Accurately predicting on-street parking occupancy is essential for urban traffic management and smart city applications. Previous studies have largely overlooked the importance of semantic correlations among parking locations, while also struggled to effectively capture long-term temporal dependencies and large-scale spatial interactions. To address these limitations, we propose a Semantics-aware Spatial-Temporal Dynamic Graph Transformer Network (SaSFormer) for on-street parking occupancy prediction. Specifically, a graph-based convolutional approach is employed to capture semantic correlations among parking locations; a ProbSparse self-attention mechanism is utilized to model temporal dependencies; and a dynamic self-attention mechanism is designed to capture spatial interactions. Experimental results on real-world datasets demonstrate that SaSFormer significantly outperforms baseline models in prediction accuracy and robustness. Notably, incorporating semantic correlations results in an additional 5.38% reduction in MAE and 6.01% reduction in RMSE, underscoring the critical role of semantic correlations in enhancing on-street parking occupancy prediction.

Accurate prediction of parking occupancy at Park-and-Ride (P+R) facilities is crucial for enabling efficient multimodal transportation systems and supporting dynamic urban mobility management. P+R parking behavior is influenced by multidimensional factors encompassing temporal patterns, meteorological conditions, socioeconomic indicators, and transit-related attributes. To address this challenging prediction task, this study proposes CvTFormer (Cross-variate and Temporal Transformer), a novel Transformer-based model specifically designed to jointly capture temporal dependencies and intricate cross-variate interactions within P+R occupancy data. CvTFormer employs a dual-attention architecture: self-attention mechanisms capture time-dependent patterns, while cross-attention mechanisms facilitate the deep integration of heterogeneous features. Experiments on a real-world dataset spanning 24 P+R facilities in Beijing demonstrate that CvTFormer consistently outperforms state-of-the-art baselines. It achieves average reductions in MAE, RMSE, and MAPE of 10.04%, 9.88%, and 13.45%, respectively, compared to the standard Transformer model. Further ablation studies highlight the importance of incorporating auxiliary features, yielding additional reductions of 6.05% in MAE, 5.34% in RMSE, and 3.15% in MAPE. These results underscore the effectiveness, robustness, and generalizability of CvTFormer for practical urban mobility applications.

With the challenge of making online decisions in the face of escalating losses, the vehicle routing problem in forest fire-fighting places great emphasis on the computing efficiency of algorithms. This paper investigates the route optimization for fire-fighting vehicles and formulates a mixed integer linear programming model to minimize the total losses of overall fire spots. A parallel genetic algorithm with variable neighborhood search (PGA-VNS) is developed, which includes a parallelized evolution process of multiple sub-populations for ensuring the fast convergency to a high-quality solution, and a multi-operator variable searching process to promote the convergency to a nearly optimal solution. By comparing the computational results for various-scale instances, the PGA-VNS is demonstrated to be more effective than the baseline algorithms (i.e., Gurobi, GA, PSO, ALNS, VNS, GA-ALNS, and GA-VNS). Furthermore, the proposed model is extended to the scenario of unmanned aerial vehicles (UAVs), taking into account limited flight range and load impact constraints. The results could support the scheduling and routing for fire engines or UAVs in minimizing the losses caused by forest fires.

Dealing with uncertain rail disruptions effectively raises a significant challenge for computational intelligence research. This article studies the bus bridging problem under demand uncertainty, where the passenger demand is represented as parametric interval-valued fuzzy variables and their associated uncertainty distribution sets. A distributionally robust fuzzy optimization model is proposed to minimize the maximum travel time and to search for the optimal scheme for vehicle allocation, route selection, and frequency determination. To solve the proposed robust model, we discuss the computational issues concerning credibilistic constraints, turning the robust counterpart model into computationally tractable equivalent formulations. The proposed approach is verified, and the resulting method is validated with a report on uncertain parameters in a real-world disrupted event of Shanghai Rail Line 1. Experimental results show that the distributionally robust fuzzy optimization approach can provide a better uncertainty-immunized solution.

Thanks to the booming development of artificial intelligence, 5G technology, and intelligent manufacturing technology, numerous intelligent edge devices contained in the industrial Internet of Things (IIoT) are endowed with the ability to mine knowledge from perceived massive data. Knowledge-driven IIoT plays an unprecedented role in application fields such as cyber-physical systems and Industry 4.0. However, knowledge is generally scattered across the distributed edge devices of IIoT. Therefore, in order to further achieve the edge intelligence in IIoT, it is very important to explore an efficient edge knowledge sharing method. In this article, we establish a decentralized knowledge sharing platform in IIoT. First, for public knowledge, a dynamics model that can quantitatively describe its sharing process is established by using the system dynamics theory. Furthermore, a control method for maximizing public knowledge sharing under constraints based on the optimal control theory is presented. Second, for private knowledge, a trusted transaction control method based on blockchain technology is proposed. By developing both smart contract and lightweight consensus mechanism, the efficient peer-to-peer sharing of private knowledge is realized, and the integrity of knowledge and the privacy of participants are protected. The results of extensive experiments show that the proposed method can eliminate the obstacles of knowledge sharing among edge devices in IIoT, and further promote the development of edge intelligence empowered 5G-enabled IIoT applications.

For subway regenerative energy optimization, the essential idea is to maximize the amount of regenerative energy absorption (AREA) between accelerating trains and braking trains by adjusting train timetable. The bottlenecks preventing the theoretical research from being used to industrial practice lie in the time-consuming simulation of AREA and the constant setting of headway threshold, which seriously reduces the computational efficiency and compresses the solution space. For addressing these issues, we formulate neural networks to speed up the AREA simulation, develop a bisection algorithm to dynamically adjust the headway threshold under moving block mechanism, and then establish a hybrid heuristic method integrating neural networks, genetic algorithm, variable neighborhood search and simulated annealing to determine the timetable, including dwell time distribution (DTD) at stations and headway time distribution (HTD) among consecutive trains. The effectiveness of the proposed method is confirmed by numerical experiments based on the real ATO data of Beijing Subway Changping Line. The results reveal that the neural networks could save 98.96% of the computational time at the expense of 1.76% of the accuracy loss, such that the hybrid heuristic algorithm could save 75.68% of the computational time. Benchmarked with the constant headway constraints used in literature, the variable headway constraints could achieve an average improvement of AREA by 8.94%. In addition, the joint optimization of DTD and HTD could increase the AREA by 15.28% on average, compared to the single optimization of HTD. This research is of great significance to improve the practicability of subway regenerative energy optimization algorithms.

Condition-based maintenance has been developed as a very efficient strategy for guaranteeing multicomponent system performance and preventing unexpected failures. However, there are shortcomings in the existing condition-based maintenance optimization models. First, the existing models do not utilize the accelerated degradation testing (accelerated degradation testing) data obtained at the stage of component development. Second, most of these models assume perfect repair instead of imperfect repair. Third, the degradation models used in these condition-based maintenance models cannot consider the epistemic uncertainty. Motivated by these problems, this paper presents a new condition-based maintenance optimization model for multicomponent systems with imperfect repair. An integrated degradation prediction framework utilizing both ADT data and field data is presented to timely update the parameters in the proposed model. In order to solve the proposed multivariable, nonlinear programming model, a novel genetic algorithm with self-crossover operation and shift-mutation operation is developed. Numerical examples and comparisons are conducted to evaluate the performance of the proposed model. Results show that the proposed model can evaluate the degradation process of components accurately and achieve lower total maintenance cost.