A station-level electric vehicle (EV) charging market can be modeled as a Stackelberg game, where the charging station (CS) sets prices as the leader and multiple EVs respond with their charging decisions. Although the Stackelberg equilibrium (SE) is important for efficient market operation, existing data-driven methods typically lack either convergence guarantees to SE or the ability to handle practical multi-follower scenarios. To address these challenges, this paper proposes an SE-based multi-agent reinforcement learning method, named SE-MARL, which combines an exact multi-follower Stackelberg Q-learning foundation with a scalable neural implementation. Under idealized finite-state and finite-action assumptions, we prove that the underlying exact tabular update converges to the Stackelberg Bellman fixed point. For scalable implementation, SE-MARL further employs neural networks, replay buffers, and a monotonic mixing network, which enables decentralized follower action generation and mitigates the difficulty of centralized joint-action sampling under partial observability. We also develop a quality-of-service (QoS) evaluation model to reflect EVs’ price sensitivity and queuing concerns in CS pricing optimization. Experiments based on real-world electricity market data show close empirical agreement with the backward-induction benchmark in the reported performance and the reported pricing and charging trajectories, while SE-MARL improves CS profits and EV satisfaction over state-of-the-art baselines.
Linear motor (LM) systems are highly promising for advanced transportation and high-performance motion applications, yet their tracking performance is often degraded by parametric variations, external disturbances, and implementation constraints. This article develops an adaptive disturbance-rejection tracking scheme with practical prescribed-time performance for LM systems subject to composite uncertainties. The proposed method develops a co-design framework by embedding indirect parameter adaptation and integral sliding-mode observation within a barrier-Lyapunov-function-based backstepping architecture, so that parametric mismatch and lumped disturbances are compensated in a coordinated manner while the state-boundedness requirement for observer feasibility is rendered a verifiable closed-loop property. The design requires only bounded uncertainties and does not require disturbance differentiability. A prescribed-time error transformation is further introduced to achieve globally feasible prescribed-performance tracking under uncertain initial conditions. The resulting design enables specification-level decoupling between the prescribed settling time and the steady-state accuracy bound, enabling transparent tuning of transient response and steady-state accuracy. Lyapunov analysis establishes bounded closed-loop signals and practical prescribed-time tracking performance. Experiments on an iron-core LM platform demonstrate improved tracking accuracy, robustness against initial offsets and disturbances, and real-time implementability.
Driven by low-carbon policies, the fuel cell (FC) vehicle industry is embracing rapid development opportunities. This study focuses on commercial vehicle FC systems to conduct power optimization research under variable altitude environments, aiming to resolve the significant net power attenuation issue in altitude-changing scenarios for commercial vehicles. Models for the FC air supply system and atmospheric environment were established, revealing power attenuation with altitude. Analysis characterized compressed air temperature and intercooler thermal loss across altitudes, constructing a four-dimensional mapping of load, air supply, stack performance, and parasitic power. Targeting maximum net system power, an optimal offline algorithm determined ideal air supply parameters under varying altitudes. A dual-degree-of-freedom hierarchical online optimization architecture was designed, proposing an altitude-adaptive strategy integrating offline calibration with online correction. The results show that compared with conventional strategies, 2.02% to 9.07% power enhancement at 4000m altitude. Under custom variable-altitude conditions, the proposed strategy achieves 1.53% to 3.29% power optimization superiority over comparative methods. Finally, the real-time performance of the optimization strategy was validated through hardware-in-the-loop (HIL) testing platform experiments.
Sizing of passive components in electric vehicle drive systems is a critical consideration, as it influences numerous factors, including cost, power density, performance, and efficiency. Traditionally, component sizing has been carried out using empirical design rules under worst-case considerations, which may result in over-sized designs. A systematic approach to sizing is essential to ensure effective hardware utilization without compromising system stability. An impedance-based stability analysis is presented to determine the optimal sizing of system components. Given the nonlinear nature of power converters, subsystems of the electric drive are linearized to obtain small-signal models, which are amenable to linear time-invariant (LTI) system analysis. Further, sensitivity of the overall system stability to parametric variations is studied by sweeping both control-stage and power-stage parameters. Additionally, critical points in the torque-speed regions are identified, and control-loop gain scheduling is explored as an alternative for lowering component size requirements. Results from time-domain simulations verify the correctness of the developed small-signal models. Detailed experimental studies are conducted to assess the stability of both the individual converter and the combined system. Experimental results closely match the simulations, validating the developed sizing and tuning methodology.
High-speed cornering is one of the representative driving scenarios for high-performance electric vehicles, where safety, maneuverability, and real-time control must be fulfilled. This study designs an autonomous cornering trajectory planning and tracking framework for high-performance electric vehicles, based on a continuous high-speed curvature motion planning method and a variable-constraint adaptive model predictive control approach. The proposed continuous high-speed curvature motion planning method provides a generalizable trajectory generation strategy for various racing corners, while considering vehicle dynamics constraints to plan feasible speed profiles. To address the limitations of linear control models in capturing vehicle nonlinear characteristics under extreme maneuvers, this study develops a variable-constraint adaptive model predictive control method. The core of this method lies in the integration of a lateral force intercept parameter into the prediction model, along with real-time updates of both this parameter and the tire cornering stiffness. Furthermore, the tire sideslip angle constraints are adaptively modified during the control process. To assess the limit-handling capabilities, the trajectory tracking performance is specifically evaluated under the aggressive U-turn maneuver. The overall effectiveness and superiority of the proposed approach are validated through co-simulations and driver-in-the-loop testing.
