Transportation Research Part B-Methodological最新文献

Numerous studies have compared machine learning (ML) and discrete choice models (DCMs) in predicting travel demand. However, these studies often lack generalizability as they compare models deterministically without considering contextual variations. To address this limitation, our study develops an empirical benchmark by designing a tournament model to learn the intrinsic predictive values of ML and DCMs. This novel approach enables us to efficiently summarize a large number of experiments, quantify the randomness in model comparisons, and use formal statistical tests to differentiate between the model and contextual effects. This benchmark study compares two large-scale data sources: a database compiled from literature review summarizing 136 experiments from 35 studies, and our own experiment data, encompassing a total of 6970 experiments from 105 models and 12 model families, tested repeatedly on three datasets, sample sizes, and choice categories. This benchmark study yields two key findings. Firstly, many ML models, particularly the ensemble methods and deep learning, statistically outperform the DCM family and its individual variants (i.e., multinomial, nested, and mixed logit), thus corroborating with the previous research. However, this study also highlights the crucial role of the contextual factors (i.e., data sources, inputs and choice categories), which can explain models’ predictive performance more effectively than the differences in model types alone. Model performance varies significantly with data sources, improving with larger sample sizes and lower dimensional alternative sets. After controlling all the model and contextual factors, significant randomness still remains, implying inherent uncertainty in such model comparisons. Overall, we suggest that future researchers shift more focus from context-specific and deterministic model comparisons towards examining model transferability across contexts and characterizing the inherent uncertainty in ML, thus creating more robust and generalizable next-generation travel demand models.

Electric aircraft represent a major technological breakthrough with a promise of revolutionizing aviation systems towards more sustainable and accessible services. Prominent electric aircraft prototypes feature limited seating capacity and short ranges, which make them well-suited for efficiently operating thin routes—particularly, regional routes serving remote regions—in the near future. To capitalize on this opportunity, this paper proposes an original optimization framework in support of the strategic design of subsidized air transport networks using electric aircraft. We first develop a quadratic optimization model to disaggregate air transport demand data based on demand generation and allocation properties, resulting in refined demand estimates at the territorial scale (instead of at the airport level). We then develop an integrated bi-objective optimization model for network and fleet planning, utilizing a novel time-space-energy formulation. This model aims to balance the two primary objectives of planning subsidized air transport networks: maximizing passenger surplus and minimizing system-wide subsidization costs, while incorporating detailed modeling of demand accommodation and electric aircraft operations. To address large-scale problems, we develop a solution approach involving reformulation and a tailored binary relaxation scheme. By considering a real-world case study of Sweden, we demonstrate the benefits of the proposed approach and highlight its major insights—in terms of route network, fleet, number of chargers, flight schedules, fleet assignment and environmental emissions—with a comparison of conventional, electric, and mixed fleets. Our results demonstrate that a complete substitution of first-generation electric aircraft may diminish consumer surplus, while a combined use of electric and conventional aircraft yields superior solutions, resulting in higher passenger surplus and reduced emissions for the same subsidy spending.

Vehicle-based mobile sensing, also known as drive-by sensing, efficiently surveys urban environments at low costs by leveraging the mobility of urban vehicles. While recent studies have focused on drive-by sensing for fleets of a single type, our work explores the sensing power and cost-effectiveness of a mixed fleet that consists of vehicles with distinct and complementary mobility patterns. We formulate the drive-by sensing coverage (DSC) problem, proposing a method to quantify sensing utility and an optimization procedure that determines fleet composition, sensor allocation, and vehicle routing for a given budget. Our air quality sensing case study in Longquanyi District (Chengdu, China) demonstrates that using a mixed fleet enhances sensing utilities and achieves close approximations to the target sensing distribution at a lower cost. Generalizing these insights to two additional real-world networks, our regression analysis uncovers key factors influencing the sensing power of mixed fleets. This research provides quantitative and managerial insights into drive-by sensing, showcasing a positive externality of urban transport activities.

This study investigates the starting point of parking search, presenting new findings through empirical and theoretical approaches. It introduces a probabilistic model that describes the transition from normal driving to actively searching for parking, aiming to minimize journey costs. The model is tested using real-world data collected via a smartphone app that tracks the start of parking searches. Results validate the model, showing that drivers are more likely to begin searching for parking earlier when parking spaces are scarce and driving speeds are reduced (e.g., by congestion). Additionally, various factors influence the start of the parking search, including driver age, vehicle class, and familiarity with the destination. Specific conditions such as proximity to amenities, rush hour timing, and destination familiarity prompt earlier search initiation. The study also identifies scenarios where drivers skip the search process and park immediately, influenced by factors like driving home, short parking durations, and destination familiarity.

The combination airlines operate both passenger aircraft and freighter aircraft to meet passenger and cargo demand. At present, combination airlines employ a sequential approach to allocating their capacity for passenger and cargo demand. Nevertheless, implementing an integrated resource allocation procedure has the potential to improve overall resource allocation efficiency. In this paper, we introduce an integrated model to help combination airlines integrate their aircraft routing and cargo routing decisions to maximize the expected overall profits derived from both passenger and cargo demand. We considered the stochastic nature of passenger baggage and proposed a set of individual chance constraints to ensure the robustness of the integrated solution. We reformulate the chance constraints using piecewise linear approximation to ensure solution efficiency. In addition, we proposed a column-and-row generation based solution approach that removes the through-connection related constraints at the beginning of the solution process and then adds the columns and rows during the iterations as needed. We proved that the proposed column-and-row generation approach can obtain an optimal solution for the LP relaxation problem. The model and the solution approach were tested in a number of scenarios obtained from a major Chinese combination airline. The computational results show that the combination airline can improve their expected profits by integrating capacity allocation. The results also demonstrated that the proposed column-and-row generation solution approach can decrease the solution time of the integrated model. These findings indicate that the model and the solution method are useful and efficient tools for combination airlines when planning their aircraft and cargo routes.

This paper presents a novel multi-agent deep reinforcement learning (MADRL) approach for real-time rescheduling of rail transit services with short-turnings during a complete track blockage on a double-track service corridor. The optimization problem is modeled as a Markov decision process with multiple control agents rescheduling train services on each directional line for system recovery. To ensure computational efficacy, we employ a multi-agent policy optimization solution framework in which each control agent employs a decentralized policy function for deriving local decisions and a centralized value function approximation (VFA) estimating global system state values. Both the policy functions and VFAs are represented by multi-layer artificial neural networks (ANNs). A multi-agent proximal policy optimization gradient algorithm is developed for training the policies and VFAs through iterative simulated system transitions. The proposed framework is implemented and tested with real-world scenarios with data collected from London Underground, UK. Computational results demonstrate the superiority of the developed framework in computational effectiveness compared with previous distributed control algorithms and conventional metaheuristic methods. We also provide managerial implications for train rescheduling during disruptions with different durations, locations, and passenger behaviors. Additional experiments show the scalability of the proposed MADRL framework in managing disruptions with uncertain durations with a generalized model. This study contributes to real-time rail transit management with innovative control and optimization techniques.

Sustainable aviation fuel (SAF) serves as a critical short-term measure for reducing aviation's carbon footprint. Two main policy tools, subsidy and quota, have been developed to support its usage. We build an economic model to compare the environmental and welfare impacts of these two policies. First, we find that if an airline uses the same portion of SAF under both policies, the subsidy approach results in reduced SAF costs, augmented airline output, and heightened emissions. Second, the subsidy policy is better than the quota policy in terms of consumer surplus, airline profits, SAF blender profit, and social welfare, if the traditional aviation fuel is sufficiently inexpensive and the emission regulation under the subsidy policy is adequately stringent. Finally, to demonstrate the empirical relevance of our theoretical framework, we have employed empirically-validated parameters for model calibration. Sensitivity analysis indicates that SAF production costs and market potential are pivotal factors influencing governmental policy formulation. With the reduction in SAF production expenses and the growth of the aviation sector, the government is positioned to adopt a more aggressive policy stance, characterized by higher SAF quotas and increased subsidies, to stimulate the uptake of SAF by airlines.

This paper explores the impacts of a novel business model termed platform integration, which enables passengers to simultaneously request on-demand rides from multiple ride-sourcing platforms via a third-party integrator. In particular, we employ an equilibrium model where passenger demand and driver supply are endogenously dependent on the prices and wages that emerge from the competitive interaction between two platforms, with and without an integrator. A Hotelling model is adopted to characterize passengers’ heterogeneity in service preference for different platforms. We employ the concepts of a Nash equilibrium and a shared monopoly to analyze equilibrium outcomes that can arise in various settings of demand and supply characteristics with and without platform integration. We find that how the platform should adjust its price and wage at Nash equilibrium as potential demand increases is affected by the nature of supply. We also find that the profit at Nash equilibrium can increase or decrease in supply capacity depending on the competitive situation of the platforms. We build on these equilibrium results to analyze how platform integration affects the platform’s decision-making of price and wage, and market performance. We find that platform integration can increase platform profit but reduce driver income, and may hurt passengers who have a strong preference for one certain platform, especially in the case of a less heterogeneous supply.

We investigate the value of time as a resource (VOTR) and the value of childcare time saving (VOCTS) for a married couple with children by life cycle stage. Extending the framework of DeSerpa (1971), we develop a novel intertemporal utility-maximization model that can represent trade-offs within an individual and within a couple between different activities in their life stages based on a household lifetime equilibrium, and we derive wives’ and husbands’ time values when their first child is of pre-school age and after their first child reaches school age. Applying the model to the 2004–2018 Japan Household Panel Survey, we analyze couples in two life stages to empirically find the value of time by gender. The results show that the wives’ average VOTR is greater than 4400 yen/hour with statistical significance when their first child is of pre-school age; the value, however, drastically drops to around 400 yen/hour with statistical insignificance after their first child reaches school age. Conversely, the magnitudes of the husbands’ VOTRs do not change much in different life stages. In the background mechanisms, the wives’ high and low VOTRs reflect their short and long work and commute hours, respectively, whereas the husbands reduce their work and commute hours only slightly over time. For the dual-income households that only spend the minimum required time on childcare, VOCTS is statistically insignificant when their first child is of pre-school age but is greater than 28,000 yen/hour after their first child reaches school age. Using the estimated time values for urban and transport policy simulations, we find that enabling work flexibility could help households increase welfare more compared to transportation improvement and childcare support services.

We consider a ride-hail system in which a third-party integrator receives ride requests and allocates them to ride service platforms. The ride allocation problem (RAP) is modeled as a Stackelberg game. The integrator, as the leader, chooses the allocation that maximizes its profit, by pricing the rides such that no platform (i.e., follower) can find a more profitable allocation. In pursuit of self-interest, the integrator may refuse to match as many rides as the platforms are willing to serve, thereby injecting an artificial scarcity into the system. To protect the platforms from over exploitation, an exogenous reserve price is introduced to bound their per capita profit from below. We formulate RAP as a bilevel pricing problem, and convert it to a single-level problem by dualizing the lower level. When artificial scarcity is eliminated and all reserve prices are set to zero, we prove the single-level problem can be turned into a mixed integer-linear program that equals its linear relaxation, thus becoming polynomially solvable. Moreover, this version of RAP is shown to be related to cooperative assignment games. Numerical experiments confirm that artificial scarcity negatively affects matching productivity and social welfare. The integrator is favored to take most profits, and leveraging artificial scarcity strengthens its dominance. Moreover, the tighter the supply, the more the integrator benefit from artificial scarcity. The reserve price helps redistribute benefits from the integrator to the platforms. However, demanding an excessively large reserve price may depress the platforms’ profits, while undermining system efficiency.