As Internet of Things (IoT) technology advances rapidly, intelligent substation monitoring systems become crucial for power system stability. However, the transmission of its monitoring data faces security threats like theft and tampering, so it is urgent to take effective measures to ensure the data security during transmission. This paper introduces a GAN-based steganography (DM-GAN) to secure intelligent substation monitoring data transmission. Firstly, the dual-stream generator and a multi-steganalyzer in the discriminator are adopted in DM-GAN, which balances the generator and discriminator. Secondly, the skip connections in DM-GAN are redesigned to fully integrate multi-scale features and enhance the model’s understanding of image details and structures. Furthermore, the proposed spatial-channel combined attention (SCCA) mechanism enables the generator to focus on image key regions adaptively, improving the embedded probability map quality. Finally, compared with traditional steganography HILL and GAN-based methods UT-GAN, CF-UT-GAN, Wang-Net, and Steg-GMAN, steganalyzer detection accuracy of our proposed method is reduced by 4.82%, 3.11%, 1.71%, 0.81%, and 0.55% on average, respectively. Experimental results show that DM-GAN has achieved a significant improvement in steganalysis resistance ability, and has effectiveness and superiority in substation monitoring system.

In nonlinear systems, the system response frequently fluctuates with the discrepancy of variation trends over a certain horizon. In this situation, type-2 fuzzy neural networks can be limited in capturing trend features from the point prediction process due to the inherent value mapping, which may suffer from trend blindness. To address this problem, a trend-aware-based type-2 vector fuzzy neural network (TA-T2VFNN) is designed for modeling nonlinear systems. First, a quantization framework with a hybrid encoding strategy is designed to process both the magnitude and trend information of variables. In this strategy, the trend information with the rate and direction of change is represented by vectors instead of scalars, which aims to compensate for the accumulated error caused by insufficient input information. Second, a directional attention mechanism-based vector fuzzy rule is proposed to capture the trend relationship to mimic the local fluctuation patterns. Then, the vector operation is embedded into fuzzy rules for geometric inference, which can obtain the vector features with point and trend. Third, a collaborative feedback learning algorithm is developed to update the magnitude and direction parameters of TA-T2VFNN. Then, the modeling accuracy can be maintained by obtaining trusted output points and output trends. Finally, the effectiveness of TA-T2VFNN is verified by multiple practical applications.

Large-scale wind grid integration weakens frequency regulation due to the use of converters, which decouple wind turbine rotor speed from grid frequency. Pertinent studies have primarily focused on frequency regulation through adjusting rotor speed and pitch angle on the mechanical side, or by controlling direct-axis current of the motor, but adjusting rotor current phase is also possible. Thus, based on rotor current phase adjustment, this article proposed a deep reinforcement learning coordinated control strategy for a doubly-fed induction generator (DFIG) wind turbine to coordinate the participation of the mechanical and motor sides in frequency regulation. To establish the relationship between rotor current phase and active power, a signal named driving phase angle (DPA) is constructed and introduced into the dq/abc transform in the rotor-side vector control, to change the rotor current phase angle and, furthermore, the power output. Then, the model for DPA influencing DFIG power output is established, and its static and dynamic characteristics are clarified. Furthermore, combined with deep reinforcement learning, a coordinated control strategy is proposed and solved by the deep deterministic policy gradient algorithm. Finally, the simulation results show that the proposed strategy effectively improves the performance of frequency regulation.

Recent advances in data-driven methods have significantly improved battery capacity estimation, yet most existing approaches remain constrained by their reliance on supervised learning, requiring substantial amounts of labeled cycling data that are often costly to obtain. To address this challenge, this study proposes a dual-branch network-based semi supervised framework that integrates self-supervised learning and transfer learning mechanisms. First, the framework derives meaningful degradation-aware auxiliary labels from both labeled and unlabeled samples, creating reliable self-supervised signal for model training. Second, the designed dual-branch neural network architecture employs a shared feature extractor that processes input data for both the primary capacity estimation task and the auxiliary label prediction task, enabling effective knowledge transfer between labeled and unlabeled domains through their common representation space. Third, a pseudolabel filtering strategy is proposed to dynamically select high-confidence samples from the unlabeled dataset for self-training, thereby effectively expanding the training set with high-quality pseudolabels and enhancing the capacity estimation accuracy. Finally, extensive experiments validate the framework’s superior performance, achieving a worst-case root-mean-square error of only 0.0143 Ah with merely 5% labeled data, representing a 27.04% reduction compared with the best-performing semi supervised baseline (Co-training) under the same conditions.

Coordinated secondary frequency regulation by flywheel energy storage arrays (FESAs) and thermal power units (TPUs) mitigates grid frequency oscillations caused by renewable integration. However, existing strategies often overlook stage-specific dynamics and multiobjective tradeoffs. To address this, a stage-aware, multiobjective, bilevel dynamic load allocation strategy is proposed based on grid assessment criteria. At the upper level, fuzzy adaptive control predicts TPU capacity, while field-based traction experiments establish precise physical power constraints for the FESA. An improved nondominated sorting genetic algorithm II (NSGA-II) with a comprehensive satisfaction model and smooth weights is developed to optimize power output across distinct regulation stages. At the lower level, a bidirectional coupling interface enables adaptive group selection, followed by an improved consensus algorithm based on the relative rate of change of state of charge (SOC) for rapid energy and state alignment. Validated with operational data from a 660-MW supercritical plant in Ningxia, China, results demonstrate superior robustness against 0.6 s network delays and 0.5 s unit failures. Compared to representative benchmarks, it increases regulation net profit by 21.83% and reduces energy losses by 19.8%, effectively balancing grid requirements with the operational safety and economic sustainability of storage-integrated plants.

Accurate estimation of wind turbine fatigue load is crucial to ensure wind turbines’ safety. Since direct measurement is difficult, in this article, we propose a data-mechanism hybrid estimation method for wind turbine fatigue load. First, the fatigue mechanism of wind turbine generators is analyzed, and easily measurable parameters related to fatigue loads are inferred. Then, the time-series measurements of these parameters are organized into a feature set. A signal-related noise injection strategy is introduced, and a linear consistency method is employed to enhance the regression autoencoder, leading to robust dimensionality reduction of the features. Finally, a gradient-boosting CatBoost model is used to combine the coupling relationship between the obtained features and fatigue loads to estimate the fatigue loads. The experimental results show that the coefficients of determination (R2) for real-time estimation of fatigue loads in the tower and drive train reach 0.966 and 0.982, respectively, in out-of-sample tests, with errors both below 4%, demonstrating higher estimation accuracy and stronger generalization ability.

To improve the coordinated scheduling between the sending and receiving ends of large-scale renewable energy bases (LREBs) in desert regions, this paper proposes a day-ahead multi-objective optimization method. The approach considers both the power output characteristics of the base and the load demand at the receiving end. A bi-level optimization model is constructed, aiming to minimize the standard deviation of residual load at the receiving-end grid and maximize the utilization of renewable energy. The model jointly optimizes thermal unit commitment and power transmission profiles. Simulation results demonstrate that the proposed method achieves a better trade-off between economic performance and renewable integration. Compared with single-objective strategies, it effectively alleviates peak-shaving pressure at the receiving end and reduces renewable energy curtailment at the sending end. The findings confirm the effectiveness of the proposed method for day-ahead scheduling in high-penetration renewable systems and provide theoretical support for flexible transmission and coordinated dispatch in desert LREBs.

Structural load suppression of wind turbines is one of the important means to improve the economic efficiency of wind farm operation. However, wind turbines are complex systems coupled with multiple control variables. There is no comprehensive solution for the flexible and coordinated control of multiple controllable degrees of freedom. The main difficulty lies in the greater optimization burden of multi-degree-of-freedom systems. This paper proposes a variable degree-of-freedom control system for wind turbines, and uses multivariable model predictive control to design collaborative controllers for generator torque, pitch angle and yaw angle. To avoid the increase in complexity caused by the continuous intervention of yaw control, a yaw control cut-in/cut-out strategy is set, and the non-disturbance of switching is theoretically proved. Simulation results show that the proposed variable degree-of-freedom control system can effectively realize active power control and structural load suppression of wind turbines in a large wind speed range.

The safe, punctual, and reliable operation of high-speed railway (HSR) networks is crucial for ensuring system efficiency and enhancing passenger experience. However, due to the complexity of HSR systems and long operation routes, failures in system components can lead to unexpected incidents, causing deviations from the scheduled operations and, subsequently, delays. These delays, particularly large delays, significantly impact the overall performance and passenger experience of the network. Thus, accurate prediction of train delays, especially in large-delay scenarios, is essential for optimizing train scheduling and restoring normal operations in a timely manner. To address this challenge, this study proposes a deep learning architecture based on spatio-temporal feature extraction for accurate train delay prediction. A graph attention network-long short-term memory block is introduced to capture the spatio-temporal evolution features of different trains. In addition, a sequence forecasting approach and a mixture of experts module are integrated to model the complex relationships between the target train’s delay and its previous states. The delay evolution features are incorporated into the loss function, allowing for more accurate predictions. Experimental results show that the proposed model outperforms the baseline models, achieving at least an improvement of 50.62% and 27.89% in root-mean-squared error and mean absolute error, respectively. When the error tolerance is set within 3 min, the prediction accuracy reaches 96.68%. Experimental results demonstrate the superior performance of the proposed model.

Scheduling workflows in hybrid cloud environments presents significant challenges due to the inherent complexity of workflows and the dynamic nature of cloud resources. This complexity is further increased when attempting to balance workflow performance with privacy protection. Recent efforts have leveraged deep reinforcement learning (DRL) to address these challenges. However, most of these approaches rely on single-agent models, which can lead to security issues and scalability problems due to their centralized processing. Specifically, the properties of workflows are transferred to the single agent, which risks leaking privacy information. Our paper addresses these issues by introducing MARS, a real-time workflow scheduling method that prioritizes privacy protection in hybrid clouds. MARS leverages multi-agent deep reinforcement learning (MADRL) to optimize the workflow scheduling of cloud virtual machines (VMs). The benefit of our solution is that it relies on the collaborative learning of multi-agents on multiple VMs, which could assign user data to specific cloud servers for privacy protection while sharing training experiences between agents. In our implementation, MARS aims to reduce workflow completion time and operational costs while complying with strict privacy protection guidelines. The experimental results demonstrate that MARS can significantly surpass existing methods, reducing makespan by an average of $53.18 \%$ and costs by $61.98 \%$ compared to basic techniques, and achieving $20.26 \%$ and $25.71 \%$ improvements over the latest advanced methods, respectively.

The uncertainty of risk factors in infrastructure construction presents significant challenges to the management of construction progress. To effectively control construction progress while ensuring quality and reducing costs, we propose an adaptive scheduling method for infrastructure construction progress based on digital twin. First, we establish a digital twin model that reflects the construction progress by considering relevant progress information. Subsequently, taking into account the uncertainties associated with human factors, construction processes, and environmental factors, we develop a safety assessment method for power infrastructure and formulate a construction stoppage model. Then, we design an adaptive scheduling method that considers the uncertainties related to construction stoppage risks. Finally, the effectiveness and practicality of the proposed methods are validated through a comparative analysis of different cases.

Due to extreme external environments and changes in component health status, offshore wind turbines can generate out-of-distribution (OOD) data compared to normal operation. Traditional wind power forecasting (WPF) models are typically trained using historical data. Nevertheless, these models fail to maintain robustness with respect to OOD data. This paper introduces the matching based Transformer (MatTF) to mine shared knowledge from data across different distributions, thereby addressing the OOD problem. First, an algorithm for the detection of distribution boundaries is designed to partition the training data into subsets. Then, adaptive factors are added to the self-attention module in traditional transformer to solve the distribution matching problem. Finally, the effectiveness of the model is verified through case study comparing with other WPF models.

As a critical component of wind turbines, the structural integrity of the blades is crucial for safe operation. To enhance the performance of health monitoring and fault detection systems, an acoustic feature extraction method is proposed to designed to aid in the development of a digital twin model for wind turbine blades (WTB). Our approach begins with a time index algorithm based on local maxima, which isolates single-blade data from audio recordings of the wind turbine. Then a blade acoustic signal feature extraction technique is developed by combining improved Mel Frequency Cepstral Coefficients (MFCC) and Gammatone Frequency Cepstral Coefficients (GFCC). This technique leverages anomaly contribution and the maximal information coefficient (MIC) as comprehensive indicators to filter out irrelevant and negative features. By multiplying and fusing the features, our method achieves feature complementarity and reduces computational complexity. The proposed method can be seamlessly integrated into digital twins for wind turbines.

The oscillatory movement, aerodynamic damping effect, gyroscopic effect, and other phenomena in floating wind turbines result in distinct response characteristics of output and load under varying wind-wave coupling scenarios. Designing controllers within traditional control intervals using wind speed or wind-wave metrics is insufficient for fundamentally improving the wind turbine's performance. Firstly, according to the response distribution of the floating wind turbine, the response characteristics in different scenarios are extracted based on the autoencoder. And the response characteristics are clustered to form a control interval with response as the measurement index, which provides a basis for the partitioning in the construction of digital twin models for floating wind turbines. Secondly, within these reconstructed control intervals, control parameters are optimized based on the specific efficiency improvement and load reduction requirements of each scenario. This process leads to the development of a variable parameter controller. Finally, simulation experiments using FAST in various wind-wave coupling scenarios demonstrate that the proposed variable parameter controller effectively stabilizes power output and reduces loads.

In modern urban areas, inefficiency traffic management is one of the main causes of road congestion, leading to reduced fuel efficiency and increased traffic safety hazards. Traditional researches typically focus only on enhancing the throughput of intersections by optimizing traffic signals or individual vehicle trajectories. However, these methods often overlook the dynamic nature of the traffic system and the potential benefits of vehicle platooning, limiting their effectiveness in complex traffic environments. Addressing this challenge, this article presents MARP, a Cooperative Multiagent deep reinforcement learning (DRL) System for connected autonomous vehicle (CAV) Platooning. Utilizing vehicle to vehicle (V2I) and vehicles to infrastructure (V2V) technologies, MARP integrates sensing, computing, and communication to collect and process real-time data on traffic conditions, thereby achieving dynamic synchronization between traffic signal controllers and CAV platoons. By constructing platoons that collaborates with the infrastructure through a multiagent DRL collaboration model, MARP adapts to real-time traffic flow changes, significantly optimizing the fluidity and efficiency of the entire traffic network. Detailed experiments show that MARP effectively reduces traffic congestion, shortens intersection travel times, and cuts fuel consumption and emissions, surpassing the state-of-the-art approach.

The fast-growing development of floating wind turbines demands control systems capable of both reducing output power fluctuations and fault-tolerant function. In this paper, we propose an adaptive switched sliding mode controller to enhance the performance of floating wind turbine systems in the presence of environmental uncertainties and actuator faults. A control-oriented switched linear model for floating wind turbines is introduced, considering the average dwell time. Based on the proposed controller, a full-order state observer and an adaptive law compensate for the debilitated control outputs resulting from detecting errors, disturbances, and faults. The control parameters are derived by solving stability theorems, which are proved by the Lyapunov stability theory, linear matrix inequality technique, and average dwell time technique. The proposed model and controller are validated on the NREL 5MW wind turbine and spar-buoy platform using the high-fidelity fatigue, aerodynamics, structures, and turbulence (FAST) code. The performance of the proposed controller is compared with an optimal gain-scheduling proportional-integral controller under different wind-wave combined conditions. The results show that the proposed controller improves the power quality and attenuates mechanical loads of floating wind turbine under healthy and faulty conditions.Note to Practitioners—Floating wind turbines have drawn significant interest in renewable energy. When operating in the ocean far from shore, it is of great importance for floating wind turbines to have fault-tolerance capabilities for achieving stable power quality when faults occur in one or more components. A challenging problem is how to mitigate the fault effects on the wind turbine system. This paper proposes an adaptive fault-tolerant control strategy in the case of actuator fault occurrence in floating wind turbines.

This article explores aggregative games in a network of general linear systems subject to external disturbances. To deal with external disturbances, distributed strategy-updating rules based on the internal model are proposed for the case with perfect and imperfect information, respectively. Different from the existing algorithms based on gradient dynamics, by introducing the integral of the gradient of cost functions on the basis of the passivity theory, the rules are proposed to force the strategies of all agents to evolve to the Nash equilibrium, regardless of the effect of disturbances. The convergence of the two strategy-updating rules is analyzed via the Lyapunov stability theory, passivity theory, and singular perturbation theory. Simulations are performed to illustrate the effectiveness of the proposed methods.

This article focuses on a new outlier-resistant nonfragile control issue for a class of Takagi–Sugeno fuzzy delayed neural networks with reaction–diffusion terms. Compared with the existing delayed neural networks, fuzzy control rules and reaction–diffusion phenomenon are considered simultaneously, which makes the proposed models more practical. Furthermore, when subjected to abnormal interference, measurement outputs result in measurement outliers. In order to mitigate the negative effects on the estimation error, a state estimator scheme is presented by introducing a saturation function. By using an appropriate Lyapunov–Krasovskii functional and with the help of a free-weighting matrix, sufficient conditions can be deduced to guarantee the asymptotical stability and prescribed $\mathcal {H}_{\infty }$ performance index of the disturbance attenuation of the estimation error. Next, a design strategy of an outlier-resistant nonfragile state estimator is put forward by employing some decoupling techniques. An illustrative example is exploited to illustrate the validity and feasibility of the proposed state estimator. Finally, the obtained theoretical results are applied to image encryption. The experimental analysis demonstrates that the presented encryption scheme is feasible and effective.

To reduce the computational burden and resist the denial-of-service (DoS) attacks, a resilient distributed sampled-data control scheme is proposed for multiagent systems. The agent states are sampled periodically by the sensors. DoS attacks disrupt the data communication from transmitters to controllers randomly or periodically with a limited duration time. Information on DoS attacks can be obtained by introducing novel logic processors embedded in corresponding controllers. Next, the problem of resilient control can be converted into one concerned with the upper and lower bound of the sampling interval of an aperiodic sampled-data control system. Some sufficient criteria for developing resilient distributed controllers are derived using the novel looped Lyapunov functional approach and the free-matrix-based inequality method. Finally, two illustrative examples, unmanned aerial vehicles and the two-mass-spring systems, are provided to demonstrate the efficiency of the proposed resilient distributed sampled-data control protocols against the DoS attacks.

In this article, a class of continuous-time noncooperative games in networks of double-integrator agents is explored. The existing methods require that agents communicate with their neighbors in real time. In this article, we propose two discrete-time communication schemes based on the designed continuous-time strategy-updating rule for the efficient use of communication resources. First, the property of the designed continuous-time rule is analyzed to ensure that all agents’ strategies can reach the Nash equilibrium (NE). Then, we propose, respectively, periodic and event-triggered communication schemes for the discrete-time interactions among agents. The rule in the periodic case is implemented synchronously. The rule in the event-triggered case is executed asynchronously without Zeno behaviors. All agents in both cases can reach the NE asymptotically by interacting with neighbors at discrete times. Simulations are performed in networks of Cournot competition to illustrate the effectiveness of the proposed methods.

This article focuses on the event-triggered synchronization of delayed discrete-time chaotic neural networks with quantized effect and stochastic deception attack. First, for alleviating the network communication and communication burden, an event-triggered mechanism and a logarithmic quantizer are employed, separately. Second, for integrating the impact of event-triggered scheme, quantization, and cyberattack in a unified framework, a synchronization error model is introduced. Third, based on the Lyapunov–Krasvovskii functional (LKF), some sufficient conditions are established to guarantee the synchronization of drive system and response system. Furthermore, the co-design controller and homologous event-triggered parameters are also derived according to the presented asymptotic stability condition. Finally, the availability of the proposed method is verified by some numerical examples.

The emerging Internet of Things (IoT) services put forward higher demands on the bandwidth and latency performance of the traditional cloud computing. Cloud-edge collaboration and 5G cloud are two effective schemes to solve this problem. Adopting only one scheme may lead to the insufficiency of the performance in strict scenarios. So, the integration deployment of 5G networks and edge computing is significant. This article aims to provide a comprehensive survey on the complementarity and coexistence of 5G networks and edge computing. It reviews the evolving bonds of mobile communication technologies and edge computing and summarizes five scenarios where both 5G and edge computing can be applied. Based on the analysis of the key technologies and the tech-connections of 5G and edge computing, a win-win mode of them complementing each other is proposed. Finally, the potential challenges of the new mode, including security risks, heterogeneity issues, transition stage, and service orchestration, are summarized. This survey can provide rational assistance for researchers and developers focusing on the integration deployment of 5G networks and edge computing.

In this study, the problem of fault-tolerant sampled-data H∞ control for a networked control system with random time delays and actuator faults is investigated. Stochastic variables conforming with the Bernoulli distribution are considered to depict random time delays. A state feedback sampled-data controller is designed to ensure the asymptotical stability and H∞ performance of the resulting closed-loop system. By applying the Lyapunov–Krasovskii stability theory and the reciprocally convex combination lemma, a stability criterion for a random time-varying delay system is developed that guarantees the designed controller can satisfy the requirements of stability and maneuverability. The desired controller gain is then found based on the linear matrix inequalities. Finally, as a real application, a quarter-vehicle suspension system model is provided to demonstrate the benefits and validity of the proposed control law.

The irradiation forecasting technology is important for the effective utilization of solar power. Existing irradiation forecasting methods have achieved excellent performance with a massive amount of data in a centralized way. However, concerns about privacy protection and data security, which may arise in the process of data collection and transmission from distributed points to the centralized server, pose challenges to current forecasting methods. In this article, a novel federated probabilistic forecasting scheme of solar irradiation is proposed based on deep learning, variational Bayesian inference, and federated learning (FL). In this scheme, the training data are stored and computed in local Internet of Things devices, only forecasting models are shared. Two real-world datasets from SolarGIS and National Solar Radiation Database, and one benchmark dataset of Folsom are used to verify the feasibility and performance of the federated-based scheme. Comprehensive case studies are conducted to analyze the performance of the proposed scheme in multihorizon. And the effects of using meteorological features and variational Bayesian inference are evaluated. Compared with other state-of-the-art probabilistic centralized models, when data can be shared, the proposed scheme achieves competitive forecasting performance on the basis of data privacy protection. When data sharing is unavailable, due to the cooperative nature inherent (model-sharing) of FL, the performance advantage of the proposed scheme is more obvious.

This paper focuses on the problem of dissipative filtering for nonlinear interconnected systems with interval time-varying delays. The considered nonlinear interconnected system is modeled by Takagi-Sugeno fuzzy rules. By constructing the delay dependent Lyapunov-Krasovskii functional and using new integral inequality, the delay-dependent condition is established to ensure that the derived closed-loop system is asymptotically stable with strict (Q, S, R) - α- dissipativity. In addition, a suitable filter is designed by solving a set of linear matrix inequalities. The presented method can provide better performance than the existing ones for the case of H∞ filtering. A simulation example is given to demonstrate the validity of the developed filter design technique.

This paper focuses on the problem of exponential stability for a class of neural networks with time-varying delays. A more general inequality is established which extends the auxiliary function-based integral inequality. Based on the inequality and parameter-dependent matrix inequality, an improved delaydependent stability criterion is obtained by constructing an augmented Lyapunov functional. Three numerical examples are given to illustrate the efficiency of the method.