by 13.5 pp/5.7 pp. The code for this work is available at https://github.com/Songan-Lab/ROAD.
- 1 June 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (6) , 6552-6559
ROAD : R esponsibility- O riented Reward Design for Reinforcement Learning in A utonomous D riving
Reinforcementlearning (RL) in autonomous driving employs a trial-and-error mechanism, enhancing robustness in unpredictable environments. However, crafting effective reward functions remains challenging, as conventional approaches rely heavily on manual design and demonstrate limited efficacy in complex scenarios. To address this issue, this study introduces a responsibility-oriented reward function that explicitly incorporates traffic regulations into the RL framework. Specifically, we first use a vision language model (VLM) to determine the liability in traffic collisions and propose reward signals. To mitigate VLM hallucination and ensure regulatory grounding, we propose a Traffic Regulation Knowledge Graph (TRKG) that structures unstructured traffic laws into a queryable ontology of driving scenarios and liability standards. This mechanism retrieves the precise regulatory context required to quantify accident responsibility (primary/shared/secondary), which is then used to modulate the crash penalty in the reward function. Building on these liability-informed signals, we construct a reward function and employ it to train the agent, thereby encouraging behavior that more faithfully adheres to traffic regulations. Experimental validations show that our approach achieves task success rates of 73.2% (intersection) and 54.0% (roundabout). Compared with the original policy, the success rate improves by +8.2 pp/+11.2 pp, while the ego vehicle's primary-liability share decreases by 13.5 pp/5.7 pp. The code for this work is available at https://github.com/Songan-Lab/ROAD.
- Open Access
- 12 May 2026
- Published by Elsevier in SSRN Electronic Journal
Effects of Fan-Root Clearance on Leakage-Flow Dynamics and Transition-Duct Aerodynamics in an Open Fan
The open fan is regarded as a promising alternative to high-bypass-ratio turbofan engines due to its high propulsive efficiency. However, the variable-pitch configuration at the rotor root increases the complexity of the root flow and introduces pronounced inlet distortion and unsteadiness to the downstream transition duct. To clarify the coupled inner–outer flow mechanisms at the fan root, numerical simulations of an integrated open-fan and transition duct are conducted in this study. The mechanisms of variable-pitch-induced complex flow phenomena near the open-fan root are investigated for different root clearance heights. The impact of root flow on the downstream transition duct is investigated in terms of inflow distortion and loss characteristics. RANS and URANS simulations are performed in a coupled inner–outer computational domain. The RANS parametric study identifies root clearance height as the primary factor governing transition-duct inlet distortion, motivating URANS simulations of three representative clearance heights—no clearance, 0.2% span, and 1.27% span—to resolve the unsteady rotor-root flow and its interaction with the transition-duct IGV.,The results further indicate that the root spindle configuration exerts a negligible influence on the overall aerodynamic performance of the open fan, with thrust variations within 0.2%, whereas the clearance height governs the evolution of root flow structures. Building on this understanding, the mechanisms underlying inlet distortion in the transition duct are elucidated, with leakage flow identified as the primary driver of the distortion pattern. URANS simulations are subsequently performed for three representative clearance heights to characterize the unsteady flow behavior. The vortex structures at the rotor root are first analyzed, revealing coupled interactions between leakage flow and inherent secondary flow structures. The resulting distortion and unsteadiness at the transition-duct inlet are then quantified. Finally, the influence of clearance height on the aerodynamic performance of the transition-duct IGV is assessed.,The present study clarifies the underlying mechanisms by which rotor-root secondary-flow structures modulate the transition-duct inflow, providing physical insights for the coupled aerodynamic design of open-fan rotor root and downstream core-engine components.
- 12 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessEffects of Fan-Root Clearance on Leakage-Flow Dynamics and Transition-Duct Aerodynamics in an Open Fan
The open fan is regarded as a promising alternative to high-bypass-ratio turbofan engines due to its high propulsive efficiency. However, the variable-pitch configuration at the rotor root increases the complexity of the root flow and introduces pronounced inlet distortion and unsteadiness to the downstream transition duct. To clarify the coupled inner–outer flow mechanisms at the fan root, numerical simulations of an integrated open-fan and transition duct are conducted in this study. The mechanisms of variable-pitch-induced complex flow phenomena near the open-fan root are investigated for different root clearance heights. The impact of root flow on the downstream transition duct is investigated in terms of inflow distortion and loss characteristics. RANS and URANS simulations are performed in a coupled inner–outer computational domain. The RANS parametric study identifies root clearance height as the primary factor governing transition-duct inlet distortion, motivating URANS simulations of three representative clearance heights—no clearance, 0.2% span, and 1.27% span—to resolve the unsteady rotor-root flow and its interaction with the transition-duct IGV.,The results further indicate that the root spindle configuration exerts a negligible influence on the overall aerodynamic performance of the open fan, with thrust variations within 0.2%, whereas the clearance height governs the evolution of root flow structures. Building on this understanding, the mechanisms underlying inlet distortion in the transition duct are elucidated, with leakage flow identified as the primary driver of the distortion pattern. URANS simulations are subsequently performed for three representative clearance heights to characterize the unsteady flow behavior. The vortex structures at the rotor root are first analyzed, revealing coupled interactions between leakage flow and inherent secondary flow structures. The resulting distortion and unsteadiness at the transition-duct inlet are then quantified. Finally, the influence of clearance height on the aerodynamic performance of the transition-duct IGV is assessed.,The present study clarifies the underlying mechanisms by which rotor-root secondary-flow structures modulate the transition-duct inflow, providing physical insights for the coupled aerodynamic design of open-fan rotor root and downstream core-engine components.
- Top Cited PapersOpen Access
- 5 March 2026
- Published by Elsevier in Journal of Web Semantics
- Vol. 89
Leverage Knowledge Graph and Large Language Model for law article recommendation: A case study of Chinese criminal law
- 5 March 2026
- Published by Elsevier in Journal of Web Semantics
- Vol. 89
Top Cited PapersOpen AccessLeverage Knowledge Graph and Large Language Model for law article recommendation: A case study of Chinese criminal law
- Open Access
- 24 February 2026
- Published by Global Power and Propulsion Society in Journal of the Global Power and Propulsion Society
- Vol. 10, 12-28
Effects of blade loading distribution on hub leakage flow of cantilevered stator
The hub leakage flow of cantilevered stators has a significant impact and aerodynamic stability of the axial compressor. The formation and development of the hub leakage flow are affected by the blade loading level and distribution, but the internal mechanism of this effect is not fully understood in previous studies. This paper uses a novel experiment facility with a high-speed moving endwall and numerical simulation methods to focus on the influence of a fore-loading and highly-loaded cantilevered cascade on the hub leakage flow and its mechanism, and compared with a typical middle-loading controlled diffusion airfoil cascade. The results demonstrate that although the two cascades have different blade loading levels, the radial distribution of the entropy loss coefficient at the condition of moving endwall is almost identical, which is related to the development location and radial extent of the leakage flow in the two cascades. The position of maximum hub leakage flow rate is delayed relative to the position of maximum blade loading, due to the increase of circumferential momentum by the leakage vortex in the passage. The main cause for the entropy loss is the coupling effect between the product of the leakage flow velocity and injected angle and the mainstream velocity. Based on the findings of this paper, for the hub airfoil of the cantilevered stator, the fore-loading airfoil design can achieve higher loading capacity when the outlet entropy loss distribution is similar. This provides a useful idea for the design of cantilevered stators.
- 24 February 2026
- Published by Global Power and Propulsion Society in Journal of the Global Power and Propulsion Society
- Vol. 10, 12-28
Open AccessEffects of blade loading distribution on hub leakage flow of cantilevered stator
The hub leakage flow of cantilevered stators has a significant impact and aerodynamic stability of the axial compressor. The formation and development of the hub leakage flow are affected by the blade loading level and distribution, but the internal mechanism of this effect is not fully understood in previous studies. This paper uses a novel experiment facility with a high-speed moving endwall and numerical simulation methods to focus on the influence of a fore-loading and highly-loaded cantilevered cascade on the hub leakage flow and its mechanism, and compared with a typical middle-loading controlled diffusion airfoil cascade. The results demonstrate that although the two cascades have different blade loading levels, the radial distribution of the entropy loss coefficient at the condition of moving endwall is almost identical, which is related to the development location and radial extent of the leakage flow in the two cascades. The position of maximum hub leakage flow rate is delayed relative to the position of maximum blade loading, due to the increase of circumferential momentum by the leakage vortex in the passage. The main cause for the entropy loss is the coupling effect between the product of the leakage flow velocity and injected angle and the mainstream velocity. Based on the findings of this paper, for the hub airfoil of the cantilevered stator, the fore-loading airfoil design can achieve higher loading capacity when the outlet entropy loss distribution is similar. This provides a useful idea for the design of cantilevered stators.
- 30 January 2026
- Published by Elsevier in Aerospace Science and Technology
- Vol. 173
Design optimization methods for a compressor rotor under inlet distortion
- 30 January 2026
- Published by Elsevier in Aerospace Science and Technology
- Vol. 173
Design optimization methods for a compressor rotor under inlet distortion
- 14 January 2026
- Published by Springer Nature
- p. 255-269
Revisiting Symmetric Teacher-Student Network Distillation for Anomaly Detection
- 14 January 2026
- Published by Springer Nature
- p. 255-269
Revisiting Symmetric Teacher-Student Network Distillation for Anomaly Detection
- 8 January 2026
- Published by Elsevier in Aerospace Science and Technology
- Vol. 171
Aerodynamic performance in a boundary layer ingesting fan with swirl distortion
- 8 January 2026
- Published by Elsevier in Aerospace Science and Technology
- Vol. 171
Aerodynamic performance in a boundary layer ingesting fan with swirl distortion
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 7 (9) , 1-14
ASBench: Image Anomalies Synthesis Benchmark for Anomaly Detection
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Artificial Intelligence
- Vol. 7 (9) , 1-14
ASBench: Image Anomalies Synthesis Benchmark for Anomaly Detection
- 17 December 2025
- Published by Elsevier in Aerospace Science and Technology
- Vol. 170
Comparative study of neural structure paradigms for compressor cascade flow field modeling based on the condition-encoder
- 17 December 2025
- Published by Elsevier in Aerospace Science and Technology
- Vol. 170
Comparative study of neural structure paradigms for compressor cascade flow field modeling based on the condition-encoder