Current sensor offset and scaling errors degrade the current regulation accuracy and torque performance of predictive current control in permanent magnet synchronous motor (PMSM) drives. Since the two error types exhibit different signal structures, simultaneous online compensation remains difficult for existing methods. This paper proposes an adaptive disturbance-observer-based current sensor error compensation method for PMSM drives. A unified error model is established to show that offset errors behave as constant or slowly varying terms, whereas scaling errors appear as rotor-position-dependent periodic disturbances. Based on this difference, an adaptive observer with error-type-specific update laws is developed, where an integral-type law is used for offset estimation and a rotorsynchronized law is adopted for scaling estimation. In addition, a composite current control structure combining deadbeat predictive current control and a PI regulator is designed to reduce the contamination of the estimation channel by parameter mismatches and inverter nonlinearities. The observer error system is proven to be uniformly ultimately bounded. Experimental results under steady-state, dynamic, and parameter-mismatch conditions verify that the proposed method reduces current ripple and phase-current harmonic distortion while enabling direct online compensation without multi-stage signal injection.

Though recent literature consists of various optimal triple-phase-shift (TPS) modulation techniques for dual-active-bridge (DAB) converter, yet there is a research gap in terms of optimal closed-form solution implementation for mission-profile specific operation. A satellite payload’s mission-profile $\left( {{{\mathcal{M}}_{PL}}} \right)$ specific optimal closed-form solution for DAB converter operation at minimum conduction loss is proposed in this paper. The conduction loss optimization algorithm, closed-form modulation solution derivation and ${{\mathcal{M}}_{PL}}$ specific inductor optimization are elaborately presented. An experimental validation is performed on a 400 W, 500 kHz DAB module for a satellite payload with a defined ${{\mathcal{M}}_{PL}}$. The results and analysis reveal ~1% reduction in weighted conduction loss for a typical ${{\mathcal{M}}_{PL}}$ with 0.9 dwell-time in low-power standby, while adopting the proposed TPS optimization, compared to traditional TPS conduction loss minimization, and 49.1%, 16.2%, 32.2% reduction compared to respective conduction loss minimized SPS, DPS, EPS modulations.

Simultaneous integration of solar photovoltaic (SPV) generation and electric vehicle (EV) charging creates competing overvoltage and undervoltage conditions that stress meshed distribution systems and complicate realtime voltage stability assessment. Conventional techniques such as Predictor-Corrector Continuation Power Flow (CPF) and Jacobian-Based Sensitivity Analysis (JBSA) provide accurate stability margins but are computationally intensive and less suitable for rapid operational monitoring. This study proposes an artificial neural network (ANN)-based framework for real-time voltage stability assessment in a meshed distribution system under combined SPV and EV penetration. The model simultaneously estimates the Voltage Stability Index (VSI), Load Availability Margin (LAM), and Generation Acceptability Margin (GAM) for 1698 operating scenarios derived from a modified IEEE 33-bus meshed system. The optimized ANN achieves a mean squared error of 0.00039729 and a regression coefficient of approximately 0.999, indicating high predictive accuracy. In terms of computational performance, the ANN requires only 0.15 min of execution time and 2.99 MB of memory, whereas CPF requires 0.728 min and 28.64 MB, and JBSA requires 1.185 min and 7.76 MB. This corresponds to the ANN being approximately 79% faster than CPF and 87% faster than JBSA, while reducing memory usage by 89.56% compared to CPF (using 10.44% of CPF's memory) and by 61.47% compared to JBSA (using 38.53% of JBSA's memory). The ANN maintains comparable accuracy to CPF and JBSA, with mean absolute percentage errors (MAPE) of 4.282% and 2.822%, respectively. Furthermore, the framework reliably identifies Bus-18 as a critical weak bus due to multiple loop power redistributions, a vulnerability often overlooked by conventional approaches. These results demonstrate that the proposed ANN method enables accurate, lightweight, and near real-time voltage stability monitoring suitable for practical operational deployment in meshed distribution networks.

This study investigates the potential of AI-driven MOSFET bypass systems and computer vision to mitigate partial shading losses in photovoltaic (PV) systems. Partial shading, particularly due to cloud cover, can significantly reduce the power output of PV systems. As commonly used bypass diodes have limitations due to power loss and hotspots, this study explores the use of MOSFETs as an alternative, paired with a convolutional neural network (CNN) based computer vision-based cloud detection system. The CNN model is trained to detect cloud edges and track their movement, enabling dynamic adjustment of the MOSFET bypass system. The results show that the MOSFET bypass system outperforms traditional bypass diodes in energy efficiency, particularly in medium to low intensity static shading conditions. The study also highlights the potential of predictive analysis to improve the accuracy of detection and MOSFET control for dynamic shading conditions. Findings suggest that AI-driven MOSFET bypass systems and computer vision can be an effective solution for mitigating partial shading losses in PV systems, particularly in large-scale solar arrays. Future studies can focus on refining the CNN model, exploring alternative control methods, and further investigating the energy efficiency of the system under various shading conditions.

The unified power quality conditioner (UPQC) based on modular multilevel converter (MMC) integrated with photovoltaic (PV) enables renewable integration and power quality regulation. However, its startup strategy is more complex than the conventional MMC. This article presents a startup strategy that is aimed at achieving rapid and stable operation. First, the shunt-side submodules (SMs) capacitors are charged as the system operates, thereby establishing direct current (DC) side voltage. Subsequently, voltage of series side is gradually increased through the DC side. However, the two sides arm voltages are not identical because of the distinct charging techniques. Thus, to equalize the voltage on both sides, a method controlling the number of conducting SMs on the series side is employed. In addition, an integral backstepping sliding mode control (IBSMC) is introduced to further increase the voltage during the controlled rectification stage. Then, the PV side switch is closed and the power control strategy is activated to increase the output power. Finally, the method for controlling voltage on the series-side transformer is utilized to minimize voltage drop upon its connection. The MATLAB/Simulink and real-time laboratory (RT-LAB) test results demonstrate that the proposed strategy achieves startup within 14 s and reduces the voltage rise time from 0.75 to 0.1 s. Moreover, the total harmonic distortion (THD) of grid current is reduced from 5.27% to 1.85% and the voltage recovery time is decreased from 0.11 to 0.01 s, thereby validating the effectiveness of the proposed method.

The high penetration of renewable energy sources in future power grids presents stability challenges for grid-connected inverters, particularly during large frequency drops under a wide range of short-circuit ratio and reactance-to-resistance (X/R) conditions. To address these challenges, especially for applications requiring firm power delivery, such as utility-scale renewable plants under power purchase agreements, this article proposes a tight grid-forming (TGFM) control framework. The primary objective of TGFM is to ensure precise and robust tracking of active power references, shifting the focus from passive frequency support to active and reliable power injection during grid disturbances. The proposed framework employs a grid voltage state observer to achieve real-time, high-fidelity estimation of the grid voltage's phase and frequency across diverse line impedance conditions. By integrating this real-time grid information into the controller, the inverter dynamically compensates for grid phase variations. This ensures a stable power angle, enabling the inverter to tightly track its power command and thereby preventing the overcurrent and oscillatory instabilities common during large frequency drops. The effectiveness of the proposed TGFM strategy is validated through comprehensive simulations and physical experiments under various grid conditions, demonstrating superior power tracking performance and stability.

Accurate short-term wind power forecasting remains challenging due to the spatial heterogeneity among wind turbines and the complex spatio-temporal dependencies of power fluctuations. To this end, this paper proposes a short-term wind power forecasting framework targeting intra-farm turbine-level spatio-temporal modeling. First, a grid-based exhaustive distribution search (GEDS) algorithm is developed to identify spatially representative wind turbines by exploring combinations of grid sizes and offset parameters to ensure diverse and informative inputs. Then, an enhanced spatio-temporal fusion network (ESTFN) is proposed, where spatially, the ESTFN integrates lightweight alternating convolutional enhancement and spatial attention modules to extract local heterogeneity and global spatial features. In temporal, ESTFN combines gated recurrent unit and temporal attention modules to capture short-term dynamics and long-term temporal dependencies. A learnable fusion mechanism is employed in the spatial and temporal domains, respectively, to realize the fusion of multi-scale information. To improve the balance between overall forecasting accuracy and dynamic response, a combined loss function consisting of a point-value term and a trend term is designed. Experiments on real wind farm datasets show that the framework effectively enhances spatio-temporal feature extraction and improves the forecasting performance in complex environments.

Accurate wind power forecasting is crucial for grid security and efficient dispatch, but still faces challenges such as data leakage and lag in practical applications. Real-time decomposition (RTD) has been employed to address data leakage in decomposition-based wind power forecasting. However, the tail decomposed values of RTD often exhibit high noise and lag. Analysis shows that this tail deviation primarily arises from the lack of follow-up information, making the tail more susceptible to short-term fluctuations and random noise, which increases high-frequency noise and lag. To overcome this issue, this paper proposes a real-time supplement values decomposition (RTSD) method, i.e., group forecasting of subsequent values at the tail of the real-time series, supplement values and decomposition. Subsequently, the Kalman filter with optimized parameters (OKF) is applied for smoothing, balancing the requirements of single-step and multi-step forecasting. Considering that deep learning forecasting models for decomposed sub-series can adopt different training strategies, this study proposes three training strategies-independent, parallel and ensemble-and selects the optimal one through comparative analysis. Experimental results demonstrate that RTSD significantly reduces noise and lag in RTD, and the robustness of the proposed RTSD-OKF architecture and chosen training strategy is validated across various forecasting models and decomposition methods.

Accurate state of health (SOH) estimation is essential for reliable battery management systems. Existing data-driven approaches typically rely on cycle-level statistical indicators extracted from charging data. While these features are informative, they may not fully capture the geometric structure of voltage–current trajectories during battery degradation. Conversely, purely neural representations often lack structured inductive bias and may generalize poorly across datasets with different battery chemistries. To address this limitation, this paper proposes Sig-FiLMNet, a dual-branch framework that integrates path-signature representations of charging trajectories with cycle-level statistical indicators through a feature-wise linear modulation (FiLM) mechanism. The signature branch encodes charging-path geometry, while the trunk network embeds nine scalar statistics from each cycle. The FiLM module adaptively modulates signature features for degradation-aware SOH prediction. Experiments on the Huazhong University of Science and Technology (HUST), Massachusetts Institute of Technology (MIT), and Tongji University (TJU) datasets achieve mean absolute error/root mean squared error (MAE/RMSE) of 0.55%/0.65% (7.6K parameters), 0.41%/0.52% (44.8K), and 0.45%/0.53% (135.6K), respectively. In cross-dataset transfer experiments, the proposed framework achieves an average MAE of 0.63% across 24 settings. The learned FiLM modulation patterns also provide insight into how trajectory features contribute to SOH prediction across degradation stages.

Though several peer-to-peer (P2P) electrical energy market concepts are proposed in the recent literature, their physical implementation aspect has been largely overlooked. In this paper, a practically implementable decentralized transactive energy (TE)-based P2P market framework with product differentiation (PD) has been proposed and demonstrated in a physical distribution testbed. The required entities and distribution system models, PD preferences (for both buyers and sellers), multi-agent system-based problem formulation and market clearing algorithm are clearly explained. Subsequently, the pre-deployment experimental testbed setup utilized for the demonstration, with the constituting resources, is discussed. The testbed experimentation of the proposed framework is carried out for various scenarios of PD based on seller and buyer preferences while having solar/wind generation as the renewable energy producer (REP), which demonstrates the increase in revenue for the solar (6.7%) and wind (3.6%) energy producers in buyer preference and revenue agnostic seller preference. The segregated presentation of SCADA-GUI results and metered results from the testbed respectively shows (i) successful implementation of the PD-enabled TE-based P2P market and (ii) successful dispatch/control of commercial testbed entities through communication while adhering to system constraints.

Grid-interfaced fuel cells and solar photovoltaic (PV) systems represent a synergistic approach to sustainable energy production, leveraging the complementary strengths of these technologies. Fuel cells provide consistent and reliable power by converting hydrogen into electricity, while solar PV systems harness intermittent solar energy. Integrating these systems with the grid enhances overall efficiency, reliability, and resilience of the power supply. This paper investigates the design, implementation, and performance evaluation of hybrid fuel cells, solar energy conversion systems, and battery storage systems interfaced with the electrical grid. It examines the control strategies for optimal power management under unstable grid and load conditions. Employing a fifth-order generalized integrator (FOGI), this approach adeptly addresses grid disturbances and load challenges, ensuring swift transient responses and enhanced steady-state operation. The research findings highlight the potential of grid-interfaced hybrid systems to contribute significantly to clean energy transitions, ensuring a stable and sustainable electricity supply.

This work presents an integration of the holomorphic embedding power flow method (HEPFM) with the functional expanded extreme learning machine (FEELM) to develop an online corrective contingency analyzer for evaluating 'whatif' scenarios in a floating green power solution. The HEPFM offers a rapid and accurate power flow solution, outperforming traditional search-based methods in efficiency and precision. Performance indices based on active power and voltage are computed to assess the impact of contingency events, enabling the recommendation of corrective actions to enhance system resilience under emergency conditions. Three operational modes-black start, baseline operation, and generation outage are simulated and analyzed using MATLAB/Simulink software. The FEELM method is proposed as a computationally efficient approach for generating corrective actions by utilizing generation power, energy storage power, and load power as inputs, applicable in both grid-connected and islanded modes. The intelligent HEPFM-FEELM framework demonstrates simplicity, practicality, and robustness, providing a comprehensive and efficient solution for managing contingency events in modern power systems.

Though battery-less solar-plant integrated ultra-fast charging station (EV-UFCS) solutions are theoretically preferred, there is no existing control method that simultaneously ensures solar-plant's MPPT and maximum charging energy delivery to plug-in EVs (PEVs), while adhering to instantaneous grid-side power-ramp-rate and each PEV's charging current limits. A novel coordinated-control technique is proposed in this paper to meet these critical objectives, while being constrained by the instantaneous grid-side and PEV-side limits. The theoretical modelling and an implementable algorithm for the proposed control technique are elaborately explained. Experimental validation of the proposed technique is executed on a laboratory-scale 18 kVA solid-state-transformer (SST)-based solar-aided universal EV-UFCS testbed. The experimental results clearly demonstrate that (i) solar MPPT is achieved with >99% accuracy, (ii) similar to 100% of maximum charging energies are delivered to all categories of connected PEVs, and (iii) instantaneous grid power-ramp-rate and PEV-BMS constraints are strictly adhered to, which highlight the proposed coordinated-control technique's advantages.

In recent years, commercial buildings in Singapore have experienced a significant surge in energy consumption, primarily driven by the increasing demand for cooling systems due to global warming and population growth. Chillers, being major energy consumers, play a crucial role in maintaining optimal indoor temperatures. To address this challenge, various chiller models, including DOE-2 and modified Gordon-Ng models, have been explored to simplify complex chiller systems. However, selecting the most suitable model for different applications remains challenging due to discrepancies in performance indicators. This study presents a comprehensive comparison of these models, focusing on their predictive performance capabilities. The analysis serves as a foundational step for subsequent chiller optimization strategies aimed at reducing overall energy consumption in chiller plants. The results indicate that the modified Gordon-Ng model exhibits superior predictive accuracy compared to other models. This research is particularly pertinent in the context of Singapore’s sustainability goals, including greening 80% of its buildings by gross floor area by 2030 and achieving net-zero emissions by 2050. Optimizing chiller performance can significantly contribute to these objectives, making this study a valuable contribution to the field of energy efficiency in commercial buildings.

The reliability of power converters is always a key concern when designing new systems. In this paper, a new Fault-Tolerant Active Neutral Point Clamped (FT-ANPC) converter has been introduced. Under normal conditions, this converter produces a seven-level output voltage using six unidirectional switches and three bidirectional ones. To ensure its robustness, the converter performs when open-circuit faults (OCFs) occur, which has been analyzed, focusing on the impact of each switch. The converter is controlled using a carrier-based modulation technique. The proposed FT-ANPC converter has been tested in normal operation and under fault conditions to validate its performance. Power quality in different scenarios has also been evaluated using MATLAB.

Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.

The modular multilevel converter based active power filter (MMC-APF) integrated with the photovoltaic (PV) not only compensates for grid harmonics and reactive power but also optimally leverages the energy output of the PV source. However, due to the inherent volatility of PV systems, the MMC encounters power mismatches. In addressing the power mismatch issue, the parameter selection process posed a critical challenge, requiring careful tuning to ensure optimal system performance. To address the issue of complex parameter tuning in the power control process of MMC-UPQC with PV, this paper proposes a phasor particle swarm optimization (PPSO) based proportional integral resonant (PIR) power control strategy for MMC-APF with PV, to better eliminate power mismatches and restore current quality. First, the causes of power mismatches in PV submodules (PSMs) are analyzed. And a strategy for eliminating power mismatches is designed by injecting current. Then, the integration of PIR control enables the proposed strategy to compensate for current harmonics and improve current quality. Finally, the PPSO algorithm is applied to optimize the PIR controller parameters, ensuring optimal performance. The proposed PPSO-optimized power control strategy is validated on the MATLAB/Simulink platform, demonstrating its effectiveness and superiority.

Traditional islanding detection and low-voltage ride-through (LVRT) mechanisms are often implemented separately in inverter control. This separation can lead to conflicts during grid disturbances, especially in systems with high-penetration inverter-based resources (IBRs). Grid-forming inverters (GFMIs), particularly synchronverter-based designs, must satisfy both anti-islanding and LVRT requirements under evolving grid codes. This paper proposes a seamless integration strategy that combines passive islanding detection with LVRT functionality for synchronverter-based GFMIs. A coordinated control framework is developed, unifying a novel passive islanding detection function (IDF) with a voltage-dependent LVRT response. The IDF uses virtual flux, inverter current, and virtual speed deviation as key indicators. A time-domain logic discerns between unintentional islanding and grid faults within 30 ms. This ensures the timely activation of either disconnection or LVRT protocols. Simulation results on an IEEE 1547-based test system validate the approach. The method maintains voltage support during fault-induced sags and detects islanding reliably within 80 ms. It significantly reduces the non-detection zone (NDZ) without compromising system stability. The proposed strategy provides an efficient and practical solution for next-generation synchronverter-controlled GFM Inverters.

There is near global acknowledgement that a shift to green energy is imminent and urgent. This paper explores the feasibility and impact of producing green hydrogen as a decentralised energy carrier through electrolysis. Countries like Singapore are exploring options of using hydrogen for electric vehicles and energy production applications. Two notable electrolysers dominate the market, namely, the Alkaline electrolyser (AEL) and the Proton Exchange Membrane water electrolyser (PEM-WE). Although both technologies have been integral to industrial hydrogen production for decades, debates persist regarding their relative advantages and shortcomings. This research study evaluates the characteristics of both types of electrolyser, focusing on operational efficiency and economic viability. To assess efficiency, the effect of varying current density, temperature, and pressure, was studied. As for cost, a Life Cycle Cost Analysis was conducted using MS Excel. Additionally, we developed a Techno-Economic Recommendation Calculator in a Graphical User Interface on MATLAB, which integrates both the efficiency and cost evaluation outcomes presented in this study. The findings indicate that PEM-WEs demonstrate superior cost-effectiveness and energy efficiency in the long term compared to their Alkaline counterparts. This study aims to provide valuable insights for stakeholders making informed decisions in the transition to a hydrogen-based economy.

This article explores a 7-level reduced switch multilevel inverter (RSMLI) topology for electric vehicle (EV) powertrain application. This work is aimed at enhancing efficiency and compactness, while minimizing the cost and complexity. Traditional multilevel inverters need a large number of power electronic switches, increasing the overall system size, switching losses, and control complexity. The proposed RSMLI topology significantly reduces the number of active switches and gate drivers without compromising output voltage quality or power delivery capability. This results in lower total harmonic distortion (THD), reduced conduction and switching losses, and improved overall system reliability. The inverter operates effectively across a wide range of operating conditions typically encountered in EVs, offering robust performance and fast dynamic response. Furthermore, the reduced switch count contributes to better thermal management, increased power density, and simpler control implementation. Simulation results confirm the proposed inverter’s performance during different dynamic conditions, demonstrating superior efficiency and fast transient response. The compact nature of the design makes it highly suitable for integration in space-constrained EV applications. Comparative analysis with some of the recent multilevel inverters highlights the advantages on the basis of switch count, DC sources, power diodes, and capacitors used. The proposed RSMLI offers a promising solution for next-generation EV powertrains, promoting energy-efficient and sustainable transportation.