Rasool Peykarporsan

Low-voltage solid state DCCB design based on bypassed bidirectional thyristor-capacitor suppressor

M. Moradian, R. Peykarporsan, T. T. Lie, K. Gunawardane

IEEE Transactions on Power Electronics · 2024

In this work we propose a low-voltage solid-state DC circuit breaker (SS-DCCB) based on a novel bidirectional thyristor-capacitor (BiTriCap) suppressor. The motivation is that DC microgrids need fast, reliable current interruption with low overvoltage stress, and conventional MOV-based protection leaves measurable headroom on both metrics. The proposed design uses parallel snubber capacitors that absorb the switching transient and release the stored energy during the next switch operation, redirecting the surge into the bypassed capacitor and producing a clean zero-current crossing. The model accounts for both line and load inductances, and an ARM microcontroller manages the switch sequencing in real time. We validated the breaker experimentally at 48 VDC / 8 A — a design that scales to higher voltage and current ranges — and the prototype showed only 2 V of residual overvoltage across the main switch, with experimental waveforms matching the simulations across switching and fault scenarios.

Highlights

• Novel bidirectional thyristor-capacitor (BiTriCap) suppressor for solid-state DC circuit breakers in DC microgrids.
• Parallel snubber capacitors absorb the switching transient and release the stored energy on the next operation, producing a clean zero-current crossing.
• Real-time sequencing of the solid-state switches managed by a programmed ARM microcontroller.
• Experimentally validated at 48 VDC / 8 A with only 2 V residual overvoltage across the main switch; design scales to higher voltage and current ranges.

Bidirectional Solid State Circuit Breaker with Passive Surge Absorber for LV Applications

M. Moradian, R. Peykarporsan, T. T. Lie, K. Gunawardane

IEEE Journal of Emerging and Selected Topics in Power Electronics · 2025

In this work we propose a low-voltage solid-state circuit breaker (SSCB) based on a novel bidirectional passive technique (BPT) built around an RCD topology. The motivation is that DC microgrid protection benefits from breakers that are simple, bidirectional, and fully passive — limiting active components to the primary switch reduces cost, control complexity, and failure modes. The article details the working principles, operating modes, and the key design equations for the time intervals, voltages, and currents involved. The design was validated experimentally and in MATLAB at 48 VDC with a 10 A threshold fault current (Future Architecture of Network Project), achieving 20 µs to fully interrupt the load, 40 µs switching time, and 90 µs total fault clearing — with minimal surge voltage across the main switch.

Highlights

• Bidirectional passive technique (BPT) SSCB based on an RCD topology — only the primary switch is active.
• Fast interruption: 20 µs full load interrupt, 40 µs switching time, 90 µs total fault clearing.
• Rated for 48 VDC / 10 A (Future Architecture of Network Project).
• Validated by both MATLAB simulation and an experimental prototype.

Hierarchical switch fault diagnosis based on transformer algorithm in four-leg inverters of stand-alone wind energy conversion systems

J. Heidari, R. Peykarporsan, S. Oshnoei, T. T. Lie, L. Vandevelde, G. Crevecoeur

International Journal of Electrical Power & Energy Systems · 2026

In this work we propose a hierarchical Transformer-based fault diagnosis scheme for four-leg inverters used in stand-alone wind energy conversion systems. The motivation is that four-leg inverters are the standard solution for handling unbalanced loads in stand-alone wind setups, but their switching devices remain vulnerable to internal faults — and reliably identifying both the fault type (open-circuit vs. short-circuit) and the specific faulty switch is non-trivial under varying operating points. The proposed model uses a two-level hierarchical Transformer that first classifies the fault type and then localises the faulty switch. Datasets covering a wide range of operating scenarios were generated on an OPAL-RT hardware-in-the-loop setup to mimic real conditions. The results show that the Transformer-based diagnosis outperforms state-of-the-art machine learning algorithms, and the hierarchical two-level structure is consistently more effective than a single-level approach.

Highlights

• Hierarchical two-level Transformer-based fault diagnosis for four-leg inverters in stand-alone wind energy systems.
• Detects and classifies both open-circuit and short-circuit switch faults, then localises the specific faulty switch.
• Datasets generated with an OPAL-RT hardware-in-the-loop simulator across diverse operating points.
• Outperforms state-of-the-art ML baselines and a single-level structure on classification accuracy.

Adaptive mixed time-state dependent distributed event-triggered consensus protocol of a DC microgrids cluster

Z. H. A. Al-Tameemi, R. Peykarporsan, T. T. Lie, R. Zamora, F. Blaabjerg

Electric Power Systems Research · 2025

In this work we propose the Adaptive Mixed Time-State Dependent Distributed Event-Triggered Consensus protocol (AMDETC) for the global control layer of a DC microgrid (MG) cluster. The motivation is that traditional consensus-based secondary control demands frequent communication between MGs, which makes the cluster sensitive to communication delays and limits scalability. The proposed protocol triggers updates only when a mixed time- and state-dependent condition is satisfied, sharply reducing the inter-MG communication load. We pair AMDETC with an adaptive fixed-time consensus algorithm (AFTA) — augmented by a saturation function — to accelerate current-sharing convergence, and use the Grey Wolf Optimizer (GWO) to tune the PI controllers in the secondary layer. The approach was validated in MATLAB on a four-DC-MG cluster and on OPAL-RT under realistic conditions, achieving 100% triggering accuracy, current convergence within 0.02 s, fast voltage recovery, and accurate proportional current sharing under load changes and fault scenarios.

Highlights

• Novel AMDETC event-triggered consensus protocol cuts inter-MG communication while maintaining cluster stability.
• Adaptive fixed-time consensus algorithm (AFTA) with a saturation function accelerates current-sharing convergence.
• Grey Wolf Optimizer (GWO) tunes the secondary-layer PI controllers for improved disturbance rejection.
• Validated on a four-DC-MG cluster in MATLAB and on OPAL-RT: 100% triggering accuracy, current convergence within 0.02 s.

A Novel Model-Free Defense Scheme for Power Systems Stability Under Cyber Attacks

S. Oshnoei, R. Peykarporsan, J. Heidari, E. Mahboubi-Moghaddam, T. T. Lie

IET Generation, Transmission & Distribution · 2026

In this work we propose a model-free resilient defence scheme to protect the load frequency control (LFC) loop of a power system against false data injection (FDI) cyber-attacks. The motivation is that LFC depends on communication networks and is therefore exposed to FDI attacks, while existing defences require either a precise mathematical model of the system or extensive historical data — neither of which is always available. The proposed scheme combines a model-free observer that estimates the targeted frequency signal directly from measurement and control signals, a detector that compares the residual against a predefined threshold to flag attacks, and an event-triggered mechanism that — on detection — blocks the falsified signal and forwards the estimated one to the LFC controller. The observer's parameters are tuned by a deep reinforcement learning (DRL) algorithm. The scheme was validated in real time on Kundur's 4-unit-12-bus system using OPAL-RT and completely mitigated FDI attacks on the frequency response across multiple scenarios — with a design that is independent of the system's mathematical model, historical data, size, and complexity.

Highlights

• Model-free resilient defence scheme protecting load frequency control (LFC) against false data injection (FDI) cyber-attacks.
• Combines a model-free observer, a residual-vs-threshold detector, and an event-triggered mitigation mechanism.
• Deep reinforcement learning (DRL) tunes the observer parameters; no system model or historical data required.
• Validated in real time on Kundur's 4-unit-12-bus system with OPAL-RT — full mitigation across multiple attack scenarios.

Data-Driven Parameter Identification of Synchronous Generators: A Three-Stage Framework with State Consistency and Grid Decoupling

R. Peykarporsan, T. Govinda Waduge, T. T. Lie, M. Stommel

Sensors · 2026

In this work we propose a three-stage data-driven framework for identifying the parameters of synchronous generators in a Port-Hamiltonian (PH) modelling form. The motivation is that PH frameworks offer inherently stable, energy-consistent, and modular representations for power-system stability analysis, but their practical deployment has been held back by the absence of high-fidelity, PH-compatible parameter identification methods that work from sensor measurements. The widely used IEEE Standard 115 procedure has fundamental limitations in this setting — single-scenario identifiability, noise-sensitive derivative-based formulations, and an inability to decouple generator-internal damping from grid contributions. Our framework resolves these issues with three contributions: a differential damping technique that separates generator damping from grid damping without isolating the machine, a derivative-free state-consistency optimisation that draws on 15 diverse excitation scenarios, and a physics-based regularisation that enforces PH structure so the identified model remains compatible with Lyapunov-based stability certification. We identified the eight key generator parameters (H, D, X_d, X_q, X′_d, X′_q, T′_do, T′_qo) with errors between 1.26% and 9.10%, and validated the resulting model with RMS rotor-angle errors below 1.2° and speed errors below 0.15% — accuracy sufficient for transient stability analysis, passivity-based control design, and oscillation damping assessment, using only standard simulation tools and terminal measurements.

Highlights

• Three-stage data-driven framework for Port-Hamiltonian (PH) parameter identification of synchronous generators.
• Differential damping technique decouples generator-internal damping from grid contributions without isolating the machine.
• Derivative-free state-consistency optimisation across 15 diverse scenarios; physics-based regularisation preserves PH structure for Lyapunov-based stability certification.
• Identifies 8 generator parameters with 1.26%–9.10% error; RMS rotor-angle error < 1.2°, speed error < 0.15%.

Mitigating GFM-Induced Battery Stress with a VDCM-Controlled Hybrid Energy Storage

R. Peykarporsan, A. E. Rabbi, T. T. Lie, M. Stommel, J. Watson

IEEE PES International Meeting (PES IM) · 2026

In this work we propose a degradation-aware voltage-dependent current management (VDCM) strategy for a battery–supercapacitor hybrid energy storage system (HESS) supporting a grid-forming (GFM) inverter. The motivation is that as power systems become dominated by inverter-based resources, GFM converters take on stability duties that previously belonged to synchronous machines — and the battery behind a GFM inverter ends up exposed to high C-rates and severe current fluctuations during faults and sudden load changes, which accelerate degradation. The proposed VDCM strategy exploits the complementary characteristics of the two storage elements: the supercapacitor absorbs the fast transients (its high power density and low ESR are well suited to this role), while the battery handles the slower, energy-dense portion of the load. Simulations under a short-circuit fault and a sudden load change show that the strategy regulates voltage and frequency, significantly reduces both C-rate violations and current-fluctuation stress on the battery, and slightly improves GFM converter performance thanks to the supercapacitor's lower equivalent series resistance.

Highlights

• Degradation-aware VDCM strategy for a battery–supercapacitor HESS supporting a grid-forming (GFM) inverter.
• Offloads rapid transients to the supercapacitor to mitigate battery C-rate violations and current-fluctuation stress.
• Validated under short-circuit and sudden-load-change scenarios — voltage and frequency regulation preserved.
• Side benefit: slightly improved GFM converter performance due to the supercapacitor's lower ESR.

Model order reduction for control and stability analysis of complex dynamical systems in a DC microgrid

R. Peykarporsan, T. T. Lie, M. Stommel, J. Watson

IEEE Power & Energy Society General Meeting (PESGM) · 2024

In this work we study how to control a high-order complex nonlinear system using a simpler reduced-order model, with the goal of cutting computational and hardware requirements without sacrificing performance. The motivation is that controllers designed directly on full-order models of complex dynamical systems are often expensive to implement on real hardware, while a reduced-order surrogate — if accurate enough — lets us design simpler controllers that still work on the original plant. We applied three classical model-order reduction (MOR) techniques — modal truncation, balanced truncation, and Hankel-norm approximation — to a six-state nonlinear sample system, reducing it to four states, and then designed linear quadratic regulators (LQR) on each reduced model. The LQR controllers were evaluated in MATLAB on the original full-order system. The results show that the Hankel-norm approximation outperforms the other MOR methods on accuracy and robustness, and that an LQR designed on the reduced model can be applied directly to the original system without modification.

Highlights

• Reduced-order modelling pipeline for designing LQR controllers on high-order nonlinear systems.
• Compared three MOR methods: modal truncation, balanced truncation, and Hankel-norm approximation.
• Six-state nonlinear sample system reduced to four states; LQR controllers tested on the original plant.
• Hankel-norm approximation gave the best accuracy and robustness; the reduced-model LQR works directly on the full-order system without modification.

State identification of DC microgrids using probabilistic methods in the presence of noise

L. Kamyabi, R. Peykarporsan, T. T. Lie

IEEE PMAPS · 2024

In this work we investigate three input-output system identification methods — Balanced Proper Orthogonal Decomposition (BPOD), the Eigensystem Realization Algorithm (ERA), and Observer/Kalman Filter Identification (OKID) — applied to a six-state DC microgrid model. The motivation is that real-world deployment of model-based control depends on identification methods that remain robust when sensor measurements are noisy, but evaluations in the literature often focus on idealised conditions. We exposed the identification process to noise during data collection in MATLAB and compared the three methods under both clean and noisy conditions. The results show that BPOD has the highest noise tolerance in simulation but is constrained in real-world deployment, whereas OKID strikes the best balance of robustness and accuracy, making it the more viable choice for practical system identification.

Highlights

• Compared three input-output model identification methods (BPOD, ERA, OKID) on a six-state DC microgrid.
• Identification process evaluated under both clean and noisy conditions to bridge simulation and real-world deployment.
• BPOD showed the highest noise tolerance in simulation but has practical implementation constraints.
• OKID delivered the best balance of robustness, accuracy, and real-world viability.

A modified Class-D resonance inverter for operating in load-independent condition

R. Peykarporsan, H. M. Kazemi, M. Aeini, E. Afjei

11th Power Electronics, Drive Systems, and Technologies Conference (PEDSTC) · 2020

In this work we propose a modified Class-D resonant inverter topology that maintains zero-current switching (ZCS) under wide load variations. The motivation is well-known in resonant power electronics: the resonant frequency depends on the load, so when the load changes the circuit drifts between inductive and capacitive operation, the switch current ends up either lagging or leading, and the soft-switching property — the whole reason to choose a resonant topology in the first place — is lost. We add an auxiliary circuit to a Class-D power stage that decouples the switch current from the load, so ZCS is preserved across the entire operating range. The detailed simulations confirm the behaviour: at the IGBT switching instants, di/dt drops from a theoretically unbounded value to about 10⁶ A/s, placing the switches firmly in the soft-switching zone, with strong performance demonstrated at both light- and heavy-load extremes — a truly load-independent resonant inverter that also reduces fabrication cost, total losses, and converter size.

Highlights

• Modified Class-D resonant inverter topology that maintains zero-current switching (ZCS) under wide load variation.
• Auxiliary circuit decouples the switch current from the load, eliminating the load-dependence limitation of standard resonant inverters.
• Simulation: di/dt at IGBT switching reduced from an unbounded value to ~10⁶ A/s, placing switches in the soft-switching zone.
• Lower fabrication cost, reduced total losses, smaller converter footprint; verified under both light- and heavy-load conditions.