This study addresses the electric bus scheduling problem amid fleet electrification in urban transit systems. A mixed-integer linear programming (MILP) model is formulated for the problem based on a multi-layer time-space network. As a solution approach to this modeling framework, a Deep Reinforcement Learning (DRL) algorithm based on a Deep Q-Network (DQN) is developed to efficiently derive bus scheduling and charging actions for larger instances. While benchmarking results demonstrate MILP solution optimality in small-scale cases, the DRL heuristic scales efficiently to larger networks. An empirical evaluation using real-world datasets confirms that DRL consistently reduces operational costs, minimizes fleet size, and yields substantial computational speed-ups compared to MILP solutions. Our comparative analysis highlights key operational contrasts: MILP ensures robustness through redundant scheduling and extended idling, whereas DRL achieves efficiency via adaptive fleet management and segmented charging. These findings underscore DRL's practical advantages and its scalability, providing guidance for policy-makers and transit planners implementing electric bus scheduling solutions in metropolitan areas.
This study examines how artificial intelligence (AI)-based information system integration, conceptualized as a perceived integration capability embedded in organizational workflows, is associated with perceived overall organizational performance. It focuses on the mediating roles of dynamic capability activities, employee AI action readiness, and environmental, social, and governance (ESG) practices, defined as routinized monitoring, disclosure, and governance activities. Using survey data from 496 employees in firms with AI systems, the proposed framework is tested through structural equation modeling and supplemented by machine-learning-based predictive analysis. The results indicate that AI-based information system integration is associated with perceived overall organizational performance mainly through indirect pathways. It is positively associated with employees’ awareness of AI-supported sensing, seizing, and reconfiguring activities and with their readiness concerning the practical applicability of AI-supported information in daily work, both of which are associated with stronger ESG practices and higher perceived overall organizational performance. Its direct association with ESG practices is positive but weaker than the indirect pathways through dynamic capability activities and employee AI action readiness. Bootstrapped analyses support parallel and sequential indirect effects, while multi-group analyses show that the overall relationship pattern is stable across small and medium-sized enterprises and large enterprises. This study contributes by shifting attention from AI adoption to the perceived integration of AI into organizational workflows and by identifying dynamic capability activities, employee AI action readiness, and ESG practices as complementary pathways linking AI integration to perceived overall organizational performance.
This study addresses the challenge of evaluating Global Supply Chain (GSC) reliability under the simultaneous pressure of economic constraints (tariff-induced budget limits) and environmental constraints (carbon emission caps), compounded by stochastic logistics capacities. We propose a Stochastic Global Supply Chain (SGSC) framework and introduce Landing Reliability, a unified index that quantifies the probability of jointly satisfying demand, carbon compliance, and budget feasibility under random arc capacities. A hybrid approximation algorithm combining Monte Carlo simulation with Linear Programming (LP) is developed to estimate this index, with convergence properties established via the Central Limit Theorem. The framework is applied to a U.S. solar panel import supply chain calibrated to the current multi-layered tariff regime, with systematic sensitivity analyses on correlation structure, state-space granularity, carbon caps, and budget thresholds. Results indicate that the supply chain achieves feasibility in the majority of simulated scenarios. Failure mode decomposition reveals that under current tariffs, failures are predominantly driven by the carbon-capacity interaction, with the budget constraint rarely binding. However, when tariffs escalate under trade remedy scenarios such as antidumping and countervailing duties, Landing Reliability drops substantially and failures shift toward a three-constraint trilemma, demonstrating that tariff risk acts as an amplifier of carbon-capacity risk rather than an independent failure mode. These findings suggest that under favorable tariffs, managers should prioritize capacity expansion on low-emission corridors; when tariffs approach the budget saturation threshold, integrated strategies across cost, carbon, and capacity dimensions become necessary.
This study proposes a spatial compressive sensing approach to reconstruct full-field responses of beams and plates from measurements recorded by only a few sensors. To sparsely represent structural full-field responses, analytical dictionaries composed of sine series and orthogonal basis functions tailored to beam and plate structures are constructed. The required number of sensors is determined using the amplitude distribution of mode shape expansion coefficients, and their optimal locations are selected through the Latin hypercube sampling method. The sparse representation vector is reconstructed using the orthogonal matching pursuit algorithm. A comprehensive parametric study is conducted to investigate the effects of response complexity, boundary conditions, and sensor placement on reconstruction accuracy, followed by laboratory test validation. The results demonstrate that the proposed approach enables robust and efficient reconstruction of full-field responses from sparse measurements, offering a potential strategy for enhancing the reliability of vibration-based structural condition assessment.
Although there has been a variety of crash surrogates have been applied in traffic safety areas, formal discussions on how to validate crash surrogates for practical use remain limited. To fill this gap, this study seeks to propose a validation procedure suitable for traffic safety applications. To first establish key assumptions required for valid crash surrogates, this study reviews discussions on surrogate outcome from causal mediation, medical research, epidemiology clinical trials, and economics perspectives. A causally valid surrogate, whether full or partial, is a special type of mediator that lies on the causal pathway from the primary explanatory variable for a given research question (i.e., the “treatment”) to the final crash outcome. Moreover, it captures an identifiable portion of the treatment’s effect on the outcome, thereby enabling causal inference on the final crash outcome through the surrogate. To determine whether a candidate surrogate satisfies the above properties, the proposed validation procedure consists of three main components: (1) verifying the assumptions required for identifying the causal relationship between the candidate crash surrogate and the crash outcome; (2) quantifying the extent of surrogacy; and (3) evaluating the absence of a surrogate paradox and the comparability of the surrogate. A case study using SHRP2 Naturalistic Driving Study data is then presented to demonstrate the proposed procedure. Specifically, the case study examines whether hard braking is a causally valid surrogate for rear-end crash occurrence.
