Shuo Pan
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
Shuo Pan has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- Hai Yan
College of Metropolitan Transportation, Beijing University of Technology, Beijing 100124, China
5 shared publications Hai Yan
2 shared publicationsMingyang Hao
2 shared publicationsNing Chen
2 shared publications- Yue Hou
Department of Civil Engineering, Faculty of Science and Engineering, Swansea University, Swansea, UK
2 shared publications - Anand Sreeram
Faculty of Engineering, Nottingham Transportation Engineering Centre (NTEC), University of Nottingham, Nottingham, UK
1 shared publication
Latest Publications
- 1 March 2027
- Published by American Society of Civil Engineers (ASCE) in Journal of Transportation Engineering, Part B: Pavements
- Vol. 153 (1)
A YOLOv8-DTG-Based Framework for Pavement Distress Detection
Pavement condition is critical to traffic safety and pavement service life. Existing pavement distress detection algorithms are plagued by challenges such as low detection accuracy and inefficiency. To address these issues, this study proposes YOLOv8-DTG, a lightweight pavement distress detection algorithm based on the YOLOv8 framework. Specifically, dynamic snake convolution is integrated into the initial stages of the network, and the triplet attention mechanism is introduced to enhance feature extraction capabilities while minimizing the number of model parameters. Additionally, GSConv and VoV-GSCSP modules are introduced to reduce computational complexity and enhance feature interaction. Experimental results on the RDD 2022 dataset demonstrate that YOLOv8-DTG achieves an mAP50 of 0.729, representing an absolute improvement of 2.1 percentage points over the baseline YOLOv8 model. Additionally, YOLOv8-DTG reduces both model parameters and computational load. This study provides promising application prospects for real-time pavement distress detection in practical road maintenance scenarios.
- 1 March 2027
- Published by American Society of Civil Engineers (ASCE) in Journal of Transportation Engineering, Part B: Pavements
- Vol. 153 (1)
A YOLOv8-DTG-Based Framework for Pavement Distress Detection
Pavement condition is critical to traffic safety and pavement service life. Existing pavement distress detection algorithms are plagued by challenges such as low detection accuracy and inefficiency. To address these issues, this study proposes YOLOv8-DTG, a lightweight pavement distress detection algorithm based on the YOLOv8 framework. Specifically, dynamic snake convolution is integrated into the initial stages of the network, and the triplet attention mechanism is introduced to enhance feature extraction capabilities while minimizing the number of model parameters. Additionally, GSConv and VoV-GSCSP modules are introduced to reduce computational complexity and enhance feature interaction. Experimental results on the RDD 2022 dataset demonstrate that YOLOv8-DTG achieves an mAP50 of 0.729, representing an absolute improvement of 2.1 percentage points over the baseline YOLOv8 model. Additionally, YOLOv8-DTG reduces both model parameters and computational load. This study provides promising application prospects for real-time pavement distress detection in practical road maintenance scenarios.
- Open Access
- 1 May 2026
- Published by KEAI Publishing in High-speed Railway
Intelligent monitoring of longitudinal rail force using time series forecasting large language models
- 1 May 2026
- Published by KEAI Publishing in High-speed Railway
Open AccessIntelligent monitoring of longitudinal rail force using time series forecasting large language models
- Open Access
- 13 January 2026
- Published by MDPI AG in Inventions
- Vol. 11 (1) , 9
Optimization of Electric Bus Charging and Fleet Sizing Incorporating Traffic Congestion Based on Deep Reinforcement Learning
Amid the increasing demand to reduce carbon emissions, replacing diesel buses with electric buses has become a key development direction in public transportation. However, a significant challenge in this transition lies in developing efficient charging strategies and accurately determining the required fleet size, as existing research often fails to adequately account for the impact of real-time traffic congestion on energy consumption. To address this gap, in this study, an optimized charging strategy is proposed, and the necessary fleet size is calculated using a deep reinforcement learning (DRL) approach, which integrates actual route characteristics and dynamic traffic congestion patterns into an electric bus operation model. Modeling is conducted based on Beijing Bus Route 400 to ensure the practical applicability of the proposed method. The results demonstrate that the proposed DRL method ensures operational completion while minimizing charging time, with the algorithm showing rapid and stable convergence. In the multi-route scenarios investigated in this study, the DRL-based charging strategy requires 40% more electric buses, with this figure decreasing to 24% when fast-charging technology is adopted. This study provides bus companies with valuable electric bus procurement and route operation references.
- 13 January 2026
- Published by MDPI AG in Inventions
- Vol. 11 (1) , 9
Open AccessOptimization of Electric Bus Charging and Fleet Sizing Incorporating Traffic Congestion Based on Deep Reinforcement Learning
Amid the increasing demand to reduce carbon emissions, replacing diesel buses with electric buses has become a key development direction in public transportation. However, a significant challenge in this transition lies in developing efficient charging strategies and accurately determining the required fleet size, as existing research often fails to adequately account for the impact of real-time traffic congestion on energy consumption. To address this gap, in this study, an optimized charging strategy is proposed, and the necessary fleet size is calculated using a deep reinforcement learning (DRL) approach, which integrates actual route characteristics and dynamic traffic congestion patterns into an electric bus operation model. Modeling is conducted based on Beijing Bus Route 400 to ensure the practical applicability of the proposed method. The results demonstrate that the proposed DRL method ensures operational completion while minimizing charging time, with the algorithm showing rapid and stable convergence. In the multi-route scenarios investigated in this study, the DRL-based charging strategy requires 40% more electric buses, with this figure decreasing to 24% when fast-charging technology is adopted. This study provides bus companies with valuable electric bus procurement and route operation references.
- Open Access
- 1 January 2026
- Published by Maximum Academic Press in Digital Transportation and Safety
- Vol. 5 (3) , 248-254
Vehicle trajectory prediction using a kinematic feature enhanced-Transformer model
This study focuses on vehicle trajectory prediction on highways and urban expressways with the aim of enhancing traffic safety. Leveraging the CitySim dataset, three critical scenarios are examined: basic highway segments, highway merge/diverge areas, and urban expressway weaving zones. For each scenario, Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Graph Neural Network (GNN), and Transformer models are employed to predict trajectories 5 s ahead using 5 s of historical data. To improve prediction performance, this study proposes key enhancements centered on kinematic feature optimization, culminating in the development of the Kinematic Feature Enhanced-Transformer (KF-Transformer) model. The first enhancement modifies the attention mechanism to a relative attention framework incorporating position bias terms. The second replaces the feedforward network with a convolution-enhanced variant that integrates depthwise separable and dilated convolutions. Comparative experimental results demonstrate that the KF-Transformer model achieves superior performance, with Root Mean Square Error (RMSE) values of 1.93, 2.07, and 2.33 m recorded for basic highway segments, highway merge/diverge areas, and urban expressway weaving zones, respectively. The proposed model exhibits promising potential for application in Advanced Driver-Assistance Systems (ADAS), enabling proactive risk anticipation and effective accident mitigation.
- 1 January 2026
- Published by Maximum Academic Press in Digital Transportation and Safety
- Vol. 5 (3) , 248-254
Open AccessVehicle trajectory prediction using a kinematic feature enhanced-Transformer model
This study focuses on vehicle trajectory prediction on highways and urban expressways with the aim of enhancing traffic safety. Leveraging the CitySim dataset, three critical scenarios are examined: basic highway segments, highway merge/diverge areas, and urban expressway weaving zones. For each scenario, Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Graph Neural Network (GNN), and Transformer models are employed to predict trajectories 5 s ahead using 5 s of historical data. To improve prediction performance, this study proposes key enhancements centered on kinematic feature optimization, culminating in the development of the Kinematic Feature Enhanced-Transformer (KF-Transformer) model. The first enhancement modifies the attention mechanism to a relative attention framework incorporating position bias terms. The second replaces the feedforward network with a convolution-enhanced variant that integrates depthwise separable and dilated convolutions. Comparative experimental results demonstrate that the KF-Transformer model achieves superior performance, with Root Mean Square Error (RMSE) values of 1.93, 2.07, and 2.33 m recorded for basic highway segments, highway merge/diverge areas, and urban expressway weaving zones, respectively. The proposed model exhibits promising potential for application in Advanced Driver-Assistance Systems (ADAS), enabling proactive risk anticipation and effective accident mitigation.
- Open Access
Vibration Analysis and Vehicle Detection by MEMS Acceleration Sensors Embedded in PCC Pavement
Monitoring the vibration response of Portland cement concrete (PCC) pavement under dynamic vehicle loading is critical for road maintenance and traffic analysis. This study embedded micro-electro-mechanical systems (MEMS) accelerometer sensors in PCC pavement to capture vibration signals induced by vehicles. A thresholding method is proposed to automate vehicle detection by analyzing acceleration time-domain data, achieving precision and recall rates exceeding 85%. The study also explored various sensor placement locations and different threshold values for acceleration time-domain signals. Sensor placement optimization revealed that positioning sensors at the front or rear ends of pavement slabs maximizes vibration response, enabling low-cost and efficient detection. Experimental results demonstrated that the proposed method balances simplicity and accuracy, eliminating the need for complex denoising processes. This approach provides a cost-effective solution for real-time vehicle detection and enhances pavement performance monitoring, supporting improved maintenance and traffic management strategies.
Open AccessVibration Analysis and Vehicle Detection by MEMS Acceleration Sensors Embedded in PCC Pavement
Monitoring the vibration response of Portland cement concrete (PCC) pavement under dynamic vehicle loading is critical for road maintenance and traffic analysis. This study embedded micro-electro-mechanical systems (MEMS) accelerometer sensors in PCC pavement to capture vibration signals induced by vehicles. A thresholding method is proposed to automate vehicle detection by analyzing acceleration time-domain data, achieving precision and recall rates exceeding 85%. The study also explored various sensor placement locations and different threshold values for acceleration time-domain signals. Sensor placement optimization revealed that positioning sensors at the front or rear ends of pavement slabs maximizes vibration response, enabling low-cost and efficient detection. Experimental results demonstrated that the proposed method balances simplicity and accuracy, eliminating the need for complex denoising processes. This approach provides a cost-effective solution for real-time vehicle detection and enhances pavement performance monitoring, supporting improved maintenance and traffic management strategies.
- 1 April 2025
- Published by Emerald Publishing in Proceedings of the Institution of Civil Engineers - Engineering Sustainability
- Vol. 178 (2) , 124-134
Sustainable utilization of road assets concerning obscured traffic signs recognition
Traffic sign recognition is crucial for sustainable utilization of road assets. However, on real roads, traffic signs are often obscured, making their recognition challenging. Unfortunately, existing research lacks specific analysis of the impact of traffic sign occlusion and methods to improve recognition in such scenarios. This study quantitatively investigates the relationship between the degree of traffic sign occlusion and recognition accuracy. Moreover, a dedicated deep learning model is proposed, utilizing a multi-scale convolutional stacked input layer, to enhance the recognition of obscured traffic signs. Using the Chinese Traffic Sign Recognition Database, this study analyzes the recognition performance of four traffic sign categories: indication signs, prohibition signs, speed limit signs, and warning signs, under different occlusion levels. Three widely used deep learning methods, Visual Geometry Group, Residual Network, and Dense Convolutional Network, are compared with the dedicated model. Experimental results demonstrate a significant decrease in recognition accuracy when traffic signs are obscured. Importantly, the proposed dedicated model outperforms the other methods, achieving accuracies of 80.95%, 90.50%, and 97.22% in the scenarios of 50% occlusion, 25% occlusion, and no occlusion, respectively. This study holds implications for sustainable utilization of road assets.
- 1 April 2025
- Published by Emerald Publishing in Proceedings of the Institution of Civil Engineers - Engineering Sustainability
- Vol. 178 (2) , 124-134
Sustainable utilization of road assets concerning obscured traffic signs recognition
Traffic sign recognition is crucial for sustainable utilization of road assets. However, on real roads, traffic signs are often obscured, making their recognition challenging. Unfortunately, existing research lacks specific analysis of the impact of traffic sign occlusion and methods to improve recognition in such scenarios. This study quantitatively investigates the relationship between the degree of traffic sign occlusion and recognition accuracy. Moreover, a dedicated deep learning model is proposed, utilizing a multi-scale convolutional stacked input layer, to enhance the recognition of obscured traffic signs. Using the Chinese Traffic Sign Recognition Database, this study analyzes the recognition performance of four traffic sign categories: indication signs, prohibition signs, speed limit signs, and warning signs, under different occlusion levels. Three widely used deep learning methods, Visual Geometry Group, Residual Network, and Dense Convolutional Network, are compared with the dedicated model. Experimental results demonstrate a significant decrease in recognition accuracy when traffic signs are obscured. Importantly, the proposed dedicated model outperforms the other methods, achieving accuracies of 80.95%, 90.50%, and 97.22% in the scenarios of 50% occlusion, 25% occlusion, and no occlusion, respectively. This study holds implications for sustainable utilization of road assets.
- Open Access
- 1 January 2025
- Published by Maximum Academic Press in Digital Transportation and Safety
- Vol. 4 (1) , 1-9
Bus system optimization for timetables, routes, charging, and facilities: a summary
Bus systems plays an important role in reducing urban traffic congestion, developing green transportation, and providing more equitable transportation services. However, the rising demand for service quality from passengers and the large number of new energy buses have posed new challenges to the operation of bus systems. Integrating resources and optimizing bus systems is a challenge that requires collaboration between the government and bus operators to find effective solutions. This paper reviews the research related to bus system optimization from four aspects: bus timetable optimization, bus route optimization, bus charging optimization, and bus facility optimization. For bus timetable optimization, most of the existing research takes the minimum cost, the number of vehicles, and the maximum passenger capacity as the optimization objectives, which are solved by classical statistical methods and deep learning methods. For bus route optimization, existing research mostly focuses on different scenarios such as customized buses and autonomous buses, and uses heuristic algorithms and exact algorithms to analyze them with the objectives of cost, travel time, and carbon emission minimization. For bus charging optimization, integer programming models and heuristic algorithms are typically applied to address the optimization problems of charging stations and charging schedules, with minimum charging cost and charging time as the optimization objectives. For bus facility optimization, the research is mostly carried out from the optimization of bus stops and bus lanes. This paper can provide suggestions for the optimization and development of bus systems.
- 1 January 2025
- Published by Maximum Academic Press in Digital Transportation and Safety
- Vol. 4 (1) , 1-9
Open AccessBus system optimization for timetables, routes, charging, and facilities: a summary
Bus systems plays an important role in reducing urban traffic congestion, developing green transportation, and providing more equitable transportation services. However, the rising demand for service quality from passengers and the large number of new energy buses have posed new challenges to the operation of bus systems. Integrating resources and optimizing bus systems is a challenge that requires collaboration between the government and bus operators to find effective solutions. This paper reviews the research related to bus system optimization from four aspects: bus timetable optimization, bus route optimization, bus charging optimization, and bus facility optimization. For bus timetable optimization, most of the existing research takes the minimum cost, the number of vehicles, and the maximum passenger capacity as the optimization objectives, which are solved by classical statistical methods and deep learning methods. For bus route optimization, existing research mostly focuses on different scenarios such as customized buses and autonomous buses, and uses heuristic algorithms and exact algorithms to analyze them with the objectives of cost, travel time, and carbon emission minimization. For bus charging optimization, integer programming models and heuristic algorithms are typically applied to address the optimization problems of charging stations and charging schedules, with minimum charging cost and charging time as the optimization objectives. For bus facility optimization, the research is mostly carried out from the optimization of bus stops and bus lanes. This paper can provide suggestions for the optimization and development of bus systems.
- 25 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 25 (12) , 20793-20803
HSRA-Net: Intelligent Detection Network of Anomaly Monitoring Data in High-Speed Railway
Real-time monitoring and analysis of sensitive areas in high-speed railway (HSR) are crucial for ensuring the safe and smooth operation of high-speed trains. To address the problem of frequent missing and false alarms caused by anomaly data in HSR monitoring system, this study proposes an innovative network framework: Intelligent detection network of HSR anomaly monitoring data (HSRA-Net). The framework comprises of two modules: the data augmentation module and the anomaly detection module. The data augmentation module designs multiple alternative generative adversarial networks for sample augmentation. To achieve the end-to-end classification, the anomaly detection module improves the residual network by creating a deep residual shrinkage network with self-attention (DRSN-SA). An online monitoring system was installed and operated continuously for several years on a high-speed turnout of a continuous beam bridge to validate the proposed framework. The collected data includes displacement, stress, and temperature. The proposed framework has demonstrated excellent performance, generalizability, and deployability through sufficient model comparison. It can accurately and efficiently diagnose anomalies in the operation of the monitoring system. This study is of great significance for improving the anomaly detection task of the HSR monitoring system.
- 25 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 25 (12) , 20793-20803
HSRA-Net: Intelligent Detection Network of Anomaly Monitoring Data in High-Speed Railway
Real-time monitoring and analysis of sensitive areas in high-speed railway (HSR) are crucial for ensuring the safe and smooth operation of high-speed trains. To address the problem of frequent missing and false alarms caused by anomaly data in HSR monitoring system, this study proposes an innovative network framework: Intelligent detection network of HSR anomaly monitoring data (HSRA-Net). The framework comprises of two modules: the data augmentation module and the anomaly detection module. The data augmentation module designs multiple alternative generative adversarial networks for sample augmentation. To achieve the end-to-end classification, the anomaly detection module improves the residual network by creating a deep residual shrinkage network with self-attention (DRSN-SA). An online monitoring system was installed and operated continuously for several years on a high-speed turnout of a continuous beam bridge to validate the proposed framework. The collected data includes displacement, stress, and temperature. The proposed framework has demonstrated excellent performance, generalizability, and deployability through sufficient model comparison. It can accurately and efficiently diagnose anomalies in the operation of the monitoring system. This study is of great significance for improving the anomaly detection task of the HSR monitoring system.
- Open Access
- 17 July 2023
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 381 (2254) , 20220166
Automatic pavement texture recognition using lightweight few-shot learning
Texture is a crucial characteristic of roads, closely related to their performance. The recognition of pavement texture is of great significance for road maintenance professionals to detect potential safety hazards and carry out necessary countermeasures. Although deep learning models have been applied for recognition, the scarcity of data has always been a limitation. To address this issue, this paper proposes a few-shot learning model based on the Siamese network for pavement texture recognition with a limited dataset. The model achieved 89.8% accuracy in a four-way five-shot task classifying the pavement textures of dense asphalt concrete, micro surface, open-graded friction course and stone matrix asphalt. To align with engineering practice, global average pooling (GAP) and one-dimensional convolution are implemented, creating lightweight models that save storage and training time. Comparative experiments show that the lightweight model with GAP implemented on dense layers and one-dimensional convolution on convolutional layers reduced storage volume by 94% and training time by 99%, despite a 2.9% decrease in classification accuracy. Moreover, the model with only GAP implemented on dense layers achieved the highest accuracy at 93.5%, while reducing storage volume and training time by 83% and 6%, respectively. This article is part of the theme issue 'Artificial intelligence in failure analysis of transportation infrastructure and materials'.
- 17 July 2023
- Published by The Royal Society in Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences
- Vol. 381 (2254) , 20220166
Open AccessAutomatic pavement texture recognition using lightweight few-shot learning
Texture is a crucial characteristic of roads, closely related to their performance. The recognition of pavement texture is of great significance for road maintenance professionals to detect potential safety hazards and carry out necessary countermeasures. Although deep learning models have been applied for recognition, the scarcity of data has always been a limitation. To address this issue, this paper proposes a few-shot learning model based on the Siamese network for pavement texture recognition with a limited dataset. The model achieved 89.8% accuracy in a four-way five-shot task classifying the pavement textures of dense asphalt concrete, micro surface, open-graded friction course and stone matrix asphalt. To align with engineering practice, global average pooling (GAP) and one-dimensional convolution are implemented, creating lightweight models that save storage and training time. Comparative experiments show that the lightweight model with GAP implemented on dense layers and one-dimensional convolution on convolutional layers reduced storage volume by 94% and training time by 99%, despite a 2.9% decrease in classification accuracy. Moreover, the model with only GAP implemented on dense layers achieved the highest accuracy at 93.5%, while reducing storage volume and training time by 83% and 6%, respectively. This article is part of the theme issue 'Artificial intelligence in failure analysis of transportation infrastructure and materials'.
- 1 January 2013
- Published by Springer Nature
- p. 1264-1267
The Application of Improved RG-based Wavefront Simulation Model on 3D Bifurcated Tube Segmentation in CT Images
- 1 January 2013
- Published by Springer Nature
- p. 1264-1267
The Application of Improved RG-based Wavefront Simulation Model on 3D Bifurcated Tube Segmentation in CT Images