Accurate prediction of battery capacity degradation is crucial for the reliable operation of electric vehicles (EVs). Under real-world operating conditions, EV battery systems exhibit complex usage patterns, leading to pronounced uncertainty in the capacity degradation process. Existing methods primarily provide deterministic capacity predictions, which offer limited characterization of degradation-related uncertainty. To address these issues, this paper proposes a probabilistic prediction method for battery capacity degradation trajectories using a denoising diffusion probabilistic model (DDPM). The approach adopts a hybrid feature fusion strategy, extracting statistical and mechanism-based features from the charging data of in-service EVs. A convolutional neural network (CNN) is then employed to extract features from incremental capacity (IC) sequences, which are fused with other point-level features. Under the modulation of these mixed features, DDPM iteratively refines a noisy distribution into the true capacity distribution and samples from it to predict future degradation trajectories, effectively modeling the uncertainty in the battery degradation process. The framework is evaluated using cross-validation on field data from 20 EVs for future capacity trajectory prediction over 124 weeks, yielding a mean absolute error (MAE) of 1.05%, a root mean squared error (RMSE) of 1.34%, a confidence interval coverage rate (CR) of 94.05%, and an average interval width (AIW) of 4.09%. This study highlights a new approach for probabilistic battery health prediction, significantly enhancing the accuracy and reliability for the sustainable development of EV technologies.
Permanent magnet motors designed for lightweight electric vehicles encounter several challenges, including the high cost of rare-earth magnets, torque ripple, cogging torque, thermal demagnetization, and mechanical stresses. Moreover, these motors must deliver high power density within a compact size. To tackle these issues, this paper introduces a novel outer-rotor L-form partitioned-consequent-pole motor, designated as M-3 design in this paper, and evaluates its performance compared to two other reference designs: M-1, which is a conventional outer-rotor surface-mounted permanent magnet (SMPM) motor, and M-2, a partitioned SMPM motor. Through a comprehensive multi-physics analysis, the study investigates the electromagnetic performance, thermal behavior, and mechanical integrity of each design. The findings demonstrate that M-3 significantly reduces the volume of permanent magnets while maintaining high output power. The proposed motor shows improved harmonic quality of the back-electromotive force voltage, reduced torque ripple and cogging torque, a more uniform temperature distribution, enhanced thermal stability, and lower mechanical stress compared to the reference motors. Experimental validation confirms the simulation results, indicating that M-3 effectively addresses cost, thermal, and mechanical concerns while improving overall performance for lightweight electric vehicle applications, such as electric golf carts.
The Electro-mechanical composite braking system (EMCBS) offers an advanced technical foundation for vehicle emergency braking. However, existing control strategies fail to effectively coordinate the braking torques of heterogeneous actuators, making it difficult to simultaneously achieve accurate slip ratio control and high energy recovery. To address this issue, this paper proposes a Stackelberg game-based torque coordinated slip ratio control method for EMCBS. First, a Stackelberg game-based torque coordination framework is established, enabling coordinated optimization of braking safety and energy recovery. Second, an adaptive dynamic programming (ADP) robust slip ratio control scheme is developed to compute the EMB leader braking torque, and an improved robust value function is incorporated to enhance slip ratio tracking accuracy under uncertainties and external disturbances. Then, a Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm is employed to generate the RBS follower braking torque, using a reward function that jointly considers energy recovery and slip ratio jerk to improve regenerative performance and slip ratio steady-state behavior. Finally, simulations and hardware-in-the-loop experiments under various braking scenarios validate the effectiveness and superiority of the proposed method for EMCBS.
Low-speed direct-drive magnetic field-modulated linear machines possess advantages such as high thrust density and low thrust ripple. However, the complex three-dimensional magnetic circuitry obscures the thrust-generation mechanism. For a novel field-modulated linear generator for rail transit, a magnetomotive force-permeance (MMFP) is proposed in this paper. From the perspective of magnetic field modulation, the effects of magnetic flux density harmonics on back electromotive force and thrust are investigated. First, the three-dimensional structure and working principle of the linear machine are introduced. Then, based on the established analytical model, the influencing factors of air gap magnetic flux density harmonics, back electromotive force (EMF), and thrust under hybrid excitation are derived. The thrust and proportion generated by different orders of modulated magnetic flux density harmonics are calculated and analyzed. The finite element method is used to verify the accuracy of the proposed analytical model. Furthermore, the machine's efficiency, power factor, temperature distribution, and thrust were studied. Finally, an experimental test platform is built. No-load back EMF, load current, and thrust were tested to verify the correctness of the proposed model and the accuracy of the calculation results.
This study addresses thermal bottlenecks and localized overheating in permanent magnet linear synchronous motors (PMLSMs) under high thrust density demands. A graded cross-section (GCS) winding fabricated by additive manufacturing (AM) is proposed. First, a 3D refined lumped parameter thermal network model with electrothermal coupling is developed, enabling high-accuracy winding temperature prediction and improved analysis of temperature rise characteristics. To mitigate uneven intra-slot temperature distribution, a heat source density reconstruction method based on winding cross-sectional gradient is introduced, followed by a systematic optimization using a greedy algorithm. Multiphysics finite-element simulations and prototype experiments compare the GCS winding with conventional enameled wire and uniform cross-section AM windings. Results show that the GCS structure reduces peak winding temperature by up to 11.9 °C, suppresses thermal gradients within the slot, and increases motor thrust density by 6.4% relative to the uniform cross-section winding under identical thermal constraints. This work demonstrates the potential of thermally guided winding cross-section allocation for high-thrust-density linear drives.
