This paper presents a 25-kW dual-cell reconfigurable dual active bridge (DCR-DAB) bidirectional DC/DC converter for interfacing second-life battery-based energy storage systems with low-voltage DC buses. The proposed converter employs two identical DAB cells with input- and output-side series/parallel reconfiguration networks to accommodate wide voltage variations. This structure enables 200–1000 V operation at both ports while limiting the maximum voltage applied to each DAB cell to 500 V. For high-power operation, a half-bridge power cell (HBPC) employing parallel-connected SiC MOSFETs with matched gate and power paths, a short-circuit protection circuit, and a multi-gap power-transfer inductor are implemented. Experimental results demonstrate bidirectional power transfer in all four reconfiguration modes, stable operation under load and voltage variations, and 98.1% efficiency at the rated 800/800 V condition. The short-circuit protection limits the fault duration to less than 1 μs under an 800-V test condition. Furthermore, four identical 25-kW modules deliver approximately 81.5 kW at 650 V input and 700 V output with a maximum module-current deviation of approximately 0.35%. These results demonstrate the feasibility and scalability of the proposed DCR-DAB converter for second-life battery-based DC bus stabilization.
This paper presents an experimental investigation of electromagnetic field (EMF) exposure generated by a Dynamic Wireless Power Transfer (DWPT) system implemented within the Electric Road System (ERS) pilot project on the A35 BreBeMi highway. The study focuses on assessing compliance with international safety standards and evaluating potential risks for workers and passengers, including individuals with implantable medical devices. An extensive onboard measurement campaign was conducted inside an electric bus equipped with multiple receiver coils under different operating conditions and vehicle speeds. The EMF exposure assessment was specifically focused on the vehicle interior, with magnetic field levels recorded at multiple passenger-relevant positions and heights. Furthermore, EMC tests performed on representative implantable medical devices (PM, ICD, S-ICD) confirm the absence of interference under realistic exposure conditions. The findings provide new experimental evidence supporting the safe integration of DWPT technology in real-world transportation systems and contribute to the ongoing standardization efforts in wireless charging for electric vehicles.
This paper introduces a tri-port cascaded buck converter (TCBC) for DC power distribution system of More Electric Aircraft (MEA) with reduced device count. The energy storage device (ESD) plays a significant role in more electric aircraft, this work integrates an ESD into the high step-down converter to develop a tri-port cascaded buck converter with reduced device count. The proposed tri-port converter operates with three different modes of operation. The steady-state analysis, along with small-signal modeling for each mode, has been discussed. The proposed converter has been experimentally validated on an experimental test bed using a 270 V input and maintaining a regulated 28 V output. This paper also investigates the performance attributes, including stress, loss, and efficiency analysis, as well as sensitivity and reliability, of the proposed converter. In addition, the proposed converter has been compared with similar reported converter topologies, and is superior with advantages such as a simple circuit with independent power flow control, reduced device count and losses, and improved efficiency and reliability.
Reliable grid-connected electric vehicle (EV) charging requires coordinated operation of cascaded power factor correction and dual active bridge (PFC-DAB) converters under DC-link coupling and grid disturbances. Conventional decoupled control neglects DC-link interactions, resulting in degraded power quality, increased grid current distortion, reduced dynamic stability, and degraded zero-voltage switching (ZVS) caused by converter switching harmonics and input voltage variations. This paper proposes an Adaptive Lyapunov-Guided Multi-Objective Predictive Optimiser (AL-MOPO), a real-time triple-phase-shift (TPS) control framework for cascaded PFC-DAB converters. The constrained nonlinear optimisation problem is analytically reduced offline to explicit phase-shift expressions, which AL-MOPO evaluates at each switching period to minimise peak and root mean square (RMS) inductor currents, enforce ZVS across all switches, and track the battery current reference. An adaptive Chebyshev scalarisation automatically updates the objective weights, while a Lyapunov stability criterion guarantees closed-loop convergence. A ZVS-constrained predictive switching scheme further ensures reliable soft switching over the entire operating range. Experimental results obtained from a 4 kW, 100 kHz SiC prototype demonstrate unity power factor, 3.5% total harmonic distortion (THD), full-range ZVS operation, reduced current stress, and efficiencies of 96.86% for the PFC stage and 98.89% for the DAB stage, validating the proposed controller for high-performance EV charging.
Multi-energy load forecasting is vital for the optimal operation and scheduling of integrated energy management systems. This study presents a deterministic–probabilistic forecasting framework that integrates multi-task learning with a multi-decoder transformer (MDT) architecture to predict electricity and steam loads simultaneously. A two-stage feature screening procedure based on Spearman correlation and random forest importance is employed to identify both monotonic and nonlinear predictors. A unified MDT framework captures shared temporal dependencies across multiple loads, while a gating assisted temporal aggregation module generates task-specific representations by emphasizing informative time steps. Independent deterministic and quantile-regression heads produce point forecasts and prediction intervals, with interval estimation enhanced through a slope-based inflection-aware weighting mechanism. An inner–outer optimization strategy is adopted to coordinate loss minimization and hyperparameter tuning. The inner stage jointly minimizes deterministic and probabilistic losses, whereas the outer stage determines transformer hyperparameters and multi-task weights through a normalized objective function. The proposed framework is validated using real-world industrial data. Experimental results demonstrate that it substantially improves both point and interval forecasting performance compared with existing deterministic and probabilistic benchmarks.
