Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2236240
Bing Liu , Yuxiong Ji , Oded Cats
Public transport (PT) agencies are increasingly keen on integrating ride-hailing (RH) services with PT to improve overall mobility. Understanding the traffic flow distribution in the integrated system is vital for the policy decision-making and services design of such a system. We propose a stochastic user equilibrium (SUE) model for multimodal transport systems consisting of private car, PT and RH. The travel costs in the SUE model are investigated using a multimodal graph representation to capture the relationship of different travel modes in the integrated system. We apply the proposed model to a toy case and a real-world case. A RH subsidy strategy is compared with the benchmark to demonstrate travellers’ route and mode shifts in the integrated system. Our findings offer insights on subsidising RH services through the proposed model, and provide valuable knowledge on the planning and design of the integrated system.
{"title":"Integrating ride-hailing services with public transport: a stochastic user equilibrium model for multimodal transport systems","authors":"Bing Liu , Yuxiong Ji , Oded Cats","doi":"10.1080/23249935.2023.2236240","DOIUrl":"10.1080/23249935.2023.2236240","url":null,"abstract":"<div><div>Public transport (PT) agencies are increasingly keen on integrating ride-hailing (RH) services with PT to improve overall mobility. Understanding the traffic flow distribution in the integrated system is vital for the policy decision-making and services design of such a system. We propose a stochastic user equilibrium (SUE) model for multimodal transport systems consisting of private car, PT and RH. The travel costs in the SUE model are investigated using a multimodal graph representation to capture the relationship of different travel modes in the integrated system. We apply the proposed model to a toy case and a real-world case. A RH subsidy strategy is compared with the benchmark to demonstrate travellers’ route and mode shifts in the integrated system. Our findings offer insights on subsidising RH services through the proposed model, and provide valuable knowledge on the planning and design of the integrated system.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 329-357"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"47280471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2239375
HongSheng Qi
A stochastic lateral movement model is proposed to address the limitations of current traffic models, which fail to capture the stochastic nature of the lateral component in vehicle movement during lane keeping and lane changing. This model incorporates a lateral noise component and a lateral movement component, with parameters that have clear physical interpretations including noise intensity, driver’s sensitivity to lateral deviation, and sensitivity to noise. The model successfully describes the real-world distribution and standard deviation of lateral displacement, achieves over 70% accuracy in distinguishing between human driven vehicles and autonomous vehicles, derives the lane changing duration distribution consistent with experimental observation, and shows that the sensitivity to lateral deviation is about 7 times higher in lane changing compared to lane keeping.
{"title":"Modelling the lateral dimension of vehicles movement: a stochastic differential approach with applications","authors":"HongSheng Qi","doi":"10.1080/23249935.2023.2239375","DOIUrl":"10.1080/23249935.2023.2239375","url":null,"abstract":"<div><div>A stochastic lateral movement model is proposed to address the limitations of current traffic models, which fail to capture the stochastic nature of the lateral component in vehicle movement during lane keeping and lane changing. This model incorporates a lateral noise component and a lateral movement component, with parameters that have clear physical interpretations including noise intensity, driver’s sensitivity to lateral deviation, and sensitivity to noise. The model successfully describes the real-world distribution and standard deviation of lateral displacement, achieves over 70% accuracy in distinguishing between human driven vehicles and autonomous vehicles, derives the lane changing duration distribution consistent with experimental observation, and shows that the sensitivity to lateral deviation is about 7 times higher in lane changing compared to lane keeping.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 411-435"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"59991235","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2236236
Jana Vacková , Marek Bukáček
The standard definition of pedestrian density produces scattered values, hence, many approaches have been developed to improve the features of the estimated density. This paper provides a review of generally applied methods and presents a general framework based on various kernels that bring desired properties of density estimates (e.g. continuity) and incorporate ordinarily used methods. The developed kernel concept considers each pedestrian as a source of density distribution, parametrised by the kernel type (e.g. Gauss, cone) and kernel size. The quantitative parametric study performed on experimental data illustrates that parametrisation brings desired features, for instance, a conic kernel with a base radius in $ (0.7, 1.2) $ m produces smooth values that retain trend features. The correspondence between kernel and non-kernel methods (namely Voronoi diagram and customised inverse distance to the nearest pedestrian) is achievable for a wide range of kernel parameter. Thereby the generality of the concept is supported.
{"title":"Kernel estimates as general concept for the measuring of pedestrian density","authors":"Jana Vacková , Marek Bukáček","doi":"10.1080/23249935.2023.2236236","DOIUrl":"10.1080/23249935.2023.2236236","url":null,"abstract":"<div><div>The standard definition of pedestrian density produces scattered values, hence, many approaches have been developed to improve the features of the estimated density. This paper provides a review of generally applied methods and presents a general framework based on various kernels that bring desired properties of density estimates (e.g. continuity) and incorporate ordinarily used methods. The developed kernel concept considers each pedestrian as a source of density distribution, parametrised by the kernel type (e.g. Gauss, cone) and kernel size. The quantitative parametric study performed on experimental data illustrates that parametrisation brings desired features, for instance, a conic kernel with a base radius in $ (0.7, 1.2) $ <span><math><mo>(</mo><mn>0.7</mn><mo>,</mo><mn>1.2</mn><mo>)</mo></math></span> m produces smooth values that retain trend features. The correspondence between kernel and non-kernel methods (namely Voronoi diagram and customised inverse distance to the nearest pedestrian) is achievable for a wide range of kernel parameter. Thereby the generality of the concept is supported.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 303-328"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81006516","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2226245
Renxin Zhong , Xin-an Li , Qingnan Liang , Zhibin Chen , Tianlu Pan
This paper proposes a projected Newton-like inertial dynamics for modeling second-order day-to-day (DTD) traffic evolution with elastic travel demand. The proposed DTD model describes double dynamics of traffic flow and travel cost based on a class of second-order gradient-like dissipative dynamic systems. We use the projection operator to prevent the existence of negative flow, which is regarded as a major pitfall of the existing second-order DTD traffic models. To our knowledge, this would be the first attempt to address the problem of negative flow in the second-order DTD traffic models. Meanwhile, we show that the proposed model inherits the properties of Newton-like inertial dynamics and behaves similarly to the existing second-order DTD models. The proposed model admits a Hessian-driven component, which is closely related to the congestion externality associated with the marginal link travel cost. The proposed model also extends the existing second-order DTD models from the fixed demand case to the elastic demand case. We characterize several theoretical properties of the proposed projected second-order DTD model, such as the equivalence between its fixed points and the user equilibrium with elastic demand, the convergence of the DTD traffic evolution process, and the stability analysis with different stability concepts. We show that the proposed model can be reduced to the well-known network tatonnement model. Finally, we demonstrate the properties of the projected second-order DTD model via numerical examples.
{"title":"A projected Newton-like inertial dynamics for modeling day-to-day traffic evolution with elastic demand","authors":"Renxin Zhong , Xin-an Li , Qingnan Liang , Zhibin Chen , Tianlu Pan","doi":"10.1080/23249935.2023.2226245","DOIUrl":"10.1080/23249935.2023.2226245","url":null,"abstract":"<div><div>This paper proposes a projected Newton-like inertial dynamics for modeling second-order day-to-day (DTD) traffic evolution with elastic travel demand. The proposed DTD model describes double dynamics of traffic flow and travel cost based on a class of second-order gradient-like dissipative dynamic systems. We use the projection operator to prevent the existence of negative flow, which is regarded as a major pitfall of the existing second-order DTD traffic models. To our knowledge, this would be the first attempt to address the problem of negative flow in the second-order DTD traffic models. Meanwhile, we show that the proposed model inherits the properties of Newton-like inertial dynamics and behaves similarly to the existing second-order DTD models. The proposed model admits a Hessian-driven component, which is closely related to the congestion externality associated with the marginal link travel cost. The proposed model also extends the existing second-order DTD models from the fixed demand case to the elastic demand case. We characterize several theoretical properties of the proposed projected second-order DTD model, such as the equivalence between its fixed points and the user equilibrium with elastic demand, the convergence of the DTD traffic evolution process, and the stability analysis with different stability concepts. We show that the proposed model can be reduced to the well-known network tatonnement model. Finally, we demonstrate the properties of the projected second-order DTD model via numerical examples.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 101-129"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48792035","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2232047
Yuanyuan Wu , David Z. W. Wang , Feng Zhu
Autonomous Intersection Management (AIM) for high-level Connected and Automated Vehicles (CAVs) has evolved from rule-based to optimisation-based policies. However, at congested major-minor intersections, optimising solely for efficiency can negatively impact vehicle fairness. This study addresses this issue by proposing a deep reinforcement learning approach that optimises both traffic efficiency and fairness for AIM. In the modelled multi-objective Markov decision process, traffic fairness is measured by the difference between the crossing order and the approaching order of CAVs, while traffic efficiency is measured by average travel time. With unknown preferences of the objectives, Bellman optimality equation is generalised to obtain the optimal policies over the space of all possible preferences during the iterative training process. The effectiveness of the proposed method is evaluated in a simulated real-world intersection and compared with three benchmark policies, including the fairest policy for AIM: first-come-first-served. The learned policies perform best in reducing overall average vehicle delay, and demonstrate outstanding performance in balancing traffic fairness and efficiency.
{"title":"Traffic efficiency and fairness optimisation for autonomous intersection management based on reinforcement learning","authors":"Yuanyuan Wu , David Z. W. Wang , Feng Zhu","doi":"10.1080/23249935.2023.2232047","DOIUrl":"10.1080/23249935.2023.2232047","url":null,"abstract":"<div><div>Autonomous Intersection Management (AIM) for high-level Connected and Automated Vehicles (CAVs) has evolved from rule-based to optimisation-based policies. However, at congested major-minor intersections, optimising solely for efficiency can negatively impact vehicle fairness. This study addresses this issue by proposing a deep reinforcement learning approach that optimises both traffic efficiency and fairness for AIM. In the modelled multi-objective Markov decision process, traffic fairness is measured by the difference between the crossing order and the approaching order of CAVs, while traffic efficiency is measured by average travel time. With unknown preferences of the objectives, Bellman optimality equation is generalised to obtain the optimal policies over the space of all possible preferences during the iterative training process. The effectiveness of the proposed method is evaluated in a simulated real-world intersection and compared with three benchmark policies, including the fairest policy for AIM: first-come-first-served. The learned policies perform best in reducing overall average vehicle delay, and demonstrate outstanding performance in balancing traffic fairness and efficiency.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 247-271"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"41454744","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2236719
Zhenjie Zheng , Zhengli Wang , Xiqun Chen , Wei Ma , Bin Ran
Traffic incidents disrupt the normal flow of vehicles and induce nonrecurrent traffic congestion. It has been well accepted that the shape of the spatiotemporal region impacted by a traffic incident should be consistent with the propagation of shockwaves. Although there has been a variety of approaches that attempt to estimate the impact region of traffic incidents, most of them are not capable of producing results with guaranteed consistency. In this research, we propose an improved fuzzy clustering approach that integrates the domain knowledge of shockwave theory for freeway incidents to address this issue, which is new to the literature. Compared to the general clustering approaches, our improved fuzzy clustering approach takes control of the clustering process by leveraging the directional propagation of shockwaves in the form of constraints, which can guarantee the consistency. In addition, unlike existing studies that employ discrete variables to distinguish traffic status in case of traffic incidents, the fuzzy clustering approach uses the continuous variable to indicate the incident impact on vehicle speed. This can help to reduce the information loss and estimate the impact region more accurately. Numerical experiments are conducted to evaluate the performance of our approach using both simulation and real data. Results show that our approach is able to guarantee that the shape of the impact region is consistent with the propagation of shockwaves and achieve higher accuracy of the estimated delay induced by the incident than the current state-of-the-art approach.
{"title":"Spatiotemporal clustering for the impact region caused by a traffic incident: an improved fuzzy C-means approach with guaranteed consistency","authors":"Zhenjie Zheng , Zhengli Wang , Xiqun Chen , Wei Ma , Bin Ran","doi":"10.1080/23249935.2023.2236719","DOIUrl":"10.1080/23249935.2023.2236719","url":null,"abstract":"<div><div>Traffic incidents disrupt the normal flow of vehicles and induce nonrecurrent traffic congestion. It has been well accepted that the shape of the spatiotemporal region impacted by a traffic incident should be consistent with the propagation of shockwaves. Although there has been a variety of approaches that attempt to estimate the impact region of traffic incidents, most of them are not capable of producing results with guaranteed consistency. In this research, we propose an improved fuzzy clustering approach that integrates the domain knowledge of shockwave theory for freeway incidents to address this issue, which is new to the literature. Compared to the general clustering approaches, our improved fuzzy clustering approach takes control of the clustering process by leveraging the directional propagation of shockwaves in the form of constraints, which can guarantee the consistency. In addition, unlike existing studies that employ discrete variables to distinguish traffic status in case of traffic incidents, the fuzzy clustering approach uses the continuous variable to indicate the incident impact on vehicle speed. This can help to reduce the information loss and estimate the impact region more accurately. Numerical experiments are conducted to evaluate the performance of our approach using both simulation and real data. Results show that our approach is able to guarantee that the shape of the impact region is consistent with the propagation of shockwaves and achieve higher accuracy of the estimated delay induced by the incident than the current state-of-the-art approach.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 358-387"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"48699623","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2220423
Yaoyao Wang , Avishai (Avi) Ceder , Zhichao Cao , Silin Zhang
Fluctuating demand for public transport (PT) is one of the main reasons for unreliable PT service, and subsequent passenger frustration at being left behind at PT stops. A novel way to solve this situation is to optimally use autonomous PT vehicles with coupling and decoupling (C&D) of vehicle units to accommodate the fluctuating PT demand and reliability issues. In this way, vehicle size is added as a variable of the problem. This work proposes a new class of C&D tactics in the process of solving the problems of PT route timetabling subject to passenger demand. Resolving the optimisation problem involves determining the C&D arrangement at stops/stations to accommodate the C&D options and departure times. The validation of the model is performed by a small example and a real case study with a bilevel heuristic algorithm that manages to completely (100%) eliminate left-behind passengers using practical, even-headway, and even-load timetables.
{"title":"Optimal public transport timetabling with autonomous-vehicle units using coupling and decoupling tactics","authors":"Yaoyao Wang , Avishai (Avi) Ceder , Zhichao Cao , Silin Zhang","doi":"10.1080/23249935.2023.2220423","DOIUrl":"10.1080/23249935.2023.2220423","url":null,"abstract":"<div><div>Fluctuating demand for public transport (PT) is one of the main reasons for unreliable PT service, and subsequent passenger frustration at being left behind at PT stops. A novel way to solve this situation is to optimally use autonomous PT vehicles with coupling and decoupling (C&D) of vehicle units to accommodate the fluctuating PT demand and reliability issues. In this way, vehicle size is added as a variable of the problem. This work proposes a new class of C&D tactics in the process of solving the problems of PT route timetabling subject to passenger demand. Resolving the optimisation problem involves determining the C&D arrangement at stops/stations to accommodate the C&D options and departure times. The validation of the model is performed by a small example and a real case study with a bilevel heuristic algorithm that manages to completely (100%) eliminate left-behind passengers using practical, even-headway, and even-load timetables.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 50-100"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"46728619","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2215338
Paul (Young Joun) Ha , Sikai Chen , Jiqian Dong , Samuel Labi
Automation and connectivity based platforms have great potential for managing highway traffic congestion including bottlenecks. Speed harmonisation (SH), one of such platforms, is an Active Traffic Management (ATM) strategy that addresses flow breakdown in real-time by adjusting upstream traffic speeds. However, SH has limitations including the need for supporting roadway infrastructure that is immovable and has limited coverage; the inability to enact control beyond its range; and the dependence on human driver compliance. These issues could be addressed by leveraging connected and automated vehicles (CAVs), which can collect information and execute control along their trajectories, irrespective of drivers’ awareness or compliance. In addressing this objective, this study utilises reinforcement learning to present a CAV control model to achieve efficient speed harmonisation. The results suggest that even at low market penetration, CAVs can significantly mitigate traffic congestion bottlenecks to a greater extent compared to traditional SH approaches.
{"title":"Leveraging vehicle connectivity and autonomy for highway bottleneck congestion mitigation using reinforcement learning","authors":"Paul (Young Joun) Ha , Sikai Chen , Jiqian Dong , Samuel Labi","doi":"10.1080/23249935.2023.2215338","DOIUrl":"10.1080/23249935.2023.2215338","url":null,"abstract":"<div><div>Automation and connectivity based platforms have great potential for managing highway traffic congestion including bottlenecks. Speed harmonisation (SH), one of such platforms, is an Active Traffic Management (ATM) strategy that addresses flow breakdown in real-time by adjusting upstream traffic speeds. However, SH has limitations including the need for supporting roadway infrastructure that is immovable and has limited coverage; the inability to enact control beyond its range; and the dependence on human driver compliance. These issues could be addressed by leveraging connected and automated vehicles (CAVs), which can collect information and execute control along their trajectories, irrespective of drivers’ awareness or compliance. In addressing this objective, this study utilises reinforcement learning to present a CAV control model to achieve efficient speed harmonisation. The results suggest that even at low market penetration, CAVs can significantly mitigate traffic congestion bottlenecks to a greater extent compared to traditional SH approaches.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 1-26"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"59991171","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2239377
Bi Yu Chen , Yaohong Ma , Jiale Wang , Tao Jia , Xianglong Liu , William H. K. Lam
How to select a suitable spatial weighting scheme for convolutional graph neural networks (ConvGNNs) is challenging. In this study, we propose a ConvGNN, termed learnable graph convolutional (LGC) network, which learns spatial weightings between a road and its k-hop neighbours as learnable parameters in the spatial convolutional operator. A dynamic LGC (DLGC) network is further proposed to learn the dynamics of spatial weightings by explicitly considering the temporal correlations of spatial weightings at different times of the day. A multi-temporal DLGC (MTDLGC) network is developed for forecasting traffic variables in road networks. Results of case study suggest that the MT-DLGC network can achieve higher prediction accuracy than other state-of-the-art baselines. Both LGC and DLGC networks can be used as general spatial weighting schemes for baselines with better forecasting performance than existing spatial weighting schemes, e.g., graph attention. The source code of this study is available publicly at https://github.com/Mayaohong/MTDLGC.
{"title":"Graph convolutional networks with learnable spatial weightings for traffic forecasting applications","authors":"Bi Yu Chen , Yaohong Ma , Jiale Wang , Tao Jia , Xianglong Liu , William H. K. Lam","doi":"10.1080/23249935.2023.2239377","DOIUrl":"10.1080/23249935.2023.2239377","url":null,"abstract":"<div><div>How to select a suitable spatial weighting scheme for convolutional graph neural networks (ConvGNNs) is challenging. In this study, we propose a ConvGNN, termed learnable graph convolutional (LGC) network, which learns spatial weightings between a road and its k-hop neighbours as learnable parameters in the spatial convolutional operator. A dynamic LGC (DLGC) network is further proposed to learn the dynamics of spatial weightings by explicitly considering the temporal correlations of spatial weightings at different times of the day. A multi-temporal DLGC (MTDLGC) network is developed for forecasting traffic variables in road networks. Results of case study suggest that the MT-DLGC network can achieve higher prediction accuracy than other state-of-the-art baselines. Both LGC and DLGC networks can be used as general spatial weighting schemes for baselines with better forecasting performance than existing spatial weighting schemes, e.g., graph attention. The source code of this study is available publicly at <span>https://github.com/Mayaohong/MTDLGC</span>.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 436-465"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"49559216","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-02DOI: 10.1080/23249935.2023.2236724
Ali Reza Sattarzadeh , Ronny J. Kutadinata , Pubudu N. Pathirana , Van Thanh Huynh
Traffic flow prediction requires learning of nonlinear spatio-temporal dynamics which becomes challenging due to its inherent nonlinearity and stochasticity. Addressing this shortfall, we propose a new hybrid deep learning model based on an attention mechanism that uses multi-layered hybrid architectures to extract spatial–temporal, nonlinear characteristics. Firstly, by designing the autoregressive integral moving average (ARIMA) model, trends and linear regression are extracted; then, integration of convolutional neural network (CNN) and long short-term memory (LSTM) networks leads to better understanding of the model's correlations, serving for more accurate traffic prediction. Secondly, we develop a shuffle attention-based (SA) Conv-LSTM module to determine significance of flow sequences by allocating various weights. Thirdly, to effectively analyse short-term temporal dependencies, we utilise bidirectional LSTM (Bi-LSTM) components to capture periodic features. Experimental results illustrate that our Shuffle Attention ARIMA Conv-LSTM (SAACL) model provides better prediction than other comparable methods, particularly for short-term forecasting, using PeMS datasets.
{"title":"A novel hybrid deep learning model with ARIMA Conv-LSTM networks and shuffle attention layer for short-term traffic flow prediction","authors":"Ali Reza Sattarzadeh , Ronny J. Kutadinata , Pubudu N. Pathirana , Van Thanh Huynh","doi":"10.1080/23249935.2023.2236724","DOIUrl":"10.1080/23249935.2023.2236724","url":null,"abstract":"<div><div>Traffic flow prediction requires learning of nonlinear spatio-temporal dynamics which becomes challenging due to its inherent nonlinearity and stochasticity. Addressing this shortfall, we propose a new hybrid deep learning model based on an attention mechanism that uses multi-layered hybrid architectures to extract spatial–temporal, nonlinear characteristics. Firstly, by designing the autoregressive integral moving average (ARIMA) model, trends and linear regression are extracted; then, integration of convolutional neural network (CNN) and long short-term memory (LSTM) networks leads to better understanding of the model's correlations, serving for more accurate traffic prediction. Secondly, we develop a shuffle attention-based (SA) Conv-LSTM module to determine significance of flow sequences by allocating various weights. Thirdly, to effectively analyse short-term temporal dependencies, we utilise bidirectional LSTM (Bi-LSTM) components to capture periodic features. Experimental results illustrate that our Shuffle Attention ARIMA Conv-LSTM (SAACL) model provides better prediction than other comparable methods, particularly for short-term forecasting, using PeMS datasets.</div></div>","PeriodicalId":48871,"journal":{"name":"Transportmetrica A-Transport Science","volume":"21 1","pages":"Pages 388-410"},"PeriodicalIF":3.6,"publicationDate":"2025-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"43019537","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号