ABSTRACT Cryptocurrencies have emerged as a transformative force within the global financial ecosystem, propelled by the innovation of blockchain technology. This decentralized ledger system has garnered widespread attention for its potential to reshape traditional financial paradigms. However, the substantial electric energy consumption by cryptocurrency miners poses a significant environmental challenge, necessitating urgent intervention to mitigate its ecological impact. Renewable energies are great choices to deal with this challenge, presenting opportunities for promoting eco‐friendly practices within the blockchain industry. This article reviews the literature on different aspects including the electric energy demand of cryptocurrency, the application of renewable energies such as hydro, solar, wind, geothermal, and bio energies in cryptocurrency mining, and cryptocurrency policies, seeking to present the green solution for cryptocurrency mining. It aims to empower cryptocurrency miners, energy providers, and researchers in advocating for environmentally friendly practices within the blockchain industry. Unlike previous reviews that often focus on a single or two renewable sources or only on Bitcoin, this study provides a comprehensive comparison of five renewable technologies across multiple cryptocurrencies and ranks them based on LCOE, maturity, and operational suitability for continuous mining. With the increasing accessibility and affordability of renewable energy sources globally, there is optimism for transitioning towards cleaner power options in cryptocurrency mining. Through collaborative efforts and informed decision‐making, a future is envisioned where energy‐efficient solutions take precedence in sustainable cryptocurrency mining practices. The findings show that wind and solar energies are the best selections for cryptocurrency mining, respectively.
ABSTRACT This study examines the ambient‐adjusted rating (AAR) of flexible busbar conductors in power substations, a simple form of dynamic thermal rating that is directly applicable with existing meteorological infrastructure. One year of weather data measured at a Slovak transmission substation was used to rate the ACSR 758‐AL1/43‐ST1A conductor with two AAR models: M1, with variable ambient temperature and constant solar irradiation, and M2, with both quantities variable. The resulting dynamic ratings are compared with the static ratings of four Central European countries. The median rating of model M2 is 39% above the Slovak static value at the 80∘C conductor temperature limit, with an overload potential index of 1.91. The revision extends the original analysis in three directions requested by practice: a wind‐speed sensitivity study (0.2–2.0 m/s), a full dynamic rating model M3 that uses the measured wind speed, and a statistical assessment covering seasonal variability, rating distributions, thermal risk at static‐rating loading, and measurement uncertainty. Model M3 reaches a median of 1851 A, 24% above M2, which quantifies the capacity given up by the conservative wind assumption of the AAR. A thermal‐risk analysis shows that continuous loading at the Slovak static rating never brought the conductor above 77∘C during the analysed year. The results support AAR as a low‐cost first step for both transmission and distribution system operators, with straightforward validation through conductor temperature monitoring in the confined substation environment.
ABSTRACT Cables are widely used in renewable‐energy stations, and the reliable assessment of their insulation condition is essential for safe and stable system operation. In this study, the harmonic characteristics of the loss current in 10 kV XLPE cables containing conductor spike defects were investigated through simulation and experimental analysis. Based on the bipolar charge transport model, the nonlinear conductivity behavior of XLPE under high electric fields was analyzed, and the effects of tip curvature radius, defect location, and electrode spacing on the loss current and conductivity characteristics were evaluated. The results show that conductor spike defects cause severe local electric‐field distortion, which induces periodic fluctuations in insulation conductivity and distorts the loss‐current waveform, leading to the appearance of higher‐order harmonics. The harmonic components are dominated by the 3rd and 5th harmonics, and their contribution ratios vary with defect severity and location. As the defect becomes more severe, the amplitude and total harmonic distortion of the loss current, as well as the conductivity distortion rate, all increase significantly. Moreover, defects located at the conductor core produce higher distortion levels than those located at the shielding layer. A representative experimental condition supports the principal 3rd‐ and 5th‐harmonic signatures predicted by the simulation, providing a physical basis for further investigation of loss‐current harmonics as comparative indicators of conductor spike defects in 10 kV XLPE cables.
ABSTRACT Indoor localization using Wi‐Fi fingerprinting has gained significant attention due to its cost‐effectiveness and the widespread availability of Wi‐Fi infrastructure. However, the high dimensionality of data, along with signal fluctuations and redundant access points (APs), often degrades localization accuracy and increases computational complexity. This paper proposes an improved binary particle swarm optimization‐based feature selection strategy to identify the most informative APs for fingerprint‐based indoor localization. The approach integrates a novel hybrid transfer function combining V‐shaped and U‐shaped characteristics with an enhanced learning mechanism to balance exploration and exploitation while preserving population diversity. An improved weighted k‐nearest neighbours algorithm is employed to evaluate the selected AP subsets. Experiments conducted on the UJIIndoorLoc benchmark dataset demonstrate that the proposed method achieves an overall weighted mean localization error of 6.27 m while selecting approximately 85% fewer APs than the original feature set. The proposed AP selection strategy reduces computational overhead during localization, making the method suitable for deployment on resource‐constrained devices without compromising localization performance.
ABSTRACT Power sharing for inverter‐based resources (IBR) coordinates the flow of real power in response to changing demand. In some angle droop–based power sharing algorithms, global positioning system (GPS) communication is used to provide an accurate timing reference to the IBR power sharing controller. However, this GPS communication exposes the IBR power sharing controller to GPS spoofing attacks. This paper first proposes a decision tree detection method for GPS spoofing attacks on power sharing controller. This detection method detects a GPS spoofing attack. Moreover, a long short‐term memory (LSTM) is proposed to mitigate the impact of the GPS spoofing attacks by estimating the phase reference of the IBR controller, and taking corrective action with integral control. The LSTM mitigation outperforms a state observer–based mitigation approach proposed in the literature. The performance of the proposed algorithms is tested under various cases in a cosimulation environment with a communication layer built in Python and the modified IEEE 39‐bus benchmark power system in PSCAD/EMTDC.
ABSTRACT The majority of Ethiopia's public buses are customized from imported cargo trucks to accommodate people with various anthropometric traits. This mismatch leads to increased musculoskeletal disorders (MSDs), discomfort for drivers and an increased risk of traffic accidents. This study defines ideal seat proportions based on Ethiopian anthropometry in order to close this crucial ergonomic safety gap. Key percentiles were examined after 157 Ethiopian drivers (119 men and 38 women) had their 11 anthropometric characteristics gathered. Seat height (44.19 cm), seat depth (44.04 cm), seat width (46.69 cm), backrest height (55.12 cm) and shoulder width (46.94 cm) are the suggested ideal seat specifications. Federal transport safety experts constructed, built and assessed a prototype seat in a medium bus. The suggested model's greater comfort and safety are demonstrated by the results, which also confirm notable departures from current seat designs. In order to lower MSDs and accident rates, the current study directly informs national transportation safety regulations by offering practical ergonomic guidelines for regional vehicle modification industries. The information also creates a fundamental anthropometric database for Ethiopian ergonomic design in the future.
ABSTRACT Music mood classification supports recommendation, retrieval, and wellness applications, yet reported methods and evaluations vary widely across studies. Following the PRISMA systematic review methodology, this paper synthesises 39 peer‐reviewed works between 2009 and 2025 to summarise common practice across feature design, model families, datasets and reporting. Analysis reveals a fundamental structural split in the field: 17 papers focus on music‐centric classification (using audio, lyrics, and metadata), while 20 papers focus on user‐centric emotion recognition (using vision or physiological signals) to drive recommendation pipelines. Convolutional neural networks (CNNs) dominate the technical landscape (31 of 39), serving as a shared architecture for both audio spectrograms and facial analysis. However, this split creates a ‘mapping gap’ due to taxonomical inconsistencies between listener‐state datasets (e.g., FER2013) and music‐content datasets (e.g., DEAM). Reporting remains a significant bottleneck: 32 papers report only a single overall accuracy, masking uneven performance across mood classes, while only two papers provide granular metrics like F1‐score or recall. This study concludes that while technical pipelines are mature, the field requires standardised emotional mapping layers and richer evaluation metrics to improve the practical robustness and real‐world readiness of music mood classification systems.
ABSTRACT In this paper, a new soft‐switching structure for a buck‐boost converter based on pulse width modulation (PWM) is proposed. The proposed structure introduces a simplified soft‐switching approach for buck‐boost converters that achieve zero voltage switching (ZVS) for main switches using only a single auxiliary switch composed of minimal components. Unlike conventional methods that require complex snubber networks or multiple auxiliary elements, the proposed design ensures ZVS‐ZCS operation for all semiconductor elements, significantly reducing switching and reverse recovery losses. Furthermore, its ground‐referenced input and output facilitate practical implementation and system integration. The theoretical analysis is done for the proposed converter during this paper, and its voltage gain is calculated. Also, its duty cycle and switching conditions under zero voltage are determined. Then, design considerations are presented for the proposed converter. Finally, the extensive hardware‐in‐the‐loop (HiL) simulations are carried out for the studied converter to validate the usefulness of the proposed approach under various load conditions. Moreover, experimental results obtained from a laboratory prototype are presented to further validate the theoretical and simulation findings. The prototype achieves an output voltage of 16.5 V and peak efficiency of 97.4% at full load.
ABSTRACT Multiphase field‐wound flux switching (FWFS) machines combine the advantages of traditional flux‐switching permanent magnet machines with the benefits of multiphase architectures, including enhanced flux controllability, high fault tolerance, increased torque density and reduced torque ripple. The doubly salient structure contributes to mechanical robustness and structural simplicity, while the utilisation of non‐overlapping windings reduces the volume of copper, which also minimises the production costs of FWFS machines. This paper presents a novel five‐phase outer rotor FWFS machine with non‐overlap windings and investigates four different rotor pole topologies. All the designs are analysed with respect to flux linkage, back‐EMF, cogging torque and electromagnetic torque. Among these configurations, the 10 slot/11 pole configuration demonstrated superior performances, exhibiting high flux linkage and average torque and minimum cogging torque. A deterministic optimisation technique was applied on 10 slot/11 pole topology to further enhance its electromagnetic performance, resulting in a 44.8% increase in average torque along with significant improvements in other parameters. The optimised design was then compared against the existing inner rotor design, showing a 96% increase in torque output. Additionally, the rotor pole designs were optimised to suit direct‐drive applications, confirming the proposed five‐phase design's potential for high speed and high reliability use in electric vehicles and renewable energy systems.
ABSTRACT This concept study investigated a stepwise design optimisation strategy for achieving enhanced aerodynamic characteristics of the UAVs/aircraft airframes. This was achieved by implementing the aerodynamic design optimisation of wings and the integration of wingtip flow control devices (winglets). The strategy was implemented on two case studies, that is, a small‐scale low‐altitude UAV and a lightweight medium‐altitude aircraft. In the first step, the aerodynamic design optimisation of the existing UAV/aircraft was performed for the aerofoils (2D scale) using an evolutionary algorithm based on the flight conditions. In the next step, the (3D) wings of the UAV/aircraft were developed from the aerodynamically optimised 2D aerofoils. While up to 50% maximum lift coefficient enhancement was achieved during the aerofoil optimisation process, the CFD results indicated that the performance loss of up to 10% can be predicted transitioning from a 2D aerofoil to 3D (optimised) wings. A large proportion of the drag comes from the wingtips and body of the aircraft. Therefore, in future designs, it is critical to introduce flow treatment strategies contributing to reduced drag. For this, further aerodynamic quality enhancement (in terms of drag reduction) was attempted by introducing wingtip flow control devices (winglets). Winglets as an additional component, were designed, optimised (after a series of design optimisation processes), and integrated into the wings of the respective UAV (Case Study I) and light aircraft (Case Study II). In the final step, the original wings of the UAV and light aircraft airframes were replaced with the optimised wings generated by the stepwise performance enhancement strategy. A full‐scale UAV/aircraft performance comparison was conducted using numerical simulations, and aerodynamic parameters of the optimised designs were compared with the original baseline designs. As compared to the original baseline designs, an improvement in overall aerodynamic performance of 15% and 10% was demonstrated by the optimised UAV and light aircraft equipped with winglets, respectively. The optimisation strategy developed in this concept study can be useful in meeting the future demand for simple and cheaper dual‐use UAVs, not only for the existing design but also for newer versions. Further (experimental) investigations are planned by employing the scaled‐down wind tunnel testing.
ABSTRACT Smart cities represent transformative urban ecosystems, leveraging advanced technologies to address challenges posed by rapid urbanisation, energy efficiency demands, and sustainability goals. Despite extensive research, a comprehensive and integrated literature analysis combining AI‐based communication, energy management, cybersecurity, and IoT for smart city applications remains limited, which motivates this study. So, this study provides a comprehensive appraisal of critical enablers, including Artificial Intelligence (AI), the Internet of Things (IoT), Machine Learning (ML), and Information and Communication Technologies (ICT), in shaping the next generation of smart cities. Special emphasis is placed on the role of 5G and emerging 6G communication paradigms in enabling real‐time data exchange and seamless IoT integration. Furthermore, the paper explores the significance of advanced energy management systems in optimising resource allocation and mitigating environmental impacts. Cybersecurity, a cornerstone of resilient smart city frameworks, is examined in‐depth, highlighting strategies to safeguard interconnected infrastructures against potential threats. Through an extensive review of state‐of‐the‐art methodologies, this paper identifies existing research gaps and presents actionable recommendations to overcome challenges in the smart city domain. The insights provided aim to propel future innovations, fostering sustainable, secure, and intelligent urban environments.
ABSTRACT The rising global burden of pulmonary tuberculosis (PTB) and drug‐resistant tuberculosis (TB) necessitates rapid, accurate, and accessible diagnostic tools. This systematic review and meta‐analysis evaluates artificial intelligence (AI) models for TB detection, discrimination from other diseases, and drug‐resistance identification. Following the Preferred Reporting Items for Systematic reviews and Meta‐Analyses (PRISMA) guidelines, we systematically searched four databases (PubMed, Scopus, Web of Science, and Embase) from 2016 to 2025. Following data extraction, study quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS‐2) tool. A bivariate random‐effects model was used to estimate pooled sensitivity and specificity. Of the 2740 articles screened, 30 met the inclusion criteria of the quantitative analysis; 16, 6, and 8 articles were conducted on TB detection, discrimination, and drug‐resistance identification, respectively. The pooled sensitivities and specificities were 92% (95% CI 90%–94%) and 79% (95% CI 74%–84%) for TB detection, 87% (95% CI 79%–92%) and 76% (95% CI 66%–86%) for discrimination, and 89% (95% CI 85%–94%) and 94% (95% CI 92%–97%) for drug‐resistance identification. Key limitations include the retrospective design of the majority of studies, dataset heterogeneity, and limited external validation. These findings suggest that AI‐based applications hold significant potential as accurate tools for TB detection, discrimination, and drug‐resistance identification, while simultaneously mitigating reliance on traditional, time‐consuming methods. However, the urgent adoption of standardised reporting guidelines – specifically Standards for the Reporting of Diagnostic Accuracy Studies‐AI for diagnostic accuracy and Consolidated Standards of Reporting Trials‐AI for clinical trials – is essential. Adhering to these frameworks in multicentre studies will be crucial for bridging the gap between cutting‐edge research and the safe, transparent integration of AI into routine clinical practice.
ABSTRACT By analysing the response of low‐power radio waves reflected from human subjects, biomedical radar sensors enable remote monitoring without wearable devices and support emerging healthcare and human–machine interface applications. This paper reviews applications at the human–microwave frontier, including physiological sensing, non‐contact human–computer interfaces, driving behaviour recognition, human tracking, and early‐stage clinical studies. Despite rapid progress in radar‐based biomedical sensing, its integration into everyday life remains limited by body orientation, random motion, environmental clutter, and spectrum‐sharing constraints, all of which affect reliable signal acquisition. To address these challenges from a review perspective, this paper uses the bio‐inspired compound‐eye radio frequency (RF) vision concept as an organising framework for discussing spatial diversity, wavelength diversity, multi‐view beamforming, and data fusion as pathways towards more robust biomedical radar sensing. Recent advances in semiconductor technology have enabled compact radio‐frequency and millimetre‐wave integrated circuits with antenna‐in‐package solutions, making distributed and multi‐aperture radar sensing increasingly practical. Within this framework, compound‐eye RF vision can provide high‐fidelity depth and angular information, allowing radar systems to target specific body regions and extract physiological signals more reliably. Finally, this paper discusses indoor passive sensing based on ambient wireless signals and highlights the roles of advanced beamforming, multistatic detection, and spectrum‐efficient sensing in future human‐centred radar systems.
ABSTRACT Integrating device‐to‐device (D2D) communications into LTE‐advanced (LTE‐A) networks offers a promising solution to reduce network congestion and improve spectral efficiency. D2D communications allow user equipment (UE) to directly exchange content without involving the base station. This process helps to reduce network traffic and decrease latency. However, interference management and resource allocation remain two significant challenges. This study focuses on the optimal pairing, resource allocation and power assignment in uplink D2D‐enabled LTE‐A networks. We first formally formulate the sum‐rate maximisation problem in a heterogeneous network that includes both cellular and D2D users. We consider multiple D2D pairs in a single cell in our joint optimisation problem and evaluate the performance of the proposed binary power control (BPC)‐based exhaustive search and suboptimal pairing algorithms. We also assess the number of D2D pairs based on network performance. We prove the convergence of the proposed solutions and discuss their computational complexity. We also consider the max–min metric, reformulate the optimisation problem and propose two BPC‐based and heuristic solutions. We evaluate the performance of the proposed methods through simulation experiments. The numerical results indicate that our proposed suboptimal heuristic provides a near approximation to the exhaustive search solution. The results also show that although max–min solutions achieve lower total throughput in the network, they reach acceptable fairness and improve the system's energy efficiency.
ABSTRACT Power‐transformer protection demands rapid and reliable discrimination between internal faults and transients under non‐ideal operating conditions such as magnetizing inrush. This study presents an Adaptive Neuro‐Fuzzy Inference System (ANFIS) that combines fourth‐level Discrete Wavelet Transform (DWT) features with principled feature reduction and imbalance handling for robust classification. Three‐phase current and voltage waveforms in a power system network were simulated in MATLAB/Simulink across healthy operation and seven fault types under both ideal and noisy/disturbed regimes, yielding approximately 178,000 labelled records. From the DWT coefficients, Standard Deviation (SD) and Relative Wavelet Energy (RWE) were computed. Principal Component Analysis (PCA) and feature ranking then produced a compact seven‐feature set. Imbalance in the training partition was addressed using the Synthetic Minority Over‐Sampling Technique (SMOTE). With the ANFIS classifier, the proposed DWT feature set achieved a test accuracy of 97.22% under noise, with an operation time of 10.2 ms. The reliability of the proposed system was tested by simulating it under various internal and external faults, as well as different practical conditions, including magnetising inrush, on‐load magnetising transient, switching transient, and saturation, while varying the transformer tap positions. Finally, the designed system was validated using a secondary dataset and another power transformer rating.
ABSTRACT This study broadly assessed the energy, economic, and environmental performance of shifting from charcoal to liquefied petroleum gas (LPG) and compressed natural gas (CNG). The energy performance focuses on comparing energy consumption and thermal efficiency (ηth), while in the economy, the focus is on daily and annual fuel costs. In environmental benefits, pollutants such as carbon dioxide (CO2), carbon monoxide (CO), nitrogen oxide (NO), sulphur dioxide (SO2) and noxious (NOx) were compared between LPG, CNG and charcoal modes during water boiling tests (WBTs). Findings showed that WBT under LPG mode had an average ηth of 62.93% compared to 50.63% in CNG mode. Besides, shifting from charcoal to LPG could save around 73.7% of fuel, while charcoal to CNG could save about 72.9% of fuel. In an economic point of view, transitioning from charcoal to LPG and CNG might save up to 70.7 and 71.7% of the fuel cost each year, respectively. In environmental benefits, shifting from charcoal to LPG could reduce CO2, CO, NO and NOx by approximately 3.20; 11.13; 14.76; and 14.90 times, while shifting from charcoal to CNG could save CO2, CO, NO and NOx by approximately 3.45, 2.68, 12.18 and 12.30 times, respectively. The findings revealed that the CO2 level aligns with the recommended levels of the WHO. The CO levels from LPG, CNG and charcoal exceeded the WHO levels by 4.9, 20.4 and 54.8 times, shedding light on stakeholders to find strategies to lessen the CO levels.
ABSTRACT Detecting anomalies such as bumps and dents on reflective and non‐reflective surfaces like vehicle surfaces is difficult due to specular noise, diffuse lighting, and unpredictable reflections. Traditionally, convolutional neural networks (CNNs) are used for this task; however, they suffer from limitations such as overfitting, high dependency on labelled data, and difficulty in detecting subtle or complex anomalies. CNNs often fail to extract and isolate high‐level semantic features, leading to poor performance in critical industrial quality control applications. A potential substitute for CNNs is vision transformers (ViTs), offering better global context awareness and improved feature representation. Still, ViTs require large datasets for training and are computationally intensive, which limits their use in real‐world industrial settings. To overcome these issues, we propose a hybrid algorithm combining MobileNet V3 and ResNet‐34, enhanced with a sequential attention network (SAN‐RNI). This model integrates deflectometry‐based information with channel‐wise attention mechanisms to better detect color and texture anomalies. By emphasising important areas and reducing extraneous background information, the attention layers increase accuracy and resilience. Our approach outperforms traditional CNN‐based models in terms of accuracy and loss, as demonstrated by experimental findings on the MVTec AD and deflectometry datasets. This indicates the method's potential for dependable automated visual inspection in industrial settings.
ABSTRACT Conditional generative adversarial networks (cGANs) extend conventional GANs by incorporating class information into the image generation process, enabling controlled and class‐consistent synthesis. In the medical domain, cGANs are increasingly important for addressing the scarcity of high‐quality annotated datasets, which limits the development of robust diagnostic models. However, existing cGAN‐based approaches often suffer from unstable training, mode collapse, and insufficient preservation of fine‐grained diagnostic features. In this study, we propose Spectral Normalisation & Efficient Perceptual‐cGAN (SNEP‐cGAN), which is designed to generate high‐fidelity medical images. The model integrates spectral normalisation for training stability, residual up‐sampling and down‐sampling blocks with squeeze‐and‐excitation attention for enhanced feature representation, and a perceptual loss computed using a pretrained EfficientNetB0 network to improve visual realism. A gradient penalty is further employed to enforce Lipschitz continuity. The proposed model is evaluated on the NIH Malaria Cell Dataset and a breast cancer imaging dataset, achieving a remarkably low Fréchet Inception Distance (FID) of 4.28 and strong class‐wise diversity, notably outperforming baseline models. These results demonstrate that SNEP‐cGAN provides a robust and effective solution for medical image synthesis, offering strong potential for dataset augmentation and improved computer‐aided diagnosis in healthcare applications.
ABSTRACT A converter transformer is one of the most widely used power equipment. This paper proposes a statistical dissolved gas analysis (DGA) diagnosis method to evaluate the occurrence and type of the discharge starting stage of the converter transformer based on a statistical analysis of the gas generation amount in the converter transformer oil during operation. The method is similar to case retrieval in case‐based reasoning (CBR). The basic data, which are corresponding to the case base, are sample populations consisting of the model test data for the gas generation amount at the discharge starting stage of typical discharge models with oil‐paper insulation under various AC‐DC composite voltage ratios. The diagnosis method, which is corresponding to the case retrieval, is a robust statistical analysis. The typical discharge models are column‐plate, air gap, suspension, and needle‐plate electrode models. By the statistical DGA diagnosis, we can obtain the diagnosis probability density values and the statistical hypothesis test results for the occurrence and type of the discharge starting stage. We experimentally verified the effectiveness of the proposed method with the test data of each typical discharge model for the AC‐DC composite voltage ratios of 1:1, 1:3, and 1:5.
ABSTRACT Accurate electricity demand forecasting is crucial for grid stability and resource optimization, yet predictive models degrade over time due to data drift caused by market fluctuations, policy changes, and infrastructure shifts. Undetected drift leads to forecasting errors, inefficiencies in energy distribution, and increased operational costs. This paper presents a framework for detecting data drift in electric load time series, utilizing statistical and machine learning techniques, with a particular emphasis on the pruned exact linear time (PELT) algorithm for efficient change‐point detection. Our method is scalable and highly efficient, making it suitable for real‐time applications. Unlike traditional drift detection techniques, our approach dynamically adjusts to evolving patterns, mitigating both gradual and abrupt changes in consumption behavior. By leveraging both synthetically generated multivariate data and real‐world univariate data for electricity load‐related time series, the effectiveness of PELT is evaluated demonstrating its ability to capture structural shifts with high accuracy. The proposed framework achieved F1 score of 82% and 98% for generated temperature and humidity time series, respectively. Similarly, it achieved F1 score of 94% for real‐world electricity demand data. We also assessed the quality of PELT's detection by applying time series smoothing techniques that provided additional insights. These findings highlighted the practical implications of our approach in improving forecasting models that can potentially enable proactive decision‐making and ensure resilience in energy management systems.
