Sinan Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Sinan Li has a total of 209 co-authors. Here are the top co-authors by number of shared publications:
- Shu Yuen Ron Hui
City University of Hong Kong
33 shared publications - Siew-Chong Tan
City University of Hong Kong
24 shared publications - Jianguo Zhu
School of Electrical and Computer Engineering, The University of Sydney, Camperdown, NSW 2006, Australia
12 shared publications - Wenlong Qi
Electrical and Electronics engineering, University of Hong Kong, 25809 Hong Kong Hong Kong
10 shared publications - Huawei Yuan
City University of Hong Kong
8 shared publications - Huan Li
School of Electrical and Information Engineering, University of Sydney, Sydney, NSW, Australia
6 shared publications
Latest Publications
- Open Access
- 17 August 2026
- Published by Elsevier in Control Engineering Practice
- Vol. 177
A decoupling framework for ultra-local-model-based model-free control of power electronic systems
- 17 August 2026
- Published by Elsevier in Control Engineering Practice
- Vol. 177
Open AccessA decoupling framework for ultra-local-model-based model-free control of power electronic systems
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 41 (7) , 10987-11002
Neural-Enhanced Physics: A New Paradigm for Accurate and Dimension-Aware Magnetic Core Loss Modeling
The accurate and dimension-aware modeling of magnetic core loss is fundamentally limited by a trade-off: model-driven approaches are inherently dimension-aware but are often inaccurate; purely data-driven methods are often accurate but lack dimension-transferability. This paper resolves this dilemma by introducing a novel neural-enhanced physics (NEP) paradigm, which enables highly data-efficient adaptation to different physical core dimensions while maintaining high fidelity. We demonstrate this paradigm with the neural-enhanced dynamic circuit model (NE-DCM), where a physics-based circuit handles dimension-aware electromagnetic dynamics while a data-efficient neural network accurately captures complex material hysteresis. This architecture, based on a principled decoupling of physics and material behavior, yields a key practical advantage: a highly data-efficient workflow. The neural network is trained only once using quasi-static data for a given material, while the physics-based model requires only minimal calibration for new core dimensions. Validated on TDK N87 toroidal cores of different sizes under diverse excitations, the NE-DCM demonstrates superior accuracy, waveform generality, and—most critically—data-efficient transferability across core dimensions, outperforming both purely model-driven and purely data-driven models. The extensibility of the paradigm is further confirmed through successful validation on a second, distinctly different ferrite material (TDK T65). This work establishes the NEP paradigm as a robust and extensible foundation for the next generation of magnetic component design tools.
- 1 July 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 41 (7) , 10987-11002
Neural-Enhanced Physics: A New Paradigm for Accurate and Dimension-Aware Magnetic Core Loss Modeling
The accurate and dimension-aware modeling of magnetic core loss is fundamentally limited by a trade-off: model-driven approaches are inherently dimension-aware but are often inaccurate; purely data-driven methods are often accurate but lack dimension-transferability. This paper resolves this dilemma by introducing a novel neural-enhanced physics (NEP) paradigm, which enables highly data-efficient adaptation to different physical core dimensions while maintaining high fidelity. We demonstrate this paradigm with the neural-enhanced dynamic circuit model (NE-DCM), where a physics-based circuit handles dimension-aware electromagnetic dynamics while a data-efficient neural network accurately captures complex material hysteresis. This architecture, based on a principled decoupling of physics and material behavior, yields a key practical advantage: a highly data-efficient workflow. The neural network is trained only once using quasi-static data for a given material, while the physics-based model requires only minimal calibration for new core dimensions. Validated on TDK N87 toroidal cores of different sizes under diverse excitations, the NE-DCM demonstrates superior accuracy, waveform generality, and—most critically—data-efficient transferability across core dimensions, outperforming both purely model-driven and purely data-driven models. The extensibility of the paradigm is further confirmed through successful validation on a second, distinctly different ferrite material (TDK T65). This work establishes the NEP paradigm as a robust and extensible foundation for the next generation of magnetic component design tools.
- Open Access
- 7 May 2026
- Published by Elsevier in SSRN Electronic Journal
Accelerating Maximum Energy Efficiency Tracking in Dynamic Wireless EV Charging via Fast Model-Free Voltage Control
Stable and rapid output-voltage regulation is crucial in dynamic wireless charging (DWC) systems for electric vehicles (EVs), because transmitter-side perturb-and-observe maximum energy efficiency tracking (P&O-MEET) can reliably evaluate input current only after the receiver voltage has settled. Conventional proportional-integral (PI) control often exhibits limited disturbance rejection under motion-induced coupling variation, load changes, and phase-shift perturbations, thereby limiting effective-window delivered energy. This paper introduces ultra-local model-free control (MFC) for receiver-side voltage regulation in DWC. MFC uses real-time input/output data to estimate lumped dynamics and disturbances without requiring an accurate plant model. A small-signal analysis explains how the cascaded MFC structure reshapes the disturbance-rejection transfer function relative to a single-loop PI baseline. Simulations and experiments on a 5-W, 1/13-scale control-oriented prototype show that MFC reduces voltage settling time by 48.1% and voltage ripple by 22.0% on average under representative DWC disturbances. This faster receiver response permits a shorter P&O update period, reducing static MEET convergence time and dynamic minimum-input-current tracking time by over 40% in the tested cases. At 70 km/h scaled-equivalent speed, the MFC-based system maintains effective tracking whereas the PI-based system fails to converge within the effective charging window. Consequently, MFC yields per-pass gains of 3.69% in effective-window delivered energy and 1.88% in effective-window DC-input-to-load efficiency, indicating potential cumulative benefits over multi-segment DWC roadways.
- 7 May 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessAccelerating Maximum Energy Efficiency Tracking in Dynamic Wireless EV Charging via Fast Model-Free Voltage Control
Stable and rapid output-voltage regulation is crucial in dynamic wireless charging (DWC) systems for electric vehicles (EVs), because transmitter-side perturb-and-observe maximum energy efficiency tracking (P&O-MEET) can reliably evaluate input current only after the receiver voltage has settled. Conventional proportional-integral (PI) control often exhibits limited disturbance rejection under motion-induced coupling variation, load changes, and phase-shift perturbations, thereby limiting effective-window delivered energy. This paper introduces ultra-local model-free control (MFC) for receiver-side voltage regulation in DWC. MFC uses real-time input/output data to estimate lumped dynamics and disturbances without requiring an accurate plant model. A small-signal analysis explains how the cascaded MFC structure reshapes the disturbance-rejection transfer function relative to a single-loop PI baseline. Simulations and experiments on a 5-W, 1/13-scale control-oriented prototype show that MFC reduces voltage settling time by 48.1% and voltage ripple by 22.0% on average under representative DWC disturbances. This faster receiver response permits a shorter P&O update period, reducing static MEET convergence time and dynamic minimum-input-current tracking time by over 40% in the tested cases. At 70 km/h scaled-equivalent speed, the MFC-based system maintains effective tracking whereas the PI-based system fails to converge within the effective charging window. Consequently, MFC yields per-pass gains of 3.69% in effective-window delivered energy and 1.88% in effective-window DC-input-to-load efficiency, indicating potential cumulative benefits over multi-segment DWC roadways.
- 30 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (8) , 15747-15759
Decentralized Carbon-Aware Dispatch for Virtual Power Plants in an Unbalanced Distribution Network via Heterogeneous Reinforcement Learning
Environmentally sustainable operation is critical for distribution networks (DNs) with high distributed energy resource (DER) penetration. This article proposes a decentralized carbon-aware virtual power plant (CAVPP) framework that jointly minimizes operational cost and carbon emissions while respecting network constraints. We introduce a carbon-emission flow (CEF) model that allocates demand-side emissions and, unlike prior single-phase approximations suffering from under/overallocation, develop a realistic unbalanced DN formulation with nodal-phase emission tracing. The CAVPP coordination problem is recast as a cooperative Markov game (MG) and solved via a heterogeneous-agent soft actor–critic (HASAC) algorithm that accommodates diverse agent roles and dynamic action–observation spaces. Tested on the IEEE 123-bus unbalanced system, the proposed approach achieves an 8.76% reduction in operating cost and a 21.74% reduction in total carbon emissions relative to a state-of-the-art benchmark, demonstrating coordinated scalability and deployability for large, heterogeneous DER populations.
- 30 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (8) , 15747-15759
Decentralized Carbon-Aware Dispatch for Virtual Power Plants in an Unbalanced Distribution Network via Heterogeneous Reinforcement Learning
Environmentally sustainable operation is critical for distribution networks (DNs) with high distributed energy resource (DER) penetration. This article proposes a decentralized carbon-aware virtual power plant (CAVPP) framework that jointly minimizes operational cost and carbon emissions while respecting network constraints. We introduce a carbon-emission flow (CEF) model that allocates demand-side emissions and, unlike prior single-phase approximations suffering from under/overallocation, develop a realistic unbalanced DN formulation with nodal-phase emission tracing. The CAVPP coordination problem is recast as a cooperative Markov game (MG) and solved via a heterogeneous-agent soft actor–critic (HASAC) algorithm that accommodates diverse agent roles and dynamic action–observation spaces. Tested on the IEEE 123-bus unbalanced system, the proposed approach achieves an 8.76% reduction in operating cost and a 21.74% reduction in total carbon emissions relative to a state-of-the-art benchmark, demonstrating coordinated scalability and deployability for large, heterogeneous DER populations.
- 12 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 41 (3) , 3966-3983
A Synergistic Modulation Method for Simultaneous Active Power Decoupling and Tank Current Reduction in Single-Stage Dual Active Bridge Microinverters
Conventional single-stage dual active bridge microinverters typically suffer from bulky input capacitors for double-line frequency power decoupling, significant input current pulsation, and excessive circulating current under single-phase-shift modulation. This article proposes a novel synergistic modulation method to comprehensively address these limitations. The proposed method achieves active power decoupling under arbitrary power factor conditions by repurposing the inherent primary-side dc-blocking capacitor, thus avoiding any hardware modifications. Meanwhile, the modulation parameters are optimized to substantially reduce the tank current. A systematic theoretical analysis is conducted, and a large-signal-based automatic decoupling control loop is developed. The experimental results on a 500-VA prototype demonstrate that the proposed modulation method reduces the decoupling capacitor volume by 75%, suppresses the double-line frequency pulsation in the input current by 87%, and decreases the root-mean-square value of the tank current by 27%, thereby improving the circuit’s power density, maximum power point tracking accuracy, and overall efficiency.
- 12 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 41 (3) , 3966-3983
A Synergistic Modulation Method for Simultaneous Active Power Decoupling and Tank Current Reduction in Single-Stage Dual Active Bridge Microinverters
Conventional single-stage dual active bridge microinverters typically suffer from bulky input capacitors for double-line frequency power decoupling, significant input current pulsation, and excessive circulating current under single-phase-shift modulation. This article proposes a novel synergistic modulation method to comprehensively address these limitations. The proposed method achieves active power decoupling under arbitrary power factor conditions by repurposing the inherent primary-side dc-blocking capacitor, thus avoiding any hardware modifications. Meanwhile, the modulation parameters are optimized to substantially reduce the tank current. A systematic theoretical analysis is conducted, and a large-signal-based automatic decoupling control loop is developed. The experimental results on a 500-VA prototype demonstrate that the proposed modulation method reduces the decoupling capacitor volume by 75%, suppresses the double-line frequency pulsation in the input current by 87%, and decreases the root-mean-square value of the tank current by 27%, thereby improving the circuit’s power density, maximum power point tracking accuracy, and overall efficiency.
- 1 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 41 (1) , 1087-1101
A Modulation Method for Decoupling Capacitance Reduction in Single-Stage Dual-Active-Bridge Microinverters
Conventional single-stage dual active bridge (DAB) microinverters typically have a bulky capacitor for decoupling double-line frequency power and a significant pulsation in the input current, leading to reduced power density and degraded maximum power point tracking (MPPT) performance. This article presents a new modulation method for single-stage DAB microinverters, allowing for substantial reductions in both the required decoupling capacitance and the input current ripple without hardware modifications. This is achieved by repurposing the inherent primary-side dc-blocking capacitor as an active decoupling capacitor. The operating principles of the proposed modulation scheme are comprehensively analyzed, and an automatic decoupling control system based on large-signal modelling is designed. A 500 W experimental prototype is built for validation. Experimental results demonstrate that the proposed modulation method reduces the decoupling capacitor volume by 75% and decreases the double-line frequency pulsation in the input current by 96%, thereby improving the circuit’s power density and MPPT accuracy.
- 1 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 41 (1) , 1087-1101
A Modulation Method for Decoupling Capacitance Reduction in Single-Stage Dual-Active-Bridge Microinverters
Conventional single-stage dual active bridge (DAB) microinverters typically have a bulky capacitor for decoupling double-line frequency power and a significant pulsation in the input current, leading to reduced power density and degraded maximum power point tracking (MPPT) performance. This article presents a new modulation method for single-stage DAB microinverters, allowing for substantial reductions in both the required decoupling capacitance and the input current ripple without hardware modifications. This is achieved by repurposing the inherent primary-side dc-blocking capacitor as an active decoupling capacitor. The operating principles of the proposed modulation scheme are comprehensively analyzed, and an automatic decoupling control system based on large-signal modelling is designed. A 500 W experimental prototype is built for validation. Experimental results demonstrate that the proposed modulation method reduces the decoupling capacitor volume by 75% and decreases the double-line frequency pulsation in the input current by 96%, thereby improving the circuit’s power density and MPPT accuracy.
- 7 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Smart Grid
- Vol. 16 (6) , 5669-5672
A Tuning Friendly Deep Reinforcement Learning Method for Inverter-Based Volt/Var Control
Deep reinforcement learning (DRL) methods have been applied to power system problems in active distribution net-works, including the inverter-based volt/var control (VVC). How-ever, existing DRL-based VVC methods require hyperparameter fine-tuning, tailored algorithm selection and domain expert knowledge, depending on various power system characteristics. This letter proposes a tuning-friendly DRL method to approach optimal training performance, employing an evolution strategy in the DRL training process and thus enhancing exploration capabil-ity. The proposed method is applied to deep deterministic gradient policy and soft actor-critic, two widely used DRL algorithms. The advantages of tuning-friendly training, i.e., insensitiveness to hy-perparameter tuning, are validated by numerical experiments.
- 7 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Smart Grid
- Vol. 16 (6) , 5669-5672
A Tuning Friendly Deep Reinforcement Learning Method for Inverter-Based Volt/Var Control
Deep reinforcement learning (DRL) methods have been applied to power system problems in active distribution net-works, including the inverter-based volt/var control (VVC). How-ever, existing DRL-based VVC methods require hyperparameter fine-tuning, tailored algorithm selection and domain expert knowledge, depending on various power system characteristics. This letter proposes a tuning-friendly DRL method to approach optimal training performance, employing an evolution strategy in the DRL training process and thus enhancing exploration capabil-ity. The proposed method is applied to deep deterministic gradient policy and soft actor-critic, two widely used DRL algorithms. The advantages of tuning-friendly training, i.e., insensitiveness to hy-perparameter tuning, are validated by numerical experiments.
- 30 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 40 (9) , 13568-13583
Real-Time Optimal Control of Dual-Bridge Series Resonant DC–AC Converters
Control plays a critical role in achieving efficient operation of dual-bridge series resonant dc–ac converters (DBSRC). However, realizing real-time, optimal, and robust control of DBSRC to maximize its efficiency is generally difficult, due to the complex interplay of control variables, efficiency, power demand, and voltage conversion ratio requirements. Here, we propose an optimal control strategy for DBSRC, suitable for real-time closed-loop controller implementation. The key to the strategy is the development of a closed-form analytical solution of the optimal control variables for maximum system efficiency and output power regulation. Detailed derivation of the optimal control solution is provided. The effectiveness of the proposed control strategy is validated on a 200-W DBSRC prototype. The results confirm that the proposed control strategy can achieve real-time and robust control performance, with over 2% efficiency improvement over the conventional single-phase-shift control strategy.
- 30 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 40 (9) , 13568-13583
Real-Time Optimal Control of Dual-Bridge Series Resonant DC–AC Converters
Control plays a critical role in achieving efficient operation of dual-bridge series resonant dc–ac converters (DBSRC). However, realizing real-time, optimal, and robust control of DBSRC to maximize its efficiency is generally difficult, due to the complex interplay of control variables, efficiency, power demand, and voltage conversion ratio requirements. Here, we propose an optimal control strategy for DBSRC, suitable for real-time closed-loop controller implementation. The key to the strategy is the development of a closed-form analytical solution of the optimal control variables for maximum system efficiency and output power regulation. Detailed derivation of the optimal control solution is provided. The effectiveness of the proposed control strategy is validated on a 200-W DBSRC prototype. The results confirm that the proposed control strategy can achieve real-time and robust control performance, with over 2% efficiency improvement over the conventional single-phase-shift control strategy.
- Top Cited Papers
- 16 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Smart Grid
- Vol. 16 (4) , 3392-3407
Carbon-Aware Peer-to-Peer Energy Trading in an Unbalanced Distribution Network via a Nash Equilibrium Discovery Deep Reinforcement Learning Approach
Peer-to-peer (P2P) energy trading offers an innovative platform for prosumers to engage in bilateral electricity and carbon trades, enabling promoting the integration of distributed energy resources and supporting low-carbon operations in distribution networks. However, existing studies often rely on single-phase approximations, leading to inaccuracies in emission accounting and potential economic inefficiencies loss. To address these challenges, this paper proposes a carbon-aware P2P energy trading mechanism for unbalanced distribution networks. A novel joint electricity and carbon trading market framework is developed, integrating carbon permits trading with a continuous double auction mechanism. The trading process is formulated as a cooperative Markov game to capture the stochastic nature of market dynamics and to enable intelligent bidding decisions. A Nash equilibrium (NE) discovery multi-agent deep reinforcement learning framework is employed, utilizing a trust region policy optimization-based algorithm to enhance trading reliability and policy interpretability. Simulations on a modified IEEE-123 unbalanced distribution network demonstrate the proposed effectiveness in reducing operational costs and carbon emissions, providing rigorous NE validation and showcasing its practical applicability in complex energy markets.
- 16 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Smart Grid
- Vol. 16 (4) , 3392-3407
Top Cited PapersCarbon-Aware Peer-to-Peer Energy Trading in an Unbalanced Distribution Network via a Nash Equilibrium Discovery Deep Reinforcement Learning Approach
Peer-to-peer (P2P) energy trading offers an innovative platform for prosumers to engage in bilateral electricity and carbon trades, enabling promoting the integration of distributed energy resources and supporting low-carbon operations in distribution networks. However, existing studies often rely on single-phase approximations, leading to inaccuracies in emission accounting and potential economic inefficiencies loss. To address these challenges, this paper proposes a carbon-aware P2P energy trading mechanism for unbalanced distribution networks. A novel joint electricity and carbon trading market framework is developed, integrating carbon permits trading with a continuous double auction mechanism. The trading process is formulated as a cooperative Markov game to capture the stochastic nature of market dynamics and to enable intelligent bidding decisions. A Nash equilibrium (NE) discovery multi-agent deep reinforcement learning framework is employed, utilizing a trust region policy optimization-based algorithm to enhance trading reliability and policy interpretability. Simulations on a modified IEEE-123 unbalanced distribution network demonstrate the proposed effectiveness in reducing operational costs and carbon emissions, providing rigorous NE validation and showcasing its practical applicability in complex energy markets.