Oleksandr Rubanenko
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Oleksandr Rubanenko has a total of 76 co-authors. Here are the top co-authors by number of shared publications:
Mohit Bajaj
Department of Electrical Engineering, Graphic Era (Deemed to be University), Dehradun, 248002, India
3 shared publications- Mohit Bajaj
Department of Electrical Engineering, Graphic Era (Deemed to be University), Dehradun, India
3 shared publications - Aykut Fatih Güven
Department of Electrical and Electronics Engineering, Yalova University, Yalova, Turkey
2 shared publications - Bhabasis Mohapatra
ITER, Siksha ‘O’ Anusandhan (Deemed to be University)
2 shared publications El Moundher Aouiche
School of Electrical Engineering, Hebei University of Technology, Tianjin, China
2 shared publicationsFuta Osumanu
Department of Electrical/Electronic Engineering, Accra Technical University, Accra, Ghana
2 shared publications
Latest Publications
- Open Access
- 28 September 2026
- Published by Springer Nature in International Journal of Computational Intelligence Systems
Stochastic Modeling of Finite-Capacity Electric Vehicle Charging Infrastructures with Flexible Service Provisioning
In the recent past, the management of electric vehicle (EV) charging infrastructure received increased attention due to its significant role in enabling the transition to sustainable transportation. The current study investigates a flexible queueing mechanism for EV charging stations under two types of service disruptions. The model incorporates negative and positive arrivals, partial server breakdowns, and a threshold-based repair mechanism. The proposed modeling reveals that negative arrivals, indicating cancelled or withdrawn charge requests, interrupt ongoing services. Such disruption adversely impacts the overall operational efficiency of the charging infrastructure. For modeling purposes, the hypothesis of the quasi-birth-and-death (QBD) process in unification with the matrix geometric solution approach is utilized. The framework establishes the stationary states of the system, determines the steady-state probability vectors, and estimates numerous essential performance metrics, including average waiting time at the charging system, system utilization, reliability measures, etc. Numerical studies reveal that when the negative arrival rate increases, the average waiting time and system utilization decrease progressively, while the negative arrival rate also has a decreasing congestion effect on the system. Further investigations indicate that repair rates have a higher impact on system throughput than individual breakdown rates, emphasizing the necessity of prompt repair provisioning. Further, graphical representations demonstrate how variations in critical parameters influence system performance. These insights help improve the operational efficiency of EV charging systems in practice. The proposed research addresses how adverse demand fluctuations could impact initiatives to improve the adaptability and flexibility of EV charging infrastructures, thereby facilitating the wider adoption of EV transportation.
- 28 September 2026
- Published by Springer Nature in International Journal of Computational Intelligence Systems
Open AccessStochastic Modeling of Finite-Capacity Electric Vehicle Charging Infrastructures with Flexible Service Provisioning
In the recent past, the management of electric vehicle (EV) charging infrastructure received increased attention due to its significant role in enabling the transition to sustainable transportation. The current study investigates a flexible queueing mechanism for EV charging stations under two types of service disruptions. The model incorporates negative and positive arrivals, partial server breakdowns, and a threshold-based repair mechanism. The proposed modeling reveals that negative arrivals, indicating cancelled or withdrawn charge requests, interrupt ongoing services. Such disruption adversely impacts the overall operational efficiency of the charging infrastructure. For modeling purposes, the hypothesis of the quasi-birth-and-death (QBD) process in unification with the matrix geometric solution approach is utilized. The framework establishes the stationary states of the system, determines the steady-state probability vectors, and estimates numerous essential performance metrics, including average waiting time at the charging system, system utilization, reliability measures, etc. Numerical studies reveal that when the negative arrival rate increases, the average waiting time and system utilization decrease progressively, while the negative arrival rate also has a decreasing congestion effect on the system. Further investigations indicate that repair rates have a higher impact on system throughput than individual breakdown rates, emphasizing the necessity of prompt repair provisioning. Further, graphical representations demonstrate how variations in critical parameters influence system performance. These insights help improve the operational efficiency of EV charging systems in practice. The proposed research addresses how adverse demand fluctuations could impact initiatives to improve the adaptability and flexibility of EV charging infrastructures, thereby facilitating the wider adoption of EV transportation.
- Open Access
- 13 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Leader follower second order voltage control with disturbance observer for DC microgrids
This paper proposes an advanced secondary voltage control framework for a renewable- and electric-vehicle (EV)-integrated DC microgrid using a leader-follower multi-agent consensus architecture. To coordinate power sharing among heterogeneous battery energy storage systems (BESS), a second-order distributed voltage control strategy is developed, wherein battery voltage dynamics are modeled as second-order agents to enhance transient performance and convergence characteristics. Unlike conventional approaches that rely on random or heuristic leader assignment, this work introduces a theoretically grounded cost-index-based leader selection criterion, derived from an analytical Riccati formulation using battery ampere-hour (Ah) ratings. The proposed criterion enables the selection of an optimal battery leader that minimizes quadratic energy cost and battery stress, thereby reducing control effort and improving overall system efficiency. Following optimal leader identification, the paper addresses the challenge of continuous disturbances and uncertainties arising from heterogeneous loads, renewable intermittency, EV charging demand, and communication imperfections. To this end, a novel disturbance-observer-based heterogeneous consensus control scheme is integrated into the secondary control layer. The proposed second-order disturbance observer enables accurate estimation and compensation of unknown current and power disturbances acting on follower agents, ensuring robust voltage regulation and coordinated power sharing under dynamic operating conditions. The developed control framework is fully distributed, requires only sparse low-bandwidth communication among neighboring agents, and preserves system stability under battery disconnection/reconnection, plug-and-play operation, communication delays, and link failures. Comprehensive MATLAB/Simulink case studies on a four-bus DC microgrid comprising photovoltaic sources, wind generation, heterogeneous batteries, DC loads, and EV charging stations validate the effectiveness of the proposed approach. Simulation results demonstrate significantly reduced DC bus voltage deviation, faster convergence, improved state-of-charge (SoC) balancing, and lower power mismatch compared with conventional droop control, homogeneous consensus control, existing leader-follower strategies, and sliding-mode-based distributed controllers. The proposed method thus offers a robust, scalable, and energy-efficient secondary voltage control solution for next-generation DC microgrids with high renewable and EV penetration.
- 13 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessLeader follower second order voltage control with disturbance observer for DC microgrids
This paper proposes an advanced secondary voltage control framework for a renewable- and electric-vehicle (EV)-integrated DC microgrid using a leader-follower multi-agent consensus architecture. To coordinate power sharing among heterogeneous battery energy storage systems (BESS), a second-order distributed voltage control strategy is developed, wherein battery voltage dynamics are modeled as second-order agents to enhance transient performance and convergence characteristics. Unlike conventional approaches that rely on random or heuristic leader assignment, this work introduces a theoretically grounded cost-index-based leader selection criterion, derived from an analytical Riccati formulation using battery ampere-hour (Ah) ratings. The proposed criterion enables the selection of an optimal battery leader that minimizes quadratic energy cost and battery stress, thereby reducing control effort and improving overall system efficiency. Following optimal leader identification, the paper addresses the challenge of continuous disturbances and uncertainties arising from heterogeneous loads, renewable intermittency, EV charging demand, and communication imperfections. To this end, a novel disturbance-observer-based heterogeneous consensus control scheme is integrated into the secondary control layer. The proposed second-order disturbance observer enables accurate estimation and compensation of unknown current and power disturbances acting on follower agents, ensuring robust voltage regulation and coordinated power sharing under dynamic operating conditions. The developed control framework is fully distributed, requires only sparse low-bandwidth communication among neighboring agents, and preserves system stability under battery disconnection/reconnection, plug-and-play operation, communication delays, and link failures. Comprehensive MATLAB/Simulink case studies on a four-bus DC microgrid comprising photovoltaic sources, wind generation, heterogeneous batteries, DC loads, and EV charging stations validate the effectiveness of the proposed approach. Simulation results demonstrate significantly reduced DC bus voltage deviation, faster convergence, improved state-of-charge (SoC) balancing, and lower power mismatch compared with conventional droop control, homogeneous consensus control, existing leader-follower strategies, and sliding-mode-based distributed controllers. The proposed method thus offers a robust, scalable, and energy-efficient secondary voltage control solution for next-generation DC microgrids with high renewable and EV penetration.
- Open Access
- 5 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Adaptive long short term memory-PI based control for load frequency regulation in electric vehicle integrated two area power systems
The growing penetration of electric vehicles (EVs) and renewable energy sources has introduced significant nonlinear and stochastic disturbances in modern power systems, posing major challenges to conventional load frequency control (LFC) strategies. To address these issues, a Long Short-Term Memory-based Proportional Integral (LSTM-PI) controller is proposed to enable adaptive frequency regulation in a two-area EV-integrated system. The LSTM network dynamically tunes the proportional and integral gains in response to temporal variations in frequency and tie-line states. The learning module is trained using an Integral of Time weighted Absolute Error (ITAE) objective. Practical non-linearities such as governor deadband, generation-rate constraints, valve saturation, and EV state-of-charge limits are incorporated into both the training dataset and the online control evaluation. A constrained Model Predictive Control (CMPC-PI) is developed as a modern benchmark to compare the proposed technique under identical sampling rates and actuator limits. The proposed method shows better performance under dynamic conditions. Statistical performance analysis has been carried out through 30 runs of Monte Carlo analysis with a mean of 95% confidence intervals. Stability analysis under time-varying gains is supported by a bounded-gain Lyapunov Framework and a delay-margin analysis that addresses sensing and actuation delays. The result analysis confirms that the LSTM-PI controller provides a scalable intelligent solution for modern EV-integrated smart grid frequency regulation. The proposed intelligent adaptive controller provides a scalable, low-complexity, and stability-assured solution for robust load-frequency regulation in nonlinear, EV-integrated smart power grids.
- 5 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessAdaptive long short term memory-PI based control for load frequency regulation in electric vehicle integrated two area power systems
The growing penetration of electric vehicles (EVs) and renewable energy sources has introduced significant nonlinear and stochastic disturbances in modern power systems, posing major challenges to conventional load frequency control (LFC) strategies. To address these issues, a Long Short-Term Memory-based Proportional Integral (LSTM-PI) controller is proposed to enable adaptive frequency regulation in a two-area EV-integrated system. The LSTM network dynamically tunes the proportional and integral gains in response to temporal variations in frequency and tie-line states. The learning module is trained using an Integral of Time weighted Absolute Error (ITAE) objective. Practical non-linearities such as governor deadband, generation-rate constraints, valve saturation, and EV state-of-charge limits are incorporated into both the training dataset and the online control evaluation. A constrained Model Predictive Control (CMPC-PI) is developed as a modern benchmark to compare the proposed technique under identical sampling rates and actuator limits. The proposed method shows better performance under dynamic conditions. Statistical performance analysis has been carried out through 30 runs of Monte Carlo analysis with a mean of 95% confidence intervals. Stability analysis under time-varying gains is supported by a bounded-gain Lyapunov Framework and a delay-margin analysis that addresses sensing and actuation delays. The result analysis confirms that the LSTM-PI controller provides a scalable intelligent solution for modern EV-integrated smart grid frequency regulation. The proposed intelligent adaptive controller provides a scalable, low-complexity, and stability-assured solution for robust load-frequency regulation in nonlinear, EV-integrated smart power grids.
- Open Access
- 2 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
An unsupervised dual-phase framework combining statistical separation and hybrid level set evolution for robust motion detection and segmentation in intelligent video surveillance
Accurate detection and segmentation of moving objects constitute a fundamental challenge in computer vision, particularly for intelligent video surveillance systems operating under variable illumination, dynamic backgrounds, and environmental noise. This paper presents a fully unsupervised dual-phase motion analysis framework that effectively combines statistical independence modeling and geometric contour evolution to achieve high-precision motion detection and segmentation. In the first phase, an enhanced Fast Independent Component Analysis (Fast-ICA) algorithm is employed to perform statistical decomposition of video sequences, exploiting temporal independence to distinguish moving foregrounds from static backgrounds. This process generates an initial motion mask with strong robustness to illumination variation and noise artifacts. In the second phase, a hybrid level set segmentation model integrating the global Chan-Vese formulation and a locally adaptive Yezzi-based energy function refines object boundaries through an adaptive energy minimization process. A stabilization term and a self-regulating convergence criterion are further incorporated to ensure contour smoothness, numerical stability, and resilience to topological changes. Comprehensive experiments conducted on the CDNet-2014 benchmark dataset demonstrate that the proposed method achieves an average recall of 0.9613, precision of 0.9089, and F-measure of 0.9310, outperforming several state-of-the-art supervised, semi-supervised and unsupervised background subtraction algorithms. The proposed Fast-ICA–Level Set fusion framework thus provides a robust, adaptive, and computationally efficient solution for real-world intelligent surveillance and autonomous visual monitoring applications.
- 2 May 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessAn unsupervised dual-phase framework combining statistical separation and hybrid level set evolution for robust motion detection and segmentation in intelligent video surveillance
Accurate detection and segmentation of moving objects constitute a fundamental challenge in computer vision, particularly for intelligent video surveillance systems operating under variable illumination, dynamic backgrounds, and environmental noise. This paper presents a fully unsupervised dual-phase motion analysis framework that effectively combines statistical independence modeling and geometric contour evolution to achieve high-precision motion detection and segmentation. In the first phase, an enhanced Fast Independent Component Analysis (Fast-ICA) algorithm is employed to perform statistical decomposition of video sequences, exploiting temporal independence to distinguish moving foregrounds from static backgrounds. This process generates an initial motion mask with strong robustness to illumination variation and noise artifacts. In the second phase, a hybrid level set segmentation model integrating the global Chan-Vese formulation and a locally adaptive Yezzi-based energy function refines object boundaries through an adaptive energy minimization process. A stabilization term and a self-regulating convergence criterion are further incorporated to ensure contour smoothness, numerical stability, and resilience to topological changes. Comprehensive experiments conducted on the CDNet-2014 benchmark dataset demonstrate that the proposed method achieves an average recall of 0.9613, precision of 0.9089, and F-measure of 0.9310, outperforming several state-of-the-art supervised, semi-supervised and unsupervised background subtraction algorithms. The proposed Fast-ICA–Level Set fusion framework thus provides a robust, adaptive, and computationally efficient solution for real-world intelligent surveillance and autonomous visual monitoring applications.
- Open Access
- 17 April 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Machine learning forecasting of electroosmotic MHD flow of a buoyancy-driven ternary nanofluid in converging and diverging channels
Artificial neural networks (ANNs) are utilized to study the buoyancy-driven electroosmotic flow conveying Blood, Gold (Au), Copper (Cu), and Multi-Walled Carbon Nanotubes in convergent ciliated tube (CCT) and divergent ciliated tube (DCT) micro-vessel systems. The research includes significant physical mechanisms, such as magnetohydrodynamic, heat generation, Jeffrey fluid model, Non-dimensional electric potential, and mixed convection resulting from buoyancy force influence. The Debye–Hückel approximation is used to model the electroosmotic influence using the Helmholtz–Smoluchowski velocity. The nonlinear governing equations are solved using the finite element method (FEM) to produce data sets of high quality. An ANNs, which involves the backpropagation LevenbergMarquardt scheme (ANN-BPLMS), is constructed using a large number of data points of the FEM solutions to increase prediction accuracy. These data are then utilized to test and train the artificial neural networks model and precisely forecast velocity field and thermal profile in a broad variety of controlling parameters. The intelligent computing outcomes display strong agreement with the finite element method solutions. The results show that electroosmotic effect effectively regulate flow in both CCT and (DCT) micro-vessel systems, whereas buoyancy forces dramatically improved axial velocity field and temperature profile. Furthermore, the inclusion of \(Au\)-\(Cu\)-\(MWCNT\)/\(Blood\) ternary hybrid nanofluid significantly improves thermal heat transfer rate compared to \(Au\)-\(Cu\)/\(Blood\) hybrid nanofluid.- 17 April 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessMachine learning forecasting of electroosmotic MHD flow of a buoyancy-driven ternary nanofluid in converging and diverging channels
Artificial neural networks (ANNs) are utilized to study the buoyancy-driven electroosmotic flow conveying Blood, Gold (Au), Copper (Cu), and Multi-Walled Carbon Nanotubes in convergent ciliated tube (CCT) and divergent ciliated tube (DCT) micro-vessel systems. The research includes significant physical mechanisms, such as magnetohydrodynamic, heat generation, Jeffrey fluid model, Non-dimensional electric potential, and mixed convection resulting from buoyancy force influence. The Debye–Hückel approximation is used to model the electroosmotic influence using the Helmholtz–Smoluchowski velocity. The nonlinear governing equations are solved using the finite element method (FEM) to produce data sets of high quality. An ANNs, which involves the backpropagation LevenbergMarquardt scheme (ANN-BPLMS), is constructed using a large number of data points of the FEM solutions to increase prediction accuracy. These data are then utilized to test and train the artificial neural networks model and precisely forecast velocity field and thermal profile in a broad variety of controlling parameters. The intelligent computing outcomes display strong agreement with the finite element method solutions. The results show that electroosmotic effect effectively regulate flow in both CCT and (DCT) micro-vessel systems, whereas buoyancy forces dramatically improved axial velocity field and temperature profile. Furthermore, the inclusion of \(Au\)-\(Cu\)-\(MWCNT\)/\(Blood\) ternary hybrid nanofluid significantly improves thermal heat transfer rate compared to \(Au\)-\(Cu\)/\(Blood\) hybrid nanofluid. - Open Access
- 31 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Performance comparison of CTAB-modified NiO, ZnO, and SnO2 sensors for CO and CH4 detection in environmental and health applications
In industrial and environmental monitoring, high-sensitivity and response-time sensors are a major challenge. In this research, ZnO, NiO, and SnO2 nanostructures were prepared via a Surfactant (CTAB)-assisted hydrothermal process. Surfactant improves surface activity and porosity for controlled development. A stable p–n heterojunction was formed by including PEDOT: PSS in n-type oxides. The fabricated sensors were put through an intensive testing procedure for the detection of CO and CH4 at ambient temperature (30 °C) and at different concentrations (2–10 ppm). The maximum sensitivity of the PEDOT: PSS/SnO2 heterojunction was 92.3% for CO and 90.8% for CH4. Its response and recovery times were 10/5 s and 15/8 s, respectively. The developed sensors exhibited a linear response pattern and excellent repeatability across a 24-h evaluation, with five testing cycles of 250 s each. Characterization of structural and surface properties (XRD, EDS, BET, UV, JV, and EIS) confirmed that high crystallinity, nanoscale structure, and enhanced surface area were responsible for enhanced sensing performance. Based on the results, PEDOT: PSS heterojunction assembly combined with CTAB-assisted hydrothermal synthesis may be employed to synthesize high-performance, low-cost gas sensors for real-time detection of harmful and inflammable gases.
- 31 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessPerformance comparison of CTAB-modified NiO, ZnO, and SnO2 sensors for CO and CH4 detection in environmental and health applications
In industrial and environmental monitoring, high-sensitivity and response-time sensors are a major challenge. In this research, ZnO, NiO, and SnO2 nanostructures were prepared via a Surfactant (CTAB)-assisted hydrothermal process. Surfactant improves surface activity and porosity for controlled development. A stable p–n heterojunction was formed by including PEDOT: PSS in n-type oxides. The fabricated sensors were put through an intensive testing procedure for the detection of CO and CH4 at ambient temperature (30 °C) and at different concentrations (2–10 ppm). The maximum sensitivity of the PEDOT: PSS/SnO2 heterojunction was 92.3% for CO and 90.8% for CH4. Its response and recovery times were 10/5 s and 15/8 s, respectively. The developed sensors exhibited a linear response pattern and excellent repeatability across a 24-h evaluation, with five testing cycles of 250 s each. Characterization of structural and surface properties (XRD, EDS, BET, UV, JV, and EIS) confirmed that high crystallinity, nanoscale structure, and enhanced surface area were responsible for enhanced sensing performance. Based on the results, PEDOT: PSS heterojunction assembly combined with CTAB-assisted hydrothermal synthesis may be employed to synthesize high-performance, low-cost gas sensors for real-time detection of harmful and inflammable gases.
- Open Access
- 27 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
A bio-inspired swarm UAV framework integrating thermal sensing and optimization-based coordination for efficient search and rescue operations
Search and rescue (SAR) operations demand rapid, reliable detection of survivors in disaster-stricken environments where time, terrain, and responder safety are critical constraints. While Unmanned Aerial Vehicles (UAVs) equipped with thermal imaging offer a promising aerial solution, current approaches often struggle to balance large-area coverage, responsiveness to thermal cues, and avoidance of redundant search paths. This paper proposes a modular, bio-inspired swarm UAV framework that enables real-time thermal-based SAR through autonomous, cooperative exploration of a discretized search grid. Each UAV operates as an intelligent agent, leveraging bio-inspired optimization algorithms to determine its next target location, with decision-making grounded in a shared thermal confidence map. A novel Exploration Score metric is introduced to quantitatively assess search efficiency by integrating area coverage, redundancy minimization, and spatial dispersion of the swarm. The system is implemented in a high-fidelity PX4 + Gazebo simulation environment, featuring real-time thermal detection and multi-drone coordination. Ten bio-inspired algorithms are evaluated under identical conditions, with Particle Swarm Optimization (PSO) achieving the highest exploration score of 0.67, outperforming Grey Wolf Optimizer (0.62) and Ant Colony Optimization (0.59). PSO also achieved 80% area coverage within 60% of the mission time while maintaining a low redundancy ratio (~ 0.25) and balanced inter-drone separation (15–20 m). These results confirm the framework’s effectiveness in enabling fast, adaptive, and energy-efficient thermal search missions, significantly enhancing the probability of early survivor detection. The proposed architecture serves as a benchmarking platform for UAV swarm coordination, and the exploration score metric offers a unified performance measure for future SAR algorithm comparisons.
- 27 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessA bio-inspired swarm UAV framework integrating thermal sensing and optimization-based coordination for efficient search and rescue operations
Search and rescue (SAR) operations demand rapid, reliable detection of survivors in disaster-stricken environments where time, terrain, and responder safety are critical constraints. While Unmanned Aerial Vehicles (UAVs) equipped with thermal imaging offer a promising aerial solution, current approaches often struggle to balance large-area coverage, responsiveness to thermal cues, and avoidance of redundant search paths. This paper proposes a modular, bio-inspired swarm UAV framework that enables real-time thermal-based SAR through autonomous, cooperative exploration of a discretized search grid. Each UAV operates as an intelligent agent, leveraging bio-inspired optimization algorithms to determine its next target location, with decision-making grounded in a shared thermal confidence map. A novel Exploration Score metric is introduced to quantitatively assess search efficiency by integrating area coverage, redundancy minimization, and spatial dispersion of the swarm. The system is implemented in a high-fidelity PX4 + Gazebo simulation environment, featuring real-time thermal detection and multi-drone coordination. Ten bio-inspired algorithms are evaluated under identical conditions, with Particle Swarm Optimization (PSO) achieving the highest exploration score of 0.67, outperforming Grey Wolf Optimizer (0.62) and Ant Colony Optimization (0.59). PSO also achieved 80% area coverage within 60% of the mission time while maintaining a low redundancy ratio (~ 0.25) and balanced inter-drone separation (15–20 m). These results confirm the framework’s effectiveness in enabling fast, adaptive, and energy-efficient thermal search missions, significantly enhancing the probability of early survivor detection. The proposed architecture serves as a benchmarking platform for UAV swarm coordination, and the exploration score metric offers a unified performance measure for future SAR algorithm comparisons.
- Open Access
- 22 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 44286
Adaptive vector oriented control of doubly fed induction generator wind turbines using M5P model tree for robust dynamic performance
Wind energy systems using doubly fed induction generators (DFIGs) rely on vector-oriented control (VOC) to achieve decoupled regulation of active and reactive power. Conventional VOC strategies typically employ proportional-integral (PI) controllers for rotor current control; however, these controllers struggle under nonlinear and variable operating conditions, such as wind speed fluctuations, leading to degraded dynamic performance and power quality. To overcome these limitations, this paper proposes an intelligent VOC approach based on the M5-Pruned model tree (M5P) algorithm, complemented by a comparative study with fuzzy logic controllers (FLC). The M5P-based controller introduces a data-driven mechanism that adapts to system dynamics, ensuring improved accuracy and robustness compared to traditional PI and FLC methods. A detailed MATLAB/Simulink simulation of a grid-connected wind turbine with DFIG evaluates the controllers under varying wind profiles. Results demonstrate that the proposed M5P strategy significantly reduces overshoot and settling time, enhances power regulation, and improves overall system stability. These findings highlight the advantages of machine learning-based control in addressing the shortcomings of conventional VOC techniques for renewable energy applications.
- 22 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 44286
Open AccessAdaptive vector oriented control of doubly fed induction generator wind turbines using M5P model tree for robust dynamic performance
Wind energy systems using doubly fed induction generators (DFIGs) rely on vector-oriented control (VOC) to achieve decoupled regulation of active and reactive power. Conventional VOC strategies typically employ proportional-integral (PI) controllers for rotor current control; however, these controllers struggle under nonlinear and variable operating conditions, such as wind speed fluctuations, leading to degraded dynamic performance and power quality. To overcome these limitations, this paper proposes an intelligent VOC approach based on the M5-Pruned model tree (M5P) algorithm, complemented by a comparative study with fuzzy logic controllers (FLC). The M5P-based controller introduces a data-driven mechanism that adapts to system dynamics, ensuring improved accuracy and robustness compared to traditional PI and FLC methods. A detailed MATLAB/Simulink simulation of a grid-connected wind turbine with DFIG evaluates the controllers under varying wind profiles. Results demonstrate that the proposed M5P strategy significantly reduces overshoot and settling time, enhances power regulation, and improves overall system stability. These findings highlight the advantages of machine learning-based control in addressing the shortcomings of conventional VOC techniques for renewable energy applications.
- Open Access
- 17 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
An efficient transformer architecture with depthwise separable convolutions for high-accuracy underwater acoustic target recognition
Underwater Acoustic Target Recognition (UATR) plays a vital role in maritime security and defense, requiring accurate and efficient classification of marine vessels based on their sonar acoustic emissions. Traditional recognition systems rely on handcrafted features and shallow classifiers, which often struggle with complex acoustic patterns and impose substantial computational overhead. Although deep learning methods have improved recognition accuracy, their high computational demands hinder real-time deployment on resource-constrained platforms. We propose the Depthwise Separable Convolutional Multihead Transformer (DCMT), which combines depthwise separable convolutions for localized feature extraction with multi-head self-attention Transformer branches for global contextual modeling. The model has dual Transformer’s parallel branches having 4-head and 8-head structures for additional complementary feature processing, and their outputs are merged using Global Average and Max Pooling to form a more potent feature vector. The model incorporates the following acoustic features: Zero Crossing Rate (ZCR), Root Mean Square-Energy (RMS-Energy), Mel-Frequency Cepstral Coefficients (MFCCs), and Chroma. To enhance generalization of DCMT model, CutMix data augmentation is used to synthetically increase data variability through combining audio segments from different classes. The proposed DCMT model with 0.7 million parameters was evaluated on the public benchmark datasets DeepShip and ShipsEar achieving 0.847 Flops (G) and 0.94 Flops (G), respectively, and it reached a classification accuracy of 97.53% and 98.19%. The model also attained even better classification accuracy on the QiandaoEar22 datasets (SpeedBoat, KaiYuan, and UUV) which are 95.08%, 98.24% and 99.68%, respectively. Moreover, the proposed DCMT model achieves an average inference time of 3.8 ms and 131.6 FPS, which outperform existing models and the baseline UTAR-Transformer (4.3 ms and 230.8 FPS) in both speed and accuracy. This lightweight model exhibits strong potential for real-time deployment in marine environments.
- 17 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessAn efficient transformer architecture with depthwise separable convolutions for high-accuracy underwater acoustic target recognition
Underwater Acoustic Target Recognition (UATR) plays a vital role in maritime security and defense, requiring accurate and efficient classification of marine vessels based on their sonar acoustic emissions. Traditional recognition systems rely on handcrafted features and shallow classifiers, which often struggle with complex acoustic patterns and impose substantial computational overhead. Although deep learning methods have improved recognition accuracy, their high computational demands hinder real-time deployment on resource-constrained platforms. We propose the Depthwise Separable Convolutional Multihead Transformer (DCMT), which combines depthwise separable convolutions for localized feature extraction with multi-head self-attention Transformer branches for global contextual modeling. The model has dual Transformer’s parallel branches having 4-head and 8-head structures for additional complementary feature processing, and their outputs are merged using Global Average and Max Pooling to form a more potent feature vector. The model incorporates the following acoustic features: Zero Crossing Rate (ZCR), Root Mean Square-Energy (RMS-Energy), Mel-Frequency Cepstral Coefficients (MFCCs), and Chroma. To enhance generalization of DCMT model, CutMix data augmentation is used to synthetically increase data variability through combining audio segments from different classes. The proposed DCMT model with 0.7 million parameters was evaluated on the public benchmark datasets DeepShip and ShipsEar achieving 0.847 Flops (G) and 0.94 Flops (G), respectively, and it reached a classification accuracy of 97.53% and 98.19%. The model also attained even better classification accuracy on the QiandaoEar22 datasets (SpeedBoat, KaiYuan, and UUV) which are 95.08%, 98.24% and 99.68%, respectively. Moreover, the proposed DCMT model achieves an average inference time of 3.8 ms and 131.6 FPS, which outperform existing models and the baseline UTAR-Transformer (4.3 ms and 230.8 FPS) in both speed and accuracy. This lightweight model exhibits strong potential for real-time deployment in marine environments.
- Open Access
- 8 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
A resource-efficient machine learning framework for real-time non-intrusive load monitoring and performance optimization in solar-powered aviation systems
The integration of non-intrusive load monitoring (NILM) into solar-powered aviation systems presents a transformative approach for achieving sustainable, lightweight, and intelligent flight operations. However, the sector’s stringent constraints on weight, latency, and computational resources pose critical challenges to real-time NILM deployment. This study develops a resource-efficient machine learning framework that systematically evaluates six machine learning (ML) and deep learning (DL) models using high-resolution (200 kHz) power data that capture both transient and steady-state load characteristics across all flight phases. Advanced preprocessing techniques—comprising moving average smoothing and non-overlapping downsampling—were applied to suppress noise while preserving essential features. The comparative analysis reveals that K-Nearest neighbors (KNN) delivers the most effective balance between accuracy and computational cost, achieving an R² of 0.9403 with an execution time of 0.20 s, substantially outperforming ensemble models such as random forest (RF) and XGBoost in real-time feasibility. Conversely, the hybrid CNN-LSTM architecture attained the lowest mean squared error (MSE = 0.0048) and superior temporal sensitivity but required 271.53 s, demonstrating its suitability for offline analysis rather than onboard deployment. Through comprehensive hardware-in-the-loop validation using Opal-RT and Launchpad-F28379D DSP controllers, the framework verified appliance-level disaggregation accuracy under dynamic flight scenarios. The findings underscore a critical accuracy–efficiency trade-off in NILM model selection, establishing that traditional ML algorithms can outperform complex DL models when optimized for real-time, resource-constrained environments. This research provides actionable design insights for next-generation solar-powered aircraft energy management systems, demonstrating that model selection must prioritize computational efficiency, predictive reliability, and real-time responsiveness to enable sustainable and intelligent aviation energy control.
- 8 December 2025
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessA resource-efficient machine learning framework for real-time non-intrusive load monitoring and performance optimization in solar-powered aviation systems
The integration of non-intrusive load monitoring (NILM) into solar-powered aviation systems presents a transformative approach for achieving sustainable, lightweight, and intelligent flight operations. However, the sector’s stringent constraints on weight, latency, and computational resources pose critical challenges to real-time NILM deployment. This study develops a resource-efficient machine learning framework that systematically evaluates six machine learning (ML) and deep learning (DL) models using high-resolution (200 kHz) power data that capture both transient and steady-state load characteristics across all flight phases. Advanced preprocessing techniques—comprising moving average smoothing and non-overlapping downsampling—were applied to suppress noise while preserving essential features. The comparative analysis reveals that K-Nearest neighbors (KNN) delivers the most effective balance between accuracy and computational cost, achieving an R² of 0.9403 with an execution time of 0.20 s, substantially outperforming ensemble models such as random forest (RF) and XGBoost in real-time feasibility. Conversely, the hybrid CNN-LSTM architecture attained the lowest mean squared error (MSE = 0.0048) and superior temporal sensitivity but required 271.53 s, demonstrating its suitability for offline analysis rather than onboard deployment. Through comprehensive hardware-in-the-loop validation using Opal-RT and Launchpad-F28379D DSP controllers, the framework verified appliance-level disaggregation accuracy under dynamic flight scenarios. The findings underscore a critical accuracy–efficiency trade-off in NILM model selection, establishing that traditional ML algorithms can outperform complex DL models when optimized for real-time, resource-constrained environments. This research provides actionable design insights for next-generation solar-powered aircraft energy management systems, demonstrating that model selection must prioritize computational efficiency, predictive reliability, and real-time responsiveness to enable sustainable and intelligent aviation energy control.