Heterogeneous industrial clusters such as refiners, data centers, and manufacturing facilities represent high-density loads that should maintain operational continuity during grid emergencies while providing flexible frequency recovery support. Existing strategies-conventional under-frequency load shedding, centralized, and independent controllers- either result in disproportionate energy curtailment across clusters, imposing heavy communication overhead, or are reactive in nature. This paper introduces a fault-triggered vulnerability-aware distributed Model Predictive Control (MPC) architecture in which each cluster autonomously optimizes its operation during normal conditions, and a consensus layer based on the Alternating Direction Method of Multipliers (ADMM) activates only when frequency deviates from a defined threshold. A hysteresis approach prevents spurious activation and deactivation around the threshold. The introduced vulnerability-aware mechanism results in an adaptive consensus that redistributes the burden away from the clusters approaching cascading failure limits. Voltage constraints are enforced via linearized sensitivities derived from a Newton-Raphson AC power flow. Validation across a 24-hour simulation with three heterogeneous faults, four comparison baselines, five uncertainty scenarios, and cluster counts up to 10, demonstrated that the proposed framework reduces energy curtailment by 68% and 56% relative to independent and centralized MPC, respectively. The Jain fairness index of 0.983 confirms that the proposed architecture results in optimal burden distribution across heterogeneous clusters. The ADMM coordination operates with a duty cycle of 11.2% only, and the convergence rate improves from 72% to 85% from two to ten clusters, illustrating scalability. All frequency deviations remain within operational bounds well above the under frequency relay thresholds defined in IEEE Std C37.117-2007.
Nanocrystalline alloys, featuring high saturation flux density and low hysteresis loss, have become promising core materials for high-frequency transformers. However, the magnetization dynamics and hysteresis loss formation mechanisms of nanocrystalline alloys under kHz alternating magnetic fields are still not fully understood. To address this issue, a three-dimensional micromagnetic model of FINEMET nanocrystalline alloy is established in OOMMF and validated by static magnetization characteristics and dynamic hysteresis loss experiments. Based on the validated model, the magnetization response, domain evolution, energy conversion, and damping dissipation behaviors under high-frequency sinusoidal excitation are systematically investigated. The results show that the magnetization reversal process can be divided into five dynamic stages: initial stabilization, energy pre-accumulation and magnetization relaxation, domain nucleation and early expansion, domain-wall-dominated fast reversal, and saturation approach with structural adjustment. Energy evolution and instantaneous dissipation power analyses indicate that the energy dissipation within one cycle is mainly concentrated in the stages of domain nucleation, rapid domain-wall propagation, and local structural adjustment before saturation, with the domain-wall-dominated fast reversal stage contributing the largest loss. In addition, the hysteresis loss increases with both excitation amplitude and frequency in the simulations. Under the present micromagnetic model and parameter set, a model-predicted transition region near 28 mT is observed, corresponding to the transition of the reversal mechanism from rotation-dominated behavior to domain-nucleation- and domain-wall-motion-dominated behavior. The obtained results provide theoretical support for loss analysis of high-frequency transformer cores and optimization design of magnetic materials.
This paper presents a robust sensorless control strategy for medium-voltage (MV) induction motor (IM) drives operating through long feeder cables, as encountered in subsea and harsh-environment industrial applications. In these systems, severe voltage deviations at the motor terminals and the im practicality of installing local sensors pose major challenges to conventional control schemes. To address these issues, an H2–linear matrix inequality (LMI) full-state feedback controller is developed, using a polytopic uncertainty model to ensure robustness against large cable-parameter uncertainties. The scheme includes an extended Kalman filter (EKF) to estimate all required states and the rotor speed from inverter-side measurements only. The effectiveness of the proposed method is validated through full-scale MV simulations and experimental tests on a LV laboratory setup.
With the continuous development of aircraft technology, the power density and safety of generators are required to be higher in the aircraft power system. This paper investigates a hybrid excitation starter generator with radial additional air-gaps (HESG-RAA), which has high power density, a wide flux-regulation range, and the ability to suppress short-circuit current. The topology design and equivalent magnetic circuit of the HESG-RAA are introduced. In addition, the electromagnetic performance of the investigated HESG-RAA is analyzed, including no-load characteristics, load characteristics, and short-circuit characteristics. The mechanical strength of the rotor of the HESG-RAA is analyzed at the highest speed. Finally, a 90 kW HESG prototype is manufactured, and the test results verified the effectiveness of the HESG-RAA structure and flux-regulation principle.
Overcharging in lithium-ion batteries (LIBs) is a critical safety challenge, potentially leading to thermal runaway and catastrophic failure. The current study deeply investigates the degradation mechanisms of NMC/graphite cells in overcharge conditions up to 150% State of Charge (SoC). A detailed failure progression is characterized by correlating real-time voltage and surface temperature data with periodic Electrochemical Impedance Spectroscopy (EIS) measurements. The results reveal a sequence of events starting from lithium plating, dendritic growth, and finally a minor internal short circuit (ISC). A key finding is the identification of a clear failure signature: a sudden voltage drop that occurs at approximately 128–130% SoC, which coincides with a sharp, five-fold acceleration in the cell's temperature rise rate. Analysis of the impedance data provides a microscopic validation of this process, showing an initial increase in charge transfer resistance (RCT) due to plating, followed by a collapse of RCT and a sharp increase in ohmic resistance (RS) at the moment of the ISC. A comparative analysis of two cells confirms this failure pathway's reproducibility while highlighting the stochastic nature of the final short-circuit event's severity. A simple correlated voltage-temperature gradient based detection strategy has been proposed for safety monitoring of LIBs.
Windings are the fundamental components of electric machines. Conventional winding analysis tools include slot diagrams, star diagrams, phasor diagrams, winding functions, etc. This paper presents a unified computational tool for winding analysis that encompasses the essential idea behind these con ventional methods and at the same time facilitates computations. The idea is based on the relationship among multiple concepts specifically associated with electric machine windings, including magnetic, electrical, layer, and phase orders. Such a relationship can be described through multi-dimensional arrays. Five example windings, namely fractional slot, full-pitch, short-pitch, concentric, and hairpin windings, are presented to illustrate the pro posed framework. Furthermore, various applications scenarios, including skewing, number of turns, toroidal windings, winding balance, and end winding are discussed to show its flexibility. The representation of generic windings are also given and it is used to analyze the upper bound of possible winding configurations. Lastly, the new representation enables the inverse design of windings—a step towards EM design automation. Several key inverse design algorithms are presented and validated.
Distributed photovoltaic (DPV) generation plays an important role in the economic and low-carbon operation of modern power systems. However, the widespread data incompleteness, arising from highly dispersed sites and heterogeneous measurement conditions, poses a significant challenge to the accuracy of large-scale DPV power forecasting. To achieve accurate power forecasting with incomplete data, we extend DPV clustering using the Probability Mass Similarity Kernel (PMK) and propose a Geo-PMK dual-scale dynamic clustering method. DPV stations are initially clustered based on geographic location, followed by dynamic refinement using PMK to capture time-series similarity. From each resulting sub-cluster, a representative station is selected to characterize regional power output for forecasting. Subsequently, a Prior-Gated and Posterior-Calibrated Mixture of Experts (PP-MoE) ensemble forecasting framework is proposed. This framework dynamically integrates heterogeneous experts via a gating mechanism that fuses prior information with posterior error calibration, utilizing Top-K sparse routing for efficient forecasting. Finally, the forecasting results are scaled by capacity ratios and aggregated to obtain the total regional output. The effectiveness of the proposed method is validated using real-world data from 274 stations in Lanzhou, China. Results demonstrate that this method can maintain robustness under incomplete data scenarios and significantly improve the regional power forecasting accuracy of large-scale DPV systems.
Continuousoperation of power systems is critical to modern society, even during unexpected or emergency situations. The core method to evaluate the security of power systems is contingency analysis, a computationally-intensive process which considers how a set of potential scenarios would affect the electrical grid. This paper presents a robust security-evaluation method which enables a topology- and parameter-free estimation of an existing system-level security metric, the system aggregate megawatt contingency overload (SysAMWCO), by leveraging a deep-learning-based approach. Using the results of conventional AC-based contingency analysis for a simulated large-scale power system, a deep neural network (DNN) is trained to estimate SysAMWCO values from only bus loads, generator setpoints, and branch-connection statuses as inputs. Although DC-based analysis only considers fewer than 200 contingencies after screening (down from over 5,000), it is outperformed by this DNN framework with 10-times lower error and 30-times faster SysAMWCO evaluation time. Given these results, this framework demonstrates significant promise as a computationally-efficient surrogate for DC-based contingency-analysis and power-security evaluation in high-fidelity simulation environments with a clear pathway for future validation against industry-operational datasets.
In this study, we investigated methods to reduce carbon particles including soot in exhaust gas emissions and black carbon particles in the atmosphere. We evaluated novel designs of surface dielectric barrier discharge (SDBD) units, which have a set of discharge electrodes on their surface. We tested three types of connections with different discharge and particle incineration characteristics, including a type with a floating sub-electrode, one in which an identical voltage was applied to both the main and the sub-electrode, and another in which an external inductor and a capacitor were added. In the experimental evaluation, we compared the performance of these units in reducing carbon and analyzed the electrical behaviors of the sub-electrode using an equivalent circuit model.
Particulate matter concentrations in underground subway tunnels are often elevated due to mechanical wear and limited natural ventilation, posing potential health risks to passengers and workers. This study developed and field-validated a two-stage brush-type electrostatic precipitator integrated into subway ventilation systems to mitigate PM concentrations under real operating conditions. The system was installed in both supply- and exhaust-type ventilation rooms across interconnected tunnel sections. The ESP achieved removal efficiencies exceeding 70–90% for PM2.5 and PM10 in one pass operation, and its performance varied according to ambient particle size characteristics. Integration of the system into supply-type ventilation rooms resulted in removal of approximately 50–60% of incoming outdoor PM, corresponding to in-tunnel PM2.5 concentration reductions of 15–32% despite dynamic airflow and train induced turbulence. In exhaust-type configurations, removal efficiencies ranged from 66% to 71%, and treated exhaust air exhibited lower PM concentrations than the surrounding ambient atmosphere. The results demonstrate that integration of brush-type ESP systems into subway ventilation infrastructure can effectively reduce particulate concentrations within tunnels while limiting the emission of tunnel-derived particles to the urban environment.
The electrification of maritime electrical power systems has progressed from isolated onboard upgrades to integrated shipboard power systems, shore-to-ship interfaces, and port microgrids operating under strict safety, reliability, and regulatory constraints. In parallel, artificial intelligence, machine learning, and advanced control techniques have been increasingly reported in literature to enhance energy management, resilience, and operational efficiency. However, their technical suitability and deployment readiness remain insufficiently assessed from a system-level and operational perspective. This paper presents a system-level technical assessment of control, optimization, and AI/ML-based approaches for maritime power systems, with exclusive focus on IEEE Transactions on Industry Applications. Using a rigorously selected dataset of forty-two journal papers, the study classifies methods by maritime system context, operational layer, and method role, audits formulation transparency, and evaluates deployment realism under non-ideal operational assumptions. To support more structured comparison, a semi-quantitative rubric is used to summarize how explicitly formulation and deployment realism criteria are addressed across method classes. The results show that optimization and model predictive control dominate safety-critical decision layers due to explicit constraint enforcement, while learning-based methods are limited to auxiliary roles such as forecasting, monitoring, and decision support. The rubric-based comparison further indicates that optimization, MPC, and hybrid AI–optimization approaches achieve the strongest formulation scores, whereas supportive learning-based methods remain weaker in deployment realism because sensing non-idealities, communication constraints, and real-world feasibility are rarely treated explicitly.
Medium voltage DC (MVDC) systems represent a key enabling technology for large-scale and far-offshore wind projects and have gained extensive research attention. The paper aims to assess the costs and benefits of offshore wind farms (OWFs) connected with MVDC systems and tackle the residual DC fault currents existing in MVDC systems after blocking the converters. A comprehensive cost-benefit analysis (CBA) is conducted for OWFs employing parallel and series MVDC collection topologies and their techno-economic performance is compared with that of conventional high-voltage AC (HVAC) schemes. The CBA results verify that MVDC systems achieve lower lifecycle costs and higher efficiency, proving to be a competitive solution for offshore wind power transmission. This study provides guidance for offshore wind power developers in selecting suitable system topologies. Furthermore, this paper proposes a novel fault current suppression circuit (FCSC) to rapidly eliminate residual DC fault currents, which realizes fast fault isolation using low-cost DC switches (DCSs). The proposed scheme shortens the fault clearance time from about 500 ms to only 62 ms, which greatly improves power supply reliability and offers significant economic value, particularly in applications where reliable power supply is critical.
To address the challenges of achieving accurate short-term load forecasting in distribution networks under high load variability and stochastic user behavior, this study proposes a forecasting framework integrating sample entropy-guided secondary decomposition and deep learning. First, the Random Forest algorithm is employed to identify and select the most influential exogenous features affecting load variations. Subsequently, Time-Varying Filtering Empirical Mode Decomposition is applied to perform an initial decomposition of the load series into multiple components. To improve input quality, components with high complexity, as quantified by Sample Entropy, are further decomposed using Singular Spectrum Analysis. Next, the processed sub-sequences and selected features are fed into a bidirectional long short-term memory network, whose hyper-parameters are optimized using an Improved Dung Beetle Optimizer. Finally, the forecasts of all decomposed components are recombined to reconstruct the overall load forecast. Case studies indicate that the proposed approach improves short-term load forecasting accuracy compared with representative benchmark methods for distribution networks.
In this study, two laboratory-scale wet ESP geometries were compared under high face velocity operation. The SWEP exhibited frequent spark discharges and significant efficiency loss, achieving only 75% total suspended particle (TSP) removal efficiency at 8 m·s-1, with a distinct dip to 60% efficiency in the 0.2–0.6 $\mu$m range. In contrast, the PWEP demonstrated superior performance, suppressing spark frequency and achieving over 97.9% TSP removal efficiency at the same condition. Size-resolved analysis confirmed that the PWEP maintained submicron efficiencies above 84%, while computational fluid dynamics simulations revealed a more uniform electric field distribution and stable particle trajectories. These findings indicate that the PWEP geometry provides a more robust design for high-velocity.
Energy Hubs are widely used to connect different energy carriers to the power grid. These structures are commonly based on multiple converters, requiring high communication and control effort. Therefore, the Modular Multilevel Converter (MMC) based Energy Hub is a promising structure due to its capability to integrate multiple renewable energy sources and storage systems into a single converter. However, the MMC-based Energy Hub introduces several technical challenges, including inter-arms vertical imbalance. The proposed work focuses on the vertical imbalance that can occur in the MMC arms when energy sources and storage systems operate unevenly. Indeed, in this case, an internal power flow occurs between the two arms that is directly responsible for a larger circulating current and for an increased difference in the arm currents. Consequently, the more loaded arm is affected by larger power losses and more severe electrical stress of the semiconductors. This work proposes an innovative control strategy that leverages the fundamental frequency circulating current (FFCC) to control the active power flow from high-load arms to low-load arms. Additionally, the proposed strategy balances arm currents and ensures an even distribution of total power losses. The method is supported by an analytical formulation that relates the voltage imbalance of the arms to the amplitude and phase of the FFCC. Results of experimental tests and real-time simulation on a thirteen level MMC are presented to demonstrate the effectiveness of the proposed technique.
Hybrid-magnet multilayer interior permanent magnet synchronous machines (IPMSMs) offer reduced rare-earth magnet usage while maintaining high torque density; however, flux crowding and localized saturation in multilayer rotor structures can lead to increased core loss and torque ripples. This paper investigates hole-based rotor modification in a double U-shaped hybrid-magnet IPMSM to improve flux distribution and electromagnetic performance. A magnetic equivalent circuit (MEC) interpretation is employed to explain the influence of pole-arc hole insertion on air-gap flux and rotor saturation behaviour. Finite-element analysis (FEA) is conducted to evaluate the effects of hole location, number, size, and arrangement on torque ripple, core loss, efficiency, and mechanical integrity. Based on this investigation, a multi-objective optimization using the non-dominated sorting genetic algorithm II (NSGA-II) is performed to determine the optimal hole geometry under electromagnetic and mechanical constraints. A 1-kW prototype with a hybrid-magnet-based, Holes based topology is fabricated, and measured back-EMF and MTPA current-controlled operation validate the proposed rotor modification approach.
This paper proposes an analytical framework for calculating the maximum mechanical stress (MMS) of a single and dual-layered V-shape interior permanent magnet synchronous machine (IPMSM). For this, an analytical model based on the centrifugal force calculation is used for estimating the average bridge stress of a single-layered IPMSM. Building on this, a modified analytical model is developed to calculate the average bridge stress considering dual-layered V-shaped IPMSM. To account for the MMS, a stress concentration factor (SCF) is proposed based on the V-shaped angle. The accuracy of the developed analytical model, for both the average bridge stress and MMS, is validated through detailed finite element analysis (FEA). Additionally, a tensile experiment is proposed to validate the novel dual-layer V-shaped PMSM analytical model. It was observed that the lamination fractured in the assumed inner central bridge hotspot validating the hypothesis determined from FEA and it was determined through the comparison with the proposed analytical model that the maximum percentage error was approximately 4%, validating the model.
This paper proposes a robust design framework for surface-mounted permanent-magnet synchronous motors that simultaneously accounts for geometric manufacturing tolerances and magnet remanence variation. Incorporating remanence as a direct surrogate input requires re-evaluating all training designs at multiple remanence levels, incurring a prohibitive finite-element analysis (FEA) cost. To avoid this, the back electromotive force (back EMF) peak, an existing FEA output proportional to remanence, is repurposed as a physics-based proxy input to a deep neural network surrogate originally trained on geometric design variables. This proxy input reduces root-mean-square errors by 19–54% across all five objectives relative to a geometry-only baseline, while a targeted data augmentation requiring approximately 4.1% additional FEA evaluations enables the surrogate to generalize across the remanence range; variance decomposition across the Pareto front confirms that remanence is the dominant uncertainty source. A set-based multi-criteria ranking method then ranks candidate designs by integrating nominal performance with worst-case and standard-deviation robustness metrics. Experimental validation on a fabricated 4-kW prototype, covering the no-load tests and the torque–current characteristic, confirms that the measured quantities fall within the predicted uncertainty bounds, demonstrating the practical effectiveness of the proposed framework.
Load frequency control (LFC) is one of the most important functions in modern power systems. LFC balances power generation and power demand while maintaining frequency within a nominal range under continuously varying random loads. To improve control and measurements, LFC introduces digital communication and computer-based control. This makes power systems increasingly vulnerable to cyberattacks, posing significant threats to grid security and reliability. To mitigate these challenges, a nonlinear active disturbance rejection control (NLADRC)-based load frequency control approach is proposed for multi-area power systems. The proposed approach is investigated for various types of load-altering and denial of-service (DoS) attacks, physical disturbances, physical system nonlinearities and communication time delays. Moreover, the system states and generalized disturbances are estimated in real time using a third-order nonlinear tracking differentiator, a nonlinear extended state observer, and a nonlinear state error feedback law, which is designed to actively compensate for the lumped disturbances without the need for accurate plant models or explicit attack detection. Furthermore, a Lyapunov based stability analysis is done for the closed-loop stability under bounded cyber-physical disturbances. Moreover, various cyber-attack scenarios are extensively simulated on multi-area power system models with non-reheat, reheat and hydro turbines. The proposed method results are compared with well-known approaches, such as internal model control–based proportional integral-derivative (IMC–PID), linear active disturbance rejection controller and generalised active disturbance rejection controller, which demonstrate better frequency resilience, faster dynamic response and better robustness than existing methods. Finally, the effectiveness and practicality of the proposed control scheme are validated through real-time hardware-in-the-loop testing on the OPAL-RT platform.
