ABSTRACT Aiming at the problems of low prediction accuracy, large prediction deviation, and scarce samples that make modelling difficult under extreme weather conditions such as cold waves, this paper proposes a wind power prediction model based on time series generative adversarial network (TimeGAN)—empirical mode decomposition (EMD)—stacking ensemble. First, the TimeGAN algorithm is used to generate samples of meteorological and power data to build a database of extreme weather meteorological and power data. Next, based on EMD, the expanded wind power data are decomposed into high and low‐frequency sequences. For each subsequence obtained from decomposition and reconstruction and the corresponding meteorological factors, an improved particle swarm optimisation (IPSO) algorithm is utilised to optimise the parameters of several commonly used prediction methods and to compare their prediction accuracy. Then, combining the performance of the algorithms and the principle of maximum difference, the optimal algorithm combination is determined by using prediction error evaluation indicators and Pearson correlation coefficients to construct a stacking ensemble learning‐based multi‐algorithm combination model for extreme weather power prediction. Finally, the effectiveness of the proposed method in improving prediction performance is verified through a real cold wave case, demonstrating its good practicality.

ABSTRACT Green hydrogen (H 2 ), produced from standalone hybrid renewable energy systems, is increasingly recognized as a sustainable substitute for diesel in remote regions. Despite substantial progress, most existing studies are limited to site‐specific analyses and lack systematic regional comparison, making it difficult to generalize techno‐economic and environmental outcomes. To address this gap, this work develops a reliability‐constrained techno‐economic and environmental evaluation framework for off‐grid renewable hydrogen production and applies it to twelve representative regions across Australia using hourly time‐series simulations over a full annual horizon. The proposed framework integrates solar photovoltaic generation, wind energy, lithium‐ion battery storage, proton‐exchange‐membrane electrolysers, hydrogen storage tanks and power conversion units within a unified modelling structure. System performance is assessed using standardized indicators, including levelized cost of hydrogen (LCOH), net present cost, internal rate of return (IRR) and lifecycle emissions. The results reveal strong regional variability driven primarily by renewable resource quality and hydrogen demand scale. Among the assessed regions, Pilbara achieves the lowest LCOH of $2.92/kg and the highest IRR of 7.08%. From an environmental perspective, renewable hydrogen systems reduce lifecycle emissions by up to 95% compared with conventional grey hydrogen pathways. Sensitivity analysis further identifies electrolyser capital cost and solar capacity factor as the dominant drivers of economic competitiveness. Overall, the proposed framework enables transparent cross‐regional comparison of off‐grid renewable hydrogen systems and provides decision‐relevant insights for policymakers and investors to prioritize cost‐effective, low‐carbon hydrogen deployment in remote applications.

ABSTRACT Snow accumulation on photovoltaic (PV) systems can cause significant energy losses, especially in high‐latitude regions. To estimate these losses with high accuracy, snow loss models have been introduced and deployed on the energy production of PV modules. This study presents a comprehensive comparison of several snow loss models, including the widely used Marion model and its recent modifications, and the Townsend model, applied to four years of PV data from a high‐latitude PV site with a harsh snowy climate. By using ERA5 reanalysis data to model baseline PV power, this research demonstrates that snow loss estimation can be achieved without site‐specific irradiance or temperature measurements. This study provides the first standardised comparison of these models under identical conditions in a Nordic climate, offering practical guidance for snow loss modelling. The models are evaluated based on annual and monthly snow losses and detailed error metrics, with the focus on winter periods. The findings highlight the potential of ERA5‐based PV simulations combined with snow models for reliable forecasting in snowy regions around the world. The results indicate that models incorporating hourly snow coverage checks and snow‐depth‐dependent clearing coefficients significantly outperform the original Marion model by reducing errors by 32%.

ABSTRACT The voltage‐source‐converter (VSC)‐based high‐voltage direct current (HVDC) transmission technology has been increasingly adopted for power delivery from offshore photovoltaic (PV) power stations. Nevertheless, the offshore PV power stations can readily interact with the VSC‐HVDC system, leading to sub‐/super‐synchronous oscillations. In order to investigate the underlying mechanism, an accurate impedance model considering frequency‐coupling effects is developed in this paper. Then, a reduced‐order impedance for quantitative stability assessments is further derived. It has been revealed through the analysis that the PV power stations exhibit negative‐damping characteristics in the sub‐/super‐synchronous frequency range, which may trigger oscillations. Moreover, increasing the output power of individual PV units or the number of PV units connected to the HVDC system reduces the magnitude of the aggregated PV impedance, thereby increasing the oscillation risk. Time‐domain simulations reproduce pronounced sub‐/super‐synchronous oscillations, with dominant spectral components spreading around 20 Hz and 80 Hz. Furthermore, hardware‐in‐the‐loop (HIL) experimental tests validate that the oscillatory components are effectively suppressed by properly reducing the phase‐locked loop (PLL) parameters. When the PLL proportional gain is reduced to 0.8 of its setting value, the total harmonic distortion (THD) of the voltage at the common coupling is reduced from 11.8% to 0.72%. This indicates oscillation mitigation.

ABSTRACT Ensuring the cybersecurity of secondary control in AC microgrids is crucial for maintaining system stability and operational resilience. Cyberattacks targeting the communication links of secondary controllers can lead to severe disruptions in frequency regulation, voltage control and power sharing mechanisms. This paper proposes a cybersecurity‐resilient control strategy based on the short‐time correlation transform (STCT) to enhance the robustness of distributed secondary control against cyber threats. The proposed strategy leverages the magnification properties of STCT to accurately identify compromised communication links, isolate them from the control network and reinforce the integrity of unaffected links for reliable data transmission. The STCT‐based strategy is independent of attack type and location, making it highly adaptable to dynamic microgrid environments. The effectiveness of the proposed strategy is validated through extensive simulations on multiple microgrid topologies under various attack scenarios. Results demonstrate that the approach successfully mitigates cyberattacks, preserves system stability and restores microgrid frequency to its reference value with minimal delay. Furthermore, due to its lightweight computational requirements and compatibility with practical control implementations, the proposed strategy is well‐suited for real‐time applications. This study provides a significant advancement in microgrid cybersecurity by introducing a highly efficient, scalable and attack‐agnostic protection mechanism for distributed secondary control.

ABSTRACT In multi‐terminal flexible direct current (DC) systems, faults on the DC line develop rapidly, posing significant challenges for traditional protection schemes in terms of both speed and reliability. To address this issue, a novel transverse protection scheme based on the transient equivalent impedance difference is proposed. This scheme is particularly relevant to systems integrating large‐scale renewable energy sources, such as wind farms, where fault characteristics can differ significantly from traditional power systems. First, considering the differences between sending‐end and receiving‐end converter stations, equivalent models of the DC sides of the converter stations are constructed. By analysing the topology of the equivalent model, analytical expressions for the voltages at the converter stations on both ends of the protected line, under different fault scenarios, are derived. Based on the differences in the voltage–current ratios during internal and external faults, the protection criterion is established using the characteristic frequency point at which the difference coefficient reaches its maximum. Finally, simulation results on the RT‐LAB platform demonstrate that the proposed protection scheme has robustness against changes in fault resistance. The total operational response time is less than 3 ms, which can meet the stringent speed requirements for DC line protection, offering a reliable solution for systems with high levels of renewable energy integration.

ABSTRACT To achieve the low‐carbon transformation of energy system, a low‐carbon optimal scheduling strategy for integrated energy systems (IESs) based on improved carbon emission flow and green certificate‐carbon joint trading is proposed. Firstly, based on the multi‐energy coupling characteristics of IES, the impacts of renewable energy generations (REGs), multi‐energy coupling components, and energy storage devices on the carbon flow distribution are quantified, and an improved carbon flow calculation method for IES is proposed to solve the tracing problem of the temporal and spatial transfer of carbon emissions caused by the power loss of coupling components and the operation of energy storage. Then dynamic carbon emission factor reflecting the temporal and spatial distribution characteristics is proposed in combination with the concept of carbon potential. Moreover, to deeply explore the potential of source‐load synergy for carbon reduction, a green certificate‐carbon joint trading mechanism is constructed on the power generation side by combining the green certificate carbon reduction mechanism and carbon trading, and on the load side, the dynamic carbon emission factor is used as the carbon price correction coefficient for energy consumption prices, driving users to respond synergistically to carbon‐energy. Finally, by combining the scheduling of the source‐side units with the demand response on the load side, a bi‐level optimal scheduling model considering the deep reduction of carbon by source‐load synergy is constructed. The simulation analysis of the case study shows that the proposed method achieves carbon measurement, carbon tracing, and reasonable allocation of carbon emission responsibilities in the IES, effectively improving the system's carbon reduction capacity and economic benefits.

ABSTRACT This study examines the feasibility of transitioning to a fully renewable energy system in Iran by 2050, using the LUT Energy System Transition Model with hourly resolution and nine geographical regions. Three policy‐driven scenarios are assessed, named the best policy scenario (BPS), delayed policy scenario (DPS), and current policy scenario (CPS). The model integrates all major energy‐consuming sectors, including power, heat, transport, industry, and desalination, along with sector coupling and power‐to‐X technologies. The results indicate that a 100% renewable energy system is technically feasible and economically viable in the BPS, driven by large‐scale solar deployment. CO 2 emissions decrease from 446 MtCO 2 in 2020 to zero by 2050. Primary energy demand reaches around 3450 TWh in the BPS, with over 93% supplied by solar photovoltaics and installed capacity exceeding 1660 GW. Despite higher initial investments, the BPS achieves the lowest levelised cost of electricity at €24.7/MWh by 2050, compared to €49.4/MWh in the CPS. System flexibility is ensured through electricity and thermal storage plus hydrogen‐based seasonal balancing. Solar‐to‐X strategies play a key role in defossilising industry and transport sectors through e‐hydrogen, e‐fuels, and e‐materials. Achieving this transition requires strong policy support, including fossil fuel subsidy removal, CO 2 pricing, and infrastructure investment.

ABSTRACT Virtual power plants (VPP) can enhance power grid stability by aggregating distributed energy resources (DERs) to provide ancillary frequency regulation services. During this process, VPP tracks power regulation command (PRC) and optimally achieves generation command dispatch (GCD) among internal DERs. To overcome the limitations of conventional GCD methods in robustness, accuracy, and computational efficiency, this paper presents an improved decentralised GCD method considering acceleration and communication noise. This method integrates line capacity constraints using a linearised power flow model to avoid overloading the lines. Based on the model, an improved exact diffusion algorithm (EDA) is introduced with the acceleration based on the Nesterov accelerated gradient (NAG) and communication noise suppression using the gain function. Simulations on the IEEE 33‐bus system and a real‐world VPP in a province of China demonstrate that the proposed algorithm achieves at least 3 times faster convergence than other consistency algorithms. Compared with six mainstream methods, the proposed method improves the frequency regulation performance of VPP by 1.00% to 64.69%, while reducing operational costs by 0.46% to 5.20%. Meanwhile, under different noise situations, the noise suppression capacity is verified, and the optimal parameter selections are presented.

ABSTRACT Effective operation and maintenance (O&M) is critical to reducing the levelised cost of energy (LCOE) from wind power, yet the unstructured, free‐text nature of turbine maintenance logs presents a significant barrier to automated analysis. We address this by presenting a novel, reproducible framework for benchmarking large language models (LLMs) on classifying these complex industrial records, which is publicly available as an open‐source tool. We demonstrate the framework's utility by systematically evaluating a diverse suite of state‐of‐the‐art proprietary and open‐source LLMs, providing a practical case study for the foundational assessment of their trade‐offs in reliability, operational efficiency and model calibration. Benchmarked against a qualitatively selected high‐performing reasoning model, our study demonstrates how the tool quantifies a clear performance hierarchy to help analysts identify top models exhibiting high alignment and trustworthy confidence scores. Furthermore, we show that classification performance depends heavily on semantic ambiguity, with all models displaying higher consensus on objective component identification than on interpretive maintenance actions. Given that no model achieves perfect accuracy and calibration varies dramatically, we conclude the most responsible application is a human‐in‐the‐loop system where LLMs act as assistants to accelerate and standardise data labelling, enhancing O&M data quality and downstream reliability analysis.