Guangzhou Railway Polytechnic

During train operation, high-speed roof slipstreams exert strong shear on the airflow through the air-conditioning unit (ACU) condenser, reducing the condenser airflow rate. This decrease can trigger ACU overheating alarms or shutdowns, compromising both operational safety and passenger comfort. Conventional experimental techniques, however, struggle to accurately measure condenser airflow rate under such complex shear flow conditions. To overcome this limitation, we performed wind tunnel tests on a full-scale train ACU and developed a novel measurement strategy integrating particle image velocimetry, multiple field-of-view stitching, and airflow flux integral surface extraction. This approach enables, for the first time, reconstruction of the flow field in the ACU air-inlet region and yields high-precision measurement of condenser airflow rate under shear flow. The experiments show that increasing the incoming airflow speed from 15 to 20 m/s intensifies the shear effect, enhancing turbulent dissipation and pressure loss at the air inlets and reducing condenser airflow rate by 3.2%. Analysis further reveals that under the combined effects of shear flow and fan rotation, the windward side retains uniform flow, whereas the leeward side develops vortices and localized backflow. These effects produce a characteristic strip-like, asymmetric distribution of vertical velocity across the air inlets. This study provides valuable data on complex intake flow behavior under shear conditions and technical support for investigating airflow interactions in train ACUs and other ventilation systems.

Abstract This paper addresses the trajectory tracking problem of the 2‐DOF helicopter system by proposing an adaptive finite‐time control method based on a composite learning and switching mechanism. Unlike traditional learning control approaches, the composite learning method in this paper improves learning efficiency by introducing a prediction error term within the neural domain. Outside the neural domain, the switching mechanism ensures rapid convergence of the system state. Additionally, a barrier Lyapunov function (BLF) is employed to impose state constraints, ensuring that the system state remains within a safe range during the learning process. Through Lyapunov stability analysis, the finite‐time convergence and uniform ultimate boundedness of the system are proven. Finally, simulations and experiments validate that the proposed method effectively overcomes the challenges posed by model uncertainties and external disturbances.

This paper addresses the challenges of dynamic environments and multimodal data fusion in multimodal transport path optimization for smart ports by proposing a GL-SSL Model that integrates Graph Neural Networks (GCN), Long Short-Term Memory (LSTM), and Self-Supervised Learning (SSL). The model fully exploits the graph-structured information of port transport networks and their temporal variations, while SSL enhances feature representation, enabling efficient optimization of path planning. Experiments were conducted on multiple public datasets, including AIS data from the Port of Rotterdam, global shipping data, and port net revenue data. Results show that the GL-SSL Model achieved significant improvements in key performance metrics. Specifically, the optimized path length reached 80 km, the transport cost was reduced to 200 cost-units (a composite metric reflecting fuel consumption, equipment wear, and labor cost), and the delay rate was maintained at 0.05 (5%), all of which are substantially better than traditional algorithms and other deep learning models. Furthermore, the model demonstrated stable performance under complex scenarios such as peak traffic, adverse weather, and equipment failures, with rapid convergence of training loss and strong robustness. These findings highlight the model’s adaptability and practical application potential. Overall, this work provides effective technical support for multimodal transport path optimization in smart ports and carries important theoretical significance and broad application prospects.

Conventional commuting mode split models are characterized by inherent limitations in dynamic adaptability, primarily due to persistent dependence on periodic survey data with significant temporal gaps. A dominant transportation distance-based modeling framework for commuting mode choice is proposed, formalizing a generalized cost function. Through the application of random utility theory, probability density curves are generated to quantify mode-specific dominant distance ranges across three demographic groups: car-owning households, non-car households, and collective households. Empirical validation was conducted using Dongguan as a case study, with model parameters calibrated against 2015 resident travel survey data. Parameter updates are dynamically executed through the integration of big data sources (e.g., mobile signaling and LBS). Successful implementation has been achieved in maintaining Dongguan’s transportation models during the 2021 and 2023 iterations.

To address the challenges of manual dependency and complex management in railway data governance, and to promote data value realization, we propose DGMAS, a Data Governance Multi-Agent System based on Large Language Models (LLMs). The architectural design adopts a system comprising a mother system and three subsystems. The DGMAS comprises four critical phases: initialization, execution, re-planning, and failure attribution. Through this system, the process of railway data governance can be simulated, to help stakeholders anticipate potential issues and develop solutions, thereby supporting improved railway data governance.