To evaluate the energy savings of electric vehicles (EVs) in transportation-distribution networks (T-DNs), this article proposes a trilevel EV collaborative charging scheduling framework (Tri-CCSF) based on EV’s spatiotemporal flexibility, integrating their spatial flexibility before charging requests and charging flexibility at charging stations (CSs). Tri-CCSF is designed to deliver an integrated optimization of economic charging navigation and scheduling, considering the implications of charging load on the main power grid (MPG) and the inherent uncertainties associated with renewable energy (RE). The first level utilizes aggregated EV charging flexibility to minimize power exchange between CS and MPG. The second level develops low-cost charging plans for each EV user based on energy attributes, which helps avoiding the “curse of dimensionality” caused by large-scale EVs. The third level proposes an EV spatiotemporal dynamic energy consumption estimation model based on macroscopic traffic flow speed prediction, which informs economical charging navigation. Furthermore, the proximal policy optimization-alternating direction method of multipliers (PPO-ADMMs) algorithm is proposed to enhance the solution efficiency of the quadratic programming (QP) problem. Based on real-world map experiments, Tri-CCSF benefits T-DNs and increases RE utilization by 4.5%–36.2% and EV energy consumption estimation accuracy by 5%–7.5%.

As data privacy increasingly becomes a critical societal concern, traditional centralized training for EV charging load prediction may lead to data leakage and missing information. In this paper, a federated quantum learning (FQL) framework for charging prediction is established, which can avoid sensitive privacy concerns by sharing model parameters rather than sharing private data to obtain accurate predictive models. Quantum long short-term memory neural network (QLSTM) is designed to predict the charging load. By integrating a variational quantum circuit into the neural network architecture as a replacement for the traditional weight matrix, QLSTM model significantly enhances its ability to learn complex nonlinear patterns inherent in electric vehicle charging load data, thereby achieving improved prediction accuracy. Comparative experiments based on MAE, MSE, MAPE, and R$^{2}$ metrics demonstrate that F-QLSTM consistently outperforms traditional models such as RNN, LSTM, GRU, and Transformer, achieving higher prediction accuracy with significantly fewer trainable parameters. Specifically, it achieves a 46.1% reduction in parameters compared to LSTM, while maintaining a high R$^{2}$ score, indicating its strong ability to capture complex temporal dependencies and nonlinear patterns. Moreover, experiments conducted across multiple datasets and qubit configurations validate the robustness and generalizability of the F-QLSTM framework.

Highly flexible Automated Guided Vehicles (AGVs) are interconnected via Industrial Wireless Control Networks (IWCNs) in Multi-access Edge Computing (MEC)-assisted smart factories. The MEC alleviates the lack of computational resources in AGV systems through task offloading. However, IWCNs with limited communication resources struggle to support the highly concurrent offloading of AGVs. In the distributed AGV systems with multi-MEC servers, AGV mobility leads to uneven distribution across MEC server areas, potentially resulting in severe competition for communication resources. Therefore, in this paper, we design a Transferable joint Task Offloading and Multi-Channel Access (T2OMCA) algorithm based on multi-agent deep reinforcement learning. Specifically, AGV observations are modelled as graphs, in which edge relationships are learned through Transformer. This enables AGVs to utilize domain information to collaborate and alleviate concurrent offloading. Moreover, the T2OMCA algorithm converts network input into fixed embeddings to accommodate varying numbers of AGVs. Finally, to encourage exploration in the high-dimensional action space, the T2OMCA algorithm introduces a noisy network and a prioritized experience replay mechanism. Extensive simulations show that the T2OMCA algorithm outperforms existing algorithms in terms of average completion rate, processing delay, and access conflict rate under time-varying AGV topologies.

Federated learning (FL) is a distributed learning framework that allows clients to build models without disclosing local data. However, in resource-constrained scenarios, it is costly to participate in FL for all clients. Hence, selection strategy should be designed to select the most appropriate client groups. Current selection strategies are mainly cost and accuracy oriented, ignoring the autonomy of clients, which leads to the inability of clients to make autonomous decisions when participating in model training and updating. To realize autonomous selection of clients, we design a novel model accuracy-aware bidirectional client selection (MABCS) algorithm. The MABCS algorithm implements selection from both server and client dimensions. Specifically, the server evaluates the contributions of clients and design an accuracy-aware dynamic incentive mechanism. The client measures participation autonomy based on the reward and cost to decide whether or not to participate in FL. Thus, the client selection problem is modeled as a joint nonconvex optimization problem that maximizes the system revenue by optimizing the selection strategy and resource allocation strategy. The block coordinate descent algorithm is utilized to decouple the selection strategy and resource allocation strategy, and a linear approximation is employed to transform the selection strategy problem into a convex problem. An alternating optimization algorithm is used for the subproblems after the decomposition to obtain a near-optimal solution. Simulation results indicate that the MABCS algorithm exhibits superior convergence performance compared with other benchmark schemes.

With the increasing integration of transportation network (TN) and distribution network (DN), the growing prevalence of electric vehicles (EVs) has posed significant challenges to both TN and DN operations, such as congestion at charging stations (CSs) and overload in the DN. To address these challenges, this article proposes a bi-layer coupled model that integrates EV route navigation and CS power scheduling. The upper layer of the model optimizes the lowest-cost routes for multiple EVs to reach the CS, while the lower layer optimizes the energy-efficient scheduling of the CS to meet EV charging demands. For solving the model in the form of mixed-integer linear programming (MILP), quantum computing (QC) has demonstrated significant potential. Therefore, a hybrid quantum–classical algorithm combining the digitized-counterdiabatic quantum approximate optimization algorithm (DC-QAOA) with Benders decomposition (DC-Q-BD) is developed. This algorithm leverages the parallelism of QC and the rapid convergence of digitized-counterdiabatic evolution. It also employs collaborative processing between a classical CPU and a quantum processing unit (QPU) to handle continuous and discrete variables, respectively. Results from practical case studies demonstrate that the proposed algorithm significantly enhances solution efficiency and quality. It provides cost-effective charging routes for EVs and notably reduces the adverse impacts of EV charging on the CS. This work provides a quantum-based approach for addressing the EV charging navigation problem in coupled TN and DN.