Origin-destination Distribution Research Articles

Deep Learning Model for Short-Term Origin–Destination Distribution Prediction in Urban Rail Transit Network Considering Destination Choice Behavior

Urban rail transit (URT) has emerged as a crucial mode of transportation in metropolitan areas. For the effective operation of expanding URT networks, accurate short-term origin–destination (OD) demand distribution predictions are essential. This study introduces a novel deep-learning-based model for predicting short-term OD distribution in extensive networks, taking destination choice behaviors into account. First, we perform a comprehensive analysis of station passenger flows and OD flows from both temporal and spatial dimensions. Then, we develop the origin–destination distribution prediction (ODDP) model, combining the destination choice model (DCM) with the deep learning model (DLM). The DCM aims to understand OD distribution patterns from a behavioral perspective by transforming real-time inflows into OD distributions. Meanwhile, the DLM, employing attention and convolution layers, effectively captures the intricate temporal and spatial dynamics of passenger flows. Our model is evaluated using data from the Guangzhou Metro network in China, showing significant enhancements in prediction accuracy, model interpretability, and overall robustness. The implementation of our model promises substantial benefits for the operational efficiency of URT systems.

Improving passenger travel efficiency through a dynamic autonomous non-stop rail transit system

This study proposes a dynamic autonomous non-stop rail transit (DANRT) system to improve the travel efficiency of passengers in urban rail transit (URT) systems and solves a pertinent carriage scheduling problem derived from the DANRT system using a mathematical programming model. In the DANRT system, passengers traveling to the same destination are allocated to the same carriage(s), and each carriage can be attached to and detached from trains using the modular autonomous vehicle (MAV) technology effortlessly, which enables all trains to run non-stop throughout the focused operation period. We offer a cost-effective design for the DANRT system. To ensure safe and efficient operations, a mathematical model is proposed for the carriage scheduling problem in the DANRT system, where the number and destinations of carriages required by each station are determined. A linearization and segmentation method for the model is proposed. To examine the effectiveness of the DANRT system, we compare the travel efficiency of passengers in the DANRT system with that in the traditional system by using the origin-destination distribution of passengers on the Batong Line of Beijing Subway. The results demonstrate that passengers in DANRT can save about 2.9 % to 8.6 % of travel time compared with the traditional system. Finally, we conclude several observations and operational characteristics of the DANRT system by numerical experiments.

Inferring origin-destination distribution of agent transfer in a complex network using deep gated recurrent units

Predicting the origin-destination (OD) probability distribution of agent transfer is an important problem for managing complex systems. However, prediction accuracy of associated statistical estimators suffer from underdetermination. While specific techniques have been proposed to overcome this deficiency, there still lacks a general approach. Here, we propose a deep neural network framework with gated recurrent units (DNNGRU) to address this gap. Our DNNGRU is network-free, as it is trained by supervised learning with time-series data on the volume of agents passing through edges. We use it to investigate how network topologies affect OD prediction accuracy, where performance enhancement is observed to depend on the degree of overlap between paths taken by different ODs. By comparing against methods that give exact results, we demonstrate the near-optimal performance of our DNNGRU, which we found to consistently outperform existing methods and alternative neural network architectures, under diverse data generation scenarios.

Exploring the Spatio-Temporal and Behavioural Variations in Taxi Travel Based on Big Data during the COVID-19 Pandemic: A Case Study of New York City

The COVID-19 pandemic has caused severe social and economic chaos worldwide. To explore the impact of the COVID-19 pandemic on the travel patterns of residents, we analysed taxi trajectory data and COVID-19 pandemic data from New York City. Pearson coefficients, which were −0.7139, −0.8041, and −0.7046 during the three waves of the COVID-19, revealed a significant negative correlation between confirmed cases and taxi trips. Moran’s I was higher in drop-off areas than in pick-up areas, indicating a stronger spatial autocorrelation in drop-off areas during the study period. The hotspots of travel destinations had changed by spatial clustering, and variations in origin–destination distribution were obvious after the pandemic. Comparison of temporal and spatial dimensions before and after the pandemic revealed that strict epidemic policies directly affected travel. For instance, a week after the restrictions the taxi journeys plummeted by 95.3%, and their spatial and temporal patterns also changed. Once the anti-epidemic policy was eased or lifted, the taxi travel recovered, whereas, notably the new Omicron wave did not cause dramatic changes in taxi journeys. Despite this, travel spatial and temporal patterns did not return to pre-pandemic levels by the end of March 2022, the taxi journeys remained below half the pre-pandemic level. This study identified the profound impact of the COVID-19 outbreak on travel patterns and revealed distinct variations in behavioural responses during the pandemic and in response to subsequent policies. Strengthening targeted epidemic prevention and control measures are required to improve the balance between anti-epidemic policies and implementation efforts, that will facilitate the recovery of urban transport, work, and lifestyle of residents.

Simulating inter-individual contact in the inter-station passenger transfer system connecting multiple metro stations based on space–time path data

Determining passengers’ inter-individual contact in the metro station area (MSA) is an important issue to simulate and mitigate the spread of the Coronavirus disease 2019 (COVID-19) pandemic. Taking the inter-station passenger transfer system (IPTS) as an example, this study aimed to verify the passenger flows’ influence on the inter-individual contact around the MSA. Based on actual observed data, the passengers’ space–time paths (STP) in the network were obtained through an agent-based simulation. In this study, the direct contact model and the mediate contact model were used to describe the inter-individual contact in view of the passengers’ STP. The contact count and the exposure duration were defined as indicators to measure the contact degree of individual and the system. The results show that the time-varying trip distribution of the metro passengers significantly affected the inter-individual contact degree and the spatial distribution of contact risk region in the MSA. The intersection of passenger flow in different directions and the concentrated movement of passenger flow in the same direction increased the inter-individual contact and prolonged exposure in the morning. Through simulation experiments, the study verified the effects of controlling the flow direction and equalizing passenger flow generation measures aiming to reduce inter-individual contact and cumulative exposure duration.

Impact Estimation of Unplanned Urban Rail Disruptions on Public Transport Passengers: A Multi-Agent Based Simulation Approach.

Once unplanned urban rail disruptions occur, it is essential to evaluate the impacts on public transport passengers since impact estimation results enable transit agencies to verify whether alternative transit services have adequate capacity to evacuate the affected rail passengers and to adopt effective emergency measures in response to the disruptions. This paper focuses on estimating the impacts of unplanned rail line segment disruptions on rail passengers as well as original bus passengers, as the latter are overlooked in existing studies. A method of identifying affected rail passengers based on passenger tap-in time is proposed, which is helpful for evaluating the scale and origin-destination distribution of the affected passengers. Passengers’ response behaviors are analyzed and modeled in a multi-agent simulation system. The system realizes the simulation of the multimodal evacuation process, in which a rule-based logit model is employed to describe passengers’ travel selection behavior and the Monte Carlo method is utilized to address the issue of uncertainty in passengers’ travel selection. In particular, the original bus passengers are integrated into the simulation and interact with rail passengers. Finally, some indicators assessing the impacts on rail passengers and bus passengers are presented, and a case study based on the Ningbo urban rail transit network is conducted.

Prediction for Origin-Destination Distribution of Dockless Shared Bicycles: A Case Study in Nanjing City.

Shared bicycles are currently widely welcomed by the public due to their flexibility and convenience; they also help reduce chemical emissions and improve public health by encouraging people to engage in physical activities. However, during their development process, the imbalance between the supply and demand of shared bicycles has restricted the public's willingness to use them. Thus, it is necessary to forecast the demand for shared bicycles in different urban regions. This article presents a prediction model called QPSO-LSTM for the origin and destination (OD) distribution of shared bicycles by combining long short-term memory (LSTM) and quantum particle swarm optimization (QPSO). LSTM is a special type of recurrent neural network (RNN) that solves the long-term dependence problem existing in the general RNN, and is suitable for processing and predicting important events with very long intervals and delays in time series. QPSO is an important swarm intelligence algorithm that solves the optimization problem by simulating the process of birds searching for food. In the QPSO-LSTM model, LSTM is applied to predict the OD numbers. QPSO is used to optimize the LSTM for a problem involving a large number of hyperparameters, and the optimal combination of hyperparameters is quickly determined. Taking Nanjing as an example, the prediction model is applied to two typical areas, and the number of bicycles needed per hour in a future day is predicted. QPSO-LSTM can effectively learn the cycle regularity of the change in bicycle OD quantity. Finally, the QPSO-LSTM model is compared with the autoregressive integrated moving average model (ARIMA), back propagation (BP), and recurrent neural networks (RNNs). This shows that the QPSO-LSTM prediction result is more accurate.

A route-planning method for long-distance commuter express bus service based on OD estimation from mobile phone location data: the case of the Changping Corridor in Beijing

With the background of hyperurbanization and a jobs–housing imbalance in Beijing and other megacities in China, this study aims at developing a systematic toolkit of demand estimation and route planning for long-distance commuter bus lines. Taking the express bus services (EBS) in the Changping Corridor in Beijing as an example, this paper presents the use of a route-planning algorithm. Mobile phone location (MPL) data serves as a valid instrument for the origin–destination (OD) estimation, which provides a new perspective to identify the locations of homes and jobs. The OD distribution matrices are specified via geocoded MPL data. To minimize the aggregated travel time and attract potential passengers, this study subsamples long-distance commuting trips among the OD matrices by determining the operation distance threshold. The threshold can be solved by comparing the aggregated travel time of EBS to the travel time of local buses and private cars. Using the distance threshold and other operational parameters, we process an iterative computation to determine the length of routes, number of lines, and stop spacing and location. Compared to local buses and private cars, the planned EBS lines could reduce aggregated travel time by at least 20%. The results demonstrate that the method presented in this study is applicable and flexible. This paper may lead to new research directions for route planning of long-distance commuter buses. The limitations of this study and the future research agenda are also discussed.

Optimal Express Bus Routes Design with Limited-Stop Services for Long-Distance Commuters

This research aimed to propose a route optimization method for long-distance commuter bus service to improve the attraction of public transport as a sustainable travel mode. Taking the express bus services (EBS) in Changping Corridor in Beijing as an example, we put forward an EBS route-planning method for long-distance commuter based on a solving algorithm for vehicle routing problem with pickups and deliveries (VRPPD) to determine the length of routes, number of lines, and stop location. Mobile phone location (MPL) data served as a valid instrument for the origin–destination (OD) estimation, which provided a new perspective to identify the locations of homes and jobs. The OD distribution matrices were specified via geocoded MPL data. The optimization objective of the EBS is to minimize the total distance traveled by the lines, subject to maximum segment capacity constraints. The sensitivity analysis was done to several key factors (e.g., the segment capacity, vehicle capacity, and headway) influencing the number of lines, the length of routes. The results suggest that the scenario with the segment capacity of 4000 passengers/h has a minimum of number and length of lines, but we recommend that the transit agency adopt 3000 passengers/h as the route segment capacity because this scenario results in minimum fleet size and minimum total operation length.

A Distributed Control Method for Urban Networks Using Multi-Agent Reinforcement Learning Based on Regional Mixed Strategy Nash-Equilibrium

Urban network traffic congestion can be caused by disturbances, such as fluctuation and disequilibrium of traffic demand. This paper designs a distributed control method for preventing disturbance-based urban network traffic congestion by integrating Multi-Agent Reinforcement Learning (MARL) and regional Mixed Strategy Nash-Equilibrium (MSNE). To enhance the disturbance-rejection performance of Urban Network Traffic Control (UNTC), a regional MSNE concept is integrated, which models the competitive relationship between each agent and its neighboring agents in order to improve the decision-making process of MARL. The learning rate is enhanced with a self-adaptive ability to avoid a local optimal dilemma; Jensen-Shannon (JS) divergence is utilized to define the learning rate of the modified MARL. A two-way rectangular grid network with nine intersections is modeled via a Cell Transmission Model (CTM). A probability distribution mechanism, which can update the turn ratio of each approach dynamically and discretely, is established to represent the segmented route-decision process of the vehicles. The effectiveness of the proposed control method is evaluated through simulations in the grid network. The results show the influence of major disturbances, such as fluctuation of vehicle arrival rate, fluctuation of traffic demand (e.g. a rapidly rising flow and extreme changes in origin-destination distribution), and disequilibrium of traffic demand (e.g. different arrival flows at each boundary of the urban network), on the performance of the suggested control method. The results can be used to improve the state of the art in order to reduce urban network traffic congestion due to these disturbances.

Optimizing Strategies for the Urban Work Zone with Time Window Constraints

Work zones that move with road maintenance tasks are enclosing and have caused severe traffic jams and the significant decline of road capacity. This paper proposes an intelligent-based multi-objects road maintenance optimization strategy based on a practical origin–destination (OD) matrix and complicated work schedules over a real urban road network. It focuses on the optimization of multi short-term maintenance tasks and the minimization of average travel delay for vehicles passing through. By taking the driving characteristic into account, static and dynamic variable speed limit strategies provide access to ensure safety on the working road network. Through this view, the problem was formulated as a mixed multi-object nonlinear program (MNLP) model with respect to the time window of the related sub-maintenance task. By using actual OD distribution matrix data, a series of microscopic simulated cases were conducted to test the model’s validity. Moreover, sensitive analyses of types of parameters (e.g., traffic safety threshold, traffic flow and working efficiency) with an optimal solution were discussed considering five different scenarios.

Taxi trips distribution modeling based on Entropy-Maximizing theory: A case study in Harbin city—China

Understanding Origin–Destination distribution of taxi trips is very important for improving effects of transportation planning and enhancing quality of taxi services. This study proposes a new method based on Entropy-Maximizing theory to model OD distribution in Harbin city using large-scale taxi GPS trajectories. Firstly, a K-means clustering method is utilized to partition raw pick-up and drop-off location into different zones, and trips are assumed to start from and end at zone centers. A generalized cost function is further defined by considering travel distance, time and fee between each OD pair. GPS data collected from more than 1000 taxis at an interval of 30 s during one month are divided into two parts: data from first twenty days is treated as training dataset and last ten days is taken as testing dataset. The training dataset is used to calibrate model while testing dataset is used to validate model. Furthermore, three indicators, mean absolute error (MAE), root mean square error (RMSE) and mean percentage absolute error (MPAE), are applied to evaluate training and testing performance of Entropy-Maximizing model versus Gravity model. The results demonstrate Entropy-Maximizing model is superior to Gravity model. Findings of the study are used to validate the feasibility of OD distribution from taxi GPS data in urban system.

Time Coefficient Estimation for Hourly Origin-Destination Demand from Observed Link Flow Based on Semidynamic Traffic Assignment

Day-long origin-destination (OD) demand estimation for transportation forecasting is advantageous in terms of accuracy and reliability because it is not affected by hourly variations in the OD distribution. In this paper, we propose a method to estimate the time coefficient of day-long OD demand to estimate hourly OD demand and to predict hourly traffic for urban transportation planning of a large-scale road network that lacks discrete-time rich traffic data. The model proposed estimates the time coefficients from observed link flows given a proven day-long OD demand based on a bilevel formulation of the generalized least square and semidynamic traffic assignment (OD-modification approach). The OD-modification approach is formulated as a static user-equilibrium assignment with elastic demand, based on the residual demand at the end of each period. Our model does not require setting many parameters regarding the OD demand matrices and the discrete-time dynamic traffic assignments. Applying the model to large-scale road network demonstrates that it efficiently improves estimation accuracy because the 24-hour time coefficients of survey data are slightly biased and may be modified properly. In addition, the methods that partially relax the assumption of OD-modification approach and transform the estimated demand into demand based on departure time are examined.

An Improved Regional Traffic Volume Dynamic Feedback Control

The connection between traffic congestion and regional demand has become a consensus. A reasonable road network inflow control is essential to prevent the occurrence of congestion. With a consideration of deployment of detectors, an improved macroscopic fundamental diagram-based traffic volume dynamic control method via the feedback control is proposed in this study. Based on the origin–destination distribution of trips and connectivity of nodes, a novel method is adopted to identify the key nodes, on which the detectors are laid. An artificial neural network is adopted to predict traffic volume of those sections without detectors. As a case study, the proposed methodology is applied and estimated via simulation study to the regional road network which is located in the downtown area of Nanning, China. Adequate survey and analysis are carried out under the current road traffic conditions. It is proved that the regional traffic volume dynamic control can ensure steady and orderly regional traffic flow, and enhance the mobility during saturated traffic conditions.

Holiday Destination Choice Behavior Analysis Based on AFC Data of Urban Rail Transit

For urban rail transit, the spatial distribution of passenger flow in holiday usually differs from weekdays. Holiday destination choice behavior analysis is the key to analyze passengers’ destination choice preference and then obtain the OD (origin-destination) distribution of passenger flow. This paper aims to propose a holiday destination choice model based on AFC (automatic fare collection) data of urban rail transit system, which is highly expected to provide theoretic support to holiday travel demand analysis for urban rail transit. First, based on Guangzhou Metro AFC data collected on New Year’s day, the characteristics of holiday destination choice behavior for urban rail transit passengers is analyzed. Second, holiday destination choice models based on MNL (Multinomial Logit) structure are established for each New Year’s days respectively, which takes into account some novel explanatory variables (such as attractiveness of destination). Then, the proposed models are calibrated with AFC data from Guangzhou Metro using WESML (weighted exogenous sample maximum likelihood) estimation and compared with the base models in which attractiveness of destination is not considered. The results show that theρ2values are improved by 0.060, 0.045, and 0.040 for January 1, January 2, and January 3, respectively, with the consideration of destination attractiveness.

Optimal traffic plate scanning location for OD trip matrix and route estimation in road networks

During the last decade, there has been a substantial interest in how to determine the optimal number and locations of traffic counters for origin–destination (OD) trip matrices estimation. On the contrary, the optimal allocation of plate scanning devices has received very limited attention, even though several authors have demonstrated that plate scanning (route identification) techniques are much more informative than those based on traditional link count information. This paper provides techniques for obtaining the optimal number and location of plate scanning devices for a given prior OD distribution pattern under different situations, i.e. maximum route identifiability or budget constraints. Two rules analogous to the counting location problem are developed, and several integer linear programming models fulfilling these rules are proposed. The proposed methods are finally illustrated by their application into Nguyen–Dupuis and Cuenca networks.

The optimal location of a single ring road

Consider a radial road network such that a single ring road is the only connection between the radials. This paper shows that there is, for any fixed origin-destination distribution, a location of the ring road which minimises the impact of radial traffic flow, for almost any criterion used to assess that impact. An optimal ring road has as many relevant trip-ends inside as outside.