This paper addresses the challenge of rapid attitude stabilization for coaxial UAVs under arbitrary initial conditions following airdropped deployment. A control strategy that integrates positive invariant set (PIS) theory with sliding mode control is proposed. The proposed approach dynamically adjusts the control inputs according to the system’s initial state and fully exploits the favorable properties of the PIS, thereby ensuring that all state constraints are satisfied throughout the entire process. Flight test results demonstrate that the designed controller enables the UAV to recover a stable flight attitude from extreme initial angles within 300 ms. This study effectively resolves the challenges of robustness and rapid response during the transition from free fall to hovering flight after airdropped deployment.

Frequent plug-in/-out operations result in structural variations of DC microgrids (DCmGs), posing challenges to scalable control and often requiring costly redesigns to maintain stability. To address this issue, this paper proposes a scalable voltage control strategy for uncertain DCmGs, enabling plug-and-play functionality without controller redesign or system reconfiguration. A polytopic uncertain DCmG model is first formulated to simultaneously capture parameter uncertainties in distributed generation units (DGUs), power lines, and ZIP (i.e., impedance, current, and power) loads. A structured free-weight matrix technique is then developed to mitigate the adverse effects of line and load uncertainties on DGUs while yielding a more tractable linear matrix inequality formulation. The proposed scalable control method is implemented locally to ensure the dissipative voltage stability of each DGU, thereby preserving the dissipativity of the entire network. Numerical simulations validate the effectiveness of the proposed strategy in achieving faster convergence and reduced overshoot.

To address the challenges of difficult PID parameter tuning, model mismatch, and poor control performance in systems with large time delays for industrial temperature control, this paper proposes an Improved Hippopotamus Optimization based Fuzzy Internal Model Control PID method. First, an FIMC-PID controller is constructed, integrating the advantages of fuzzy control's low model dependency and strong disturbance rejection with the high control precision of Internal Model Control PID. Second, to solve the FIMC-PID parameter tuning problem, the Hippopotamus Optimization Algorithm is improved and the local search mechanism of the Banyan Growth Optimizeris incorporated to enhance parameter refinement, forming the IHO algorithm. Finally, MATLAB simulation experiments demonstrate that, compared to GA-PID and BTGO-IMC PID controllers, the proposed IHO-FIMC PID exhibits significantly smaller overshoot, smoother responses, and stronger robustness under both step response and model mismatch conditions, validating its potential for engineering applications.

When the measurement or intermediate exchange data are tampered by false data injection attacks (FDIAs) in cyber-physical power systems (CPPSs), their authenticity (i.e., credibility) will be destroyed, leading to the failure of distributed state estimation. To address the problem, this paper proposes a novel distributed secure state estimation method using double-layer detection for FDIAs. First, the first layer uses a chi(2)-based attack detector to check whether the measurement data are tampered and, if so, replaces the contaminated by the predicted data based on the Kalman method to improve data credibility and the accuracy of local state estimation. Then, these credible data (i.e., the checked/replaced data) are interacted with the neighboring subregions, when event-triggered mechanism is satisfied. Second, the second layer is a dynamic watermarking-based active attack detector designed to check whether the intermediate exchange data are tampered with and, if so, to replace the attacked intermediate exchange data by Kalman method. It is also found that when the Signal to Interference plus Noise Ratio (SINR) is greater than 10db, attack detection performance is positively correlated with the watermarking intensity. Third, according to the double-layer detection results, each subregion can be divided into secure set and non-secure set, which are used to reconstruct distributed state estimation model. The corresponding novel distributed secure state estimation method is designed and theoretically proved to be exponential convergence in the mean square. Finally, experimental results demonstrate the feasibility and effectiveness of the proposed double-layer attack detection and distributed secure state estimation method for CPPSs.

Aviation electrification has accelerated the adoption of onboard dc microgrids (DCMGs) in more electric aircraft (MEA). The dynamic integration and disconnection of renewable energy sources (RESs) are crucial for resource optimization and environmental adaptation but can induce voltage instability. This article proposes a plug-n-play voltage control strategy for MEAs operating under constant power loads (CPLs), ensuring voltage stability while enabling seamless RES connection and disconnection. A unified onboard DCMG model is formulated, capturing both the nonlinear behavior of CPLs and the interconnection couplings among RESs. Leveraging a structured storage function technique, sufficient conditions are derived to guarantee dissipative voltage stability at local and global levels. The proposed fully decentralized control scheme operates independently of both the number of RESs and their interconnection topology, thereby significantly enhancing the scalability for smooth plug-in/-out operations. Extensive simulations conducted across various scenarios convincingly validate the effectiveness of the proposed control strategy.

With the continuous advancement of informatization and automation in industrial control systems, the coupling between the information and physical layers has intensified, resulting in increasingly complex cross-domain attack propagation. Attackers can infiltrate from enterprise networks into the SCADA layer and further into physical devices through multi-hop strategies, thereby posing serious security threats. To address the limitations of single-layer modeling in existing methods, this paper proposes a cross-domain attack path prediction and security assessment approach based on dual reinforcement learning. The method leverages a multi-layer attack graph generated by MulVAL, which integrates cyber-physical features, and introduces a dual reward mechanism combining CVSS scores and graph centrality to guide path optimization. Experiments demonstrate its robustness and effectiveness, providing an effective and practical framework for identifying high-risk paths and assessing ICPS security.

The rapid growth of renewable energy integration and electric mobility has increased the demand for safe and reliable lithium-ion batteries, which are essential due to their high energy density, long lifespan, and efficiency. However, complex internal electrochemical reactions and external operational stress can induce minor short circuits (MSC) that are difficult to detect at early stages yet may escalate to thermal runaway, posing significant risks to large-scale energy storage systems. To address this challenge, this study proposes an unsupervised MSC fault diagnosis framework that integrates a hybrid feature extraction strategy with a deep support vector data description algorithm. The method employs two-dimensional correlation coefficients and two-dimensional wavelet transform to capture voltage consistency across cells and detect transient anomalies associated with fault development. These complementary features are fused into a multidimensional representation and processed by the deep model, which learns compact patterns of normal operating states and constructs a hypersphere for anomaly detection. The framework is validated using a laboratory module with six battery cells, demonstrating effective fault identification under varying operating conditions, fault severities, and battery chemistries, achieving a 94 % fault detection rate with a 3 % false alarm rate. Furthermore, the computational procedure relies on matrix-based feature construction and a lightweight feed-forward inference process, offering computational efficiency suitable for real-time deployment in battery management systems. Benefiting from its unsupervised and data-driven design, the framework exhibits strong generalizability under diverse conditions and provides a promising pathway for enhancing the safety and reliability of future energy storage applications.

This paper proposes a physics-informed data-driven modeling framework for coaxial helicopter dynamics, based on deep neural networks (DNNs) and the Koopman operator. First, a structured Koopman formulation is developed to ensure physical interpretability and causal consistency. Specifically, the proposed method incorporates kinematic and dynamic topological constraints into the construction of observables, while simultaneously incorporating physics-based regularization terms during network training. Second, to enhance modeling efficiency, a hyperparameter optimization strategy based on differential evolution hyperband (DEHB) is developed. This strategy automates the observable construction process by optimizing a joint objective that accounts for both accuracy and controllability. Furthermore, an adaptive horizon model predictive controller (AHMPC) is designed to reduce computational complexity by adaptively shortening the prediction horizon when the terminal-set condition is satisfied. Experimental results demonstrate that the proposed PIKoopman method achieves higher modeling accuracy under noisy conditions, and its learned model, with the aid of an adaptive prediction horizon mechanism, effectively reduces the online computational burden.

This paper investigates the voltage regulation problem for distributed generation units (DGUs) in DC microgrids subject to renewable energy fluctuations and unknown nonlinear ZIP loads. To address the complexity of system dynamics, the DGU model is first transformed into a fully actuated system (FAS) representation, which serves as the foundation for the subsequent controller design. Based on this framework, an adaptive neural network voltage regulation control strategy is proposed. Radial basis function neural networks (RBFNNs) are utilized to approximate the unknown nonlinear dynamics induced by ZIP loads, while specific adaptive laws are synthesized to estimate the unknown control gain and approximation error bounds online. Rigorous theoretical analysis based on Lyapunov stability theory guarantees that all signals in the closed-loop system remain bounded and that the output voltage tracks the reference trajectory asymptotically. Finally, numerical simulations are conducted to validate the effectiveness and robustness of the proposed control scheme under time-varying source and load conditions.

This paper presents an integrated solution for 5G and Time-Sensitive Networking (TSN), incorporating a protocol conversion mechanism, a Hybrid Stochastic Timed Fuzzy Petri Net (HSTF-PN) model, and a customized TSN node time synchronization scheme. First, a reverse protocol mapping-based conversion mechanism is introduced to address interconnection challenges between 5G and TSN protocols. Second, the HSTF-PN model, which integrates stochastic delays and fuzzy logic, is proposed to dynamically optimize network resource allocation. Third, a customized TSN node time synchronization scheme, implemented with FPGA hardware acceleration, enables high-speed data processing for precise clock alignment. Finally, the proposed 5G-TSN integration solution is validated on an FPGA platform, demonstrating its effectiveness in ensuring clock consistency, minimizing delays, and enhancing data flow efficiency.

Micro-expressions carry abundant emotional and cognitive cues, acting as significant external signals of an individual’s psychological condition. Careful observation and accurate recognition of facial micro-expressions in elderly individuals can offer essential insights for screening, tracking disease progression, and assessing the effectiveness of interventions for mild cognitive impairment. However, due to their low amplitude and predominantly localized nature, existing mainstream expression recognition algorithms often struggle to effectively extract these subtle features, leading to low recognition accuracy. To tackle this issue, this paper proposes a micro-expression recognition method based on an adaptive dynamic weak-texture amplification strategy. Given that micro-expressions exhibit small movement amplitudes, the method incorporates a dynamic weak-texture amplification technique to enhance subtle motion cues. An adaptive optimization approach with feedback regulation is employed to dynamically adjust the amplification factor. A multi-scale feature module with channel–space enhancement is then introduced to extract features from the amplified images. The channel module emphasizes inter-channel dependencies, while the spatial module captures spatial positional cues. Finally, the extracted channel and spatial features are fused to classify micro-expressions. The proposed algorithm demonstrates strong performance in both the extraction and recognition of facial micro-expression features. Our findings establish an important groundwork for investigating how micro-expressions relate to the onset of dementia, which is essential for timely detection and intervention against cognitive deterioration. The code is publicly available on GitHub: https://github.com/cxhwxhn/AA-DWT-MER

To address the collaborative optimization problem of computation offloading and resource reconfiguration in heterogeneous Industrial Internet of Things (IIoT) edge computing, this paper proposes a dynamic offloading reconstruction method integrating data encryption and load balancing. A global multiobjective coupling model is established, incorporating task latency, terminal energy consumption, encryption overhead, and load balancing degree to comprehensively characterize the collaborative relationship among multi-dimensional optimization objectives. Furthermore, a hybrid optimization algorithm based on the Decomposition-Coordination principle is designed, combining Karush-Kuhn-Tucker (KKT) conditions with Quantum-behaved Particle Swarm Optimization (QPSO). This approach efficiently solves Mixed-Integer Nonlinear Programming (MINLP) problems and enables adaptive reconfiguration of system resources. Experimental results demonstrate that compared with traditional strategies, the proposed method achieves a significantly lower total system cost. Meanwhile, it ensures data transmission security through the built-in encryption/decryption process of the model and maintains edge server loads within a balanced range, fully validating its global optimization performance in complex industrial scenarios.

This paper addresses the challenge of resilient secondary frequency control in distributed renewable energy microgrids under False Data Injection Attacks. A rigorous Lyapunov based stability analysis is conducted, and a formal theorem is established that quantifies the maximum tolerable attack intensity under both actuator and sensor side intrusions. This theoretical tolerance boundary provides a principled and interpretable threshold for real time attack detection and secure control activation. Building upon this result, we propose a resilience framework that tightly integrates the derived physical safety boundary with an adaptive reinforcement learning controller. Specifically, a Double Replay Q-learning algorithm is developed, leveraging dual Q-tables and prioritized experience replay to enhance convergence speed and robustness under nonlinear REM dynamics. Simulation results conducted on both a real world experimental REM platform and a modified IEEE 9-Node test system validate the effectiveness and scalability of the proposed approach.

Dear Editor, The crafted generalized replay attacks(GRAs)can bypass tradi-tional multiplicative watermarking(MW)-based active detection in networked control systems(NCSs),which will seriously destroy sys-tem stability or even crash.To address this problem,this letter pro-poses a novel secure control method by using MW-based detection and data compensation.First,the limitation of traditional MW-based detection method is analysed,and a novel MW-based active detec-tion scheme is proposed by adding an irreversible watermarking detection unit.Then,according to the detection result,an online data compensation scheme based on cubic spline interpolation algorithm is provided,and the maximum allowed attack rate of GRAs is given to maintain the exponential stability of NCSs.Finally,experimental results confirm the effectiveness of the proposed method.

This paper proposes a decentralized fuzzy adaptive control strategy based on the fully-actuated system (FAS) approach to address vertical–torsional coupled vibration (VTCV) and inter-stand tension coupling effects in continuous rolling mill trains (CRMT) during high-speed rolling. First, a mixed-order fully-actuated interconnected system (MOFAIS) model is developed that integrates the vertical–torsional dynamics of individual stands with inter-stand tension transmission, thereby incorporating nonlinear vibrations, tension coupling, and thickness–speed–tension (TST) chain interactions into a unified modeling framework. Next, a decentralized fuzzy adaptive controller is designed using a mixed-order fully-actuated system approach (MOFASA) to achieve effective VTCV suppression and coordinated inter-stand tension stabilization. A speed-error constraint mechanism is embedded to ensure roll speed matching across multiple stands and to maintain strip surface flatness. Lyapunov stability analysis rigorously proves that the principal tracking errors asymptotically converge to zero, while all adaptive parameter estimates and control inputs remain uniformly bounded, thereby ensuring that vibrations are suppressed and tension coupling effects are accurately compensated. Finally, a simulation model of a 1450$\, \text{mm}$ rolling mill is built on the MATLAB/Simulink platform incorporating comparative simulations to verify the robustness and engineering applicability of the proposed strategy under complex operating conditions.

The main problem addressed in this paper is the controller design under composite switching in large-scale interconnected systems. Composite switching systems are widely present in large-scale cyber-physical systems, such as microgrids and industrial process control systems, motivating the need for scalable and efficient control methods. To this end, this paper develops an integrated model for a large-scale system formed by interconnecting individual subsystems, each of which is a composite switched system, and investigates the scalable control problem for such an interconnected composite switched system (ICSS), a unified switching signal governs all subsystems, where the signal is generated through a logical function driven by random variable inputs. First, the semi-tensor product (STP) technique, combined with the dimension expansion method, is used to compress the composite switching signal, treating it as part of the state and cascading it with the states of each subsystem. This results in a new system state and an expanded interconnected system model described in the form of a linear time-invariant (LTI) system. The key contributions of this paper include the establishment of an integrated model that captures the interconnection and composite switching behavior of the large-scale system, as well as the development of a scalable distributed state feedback control algorithm that leverages this unified model. Based on this, the LTI model of the interconnected large system under distributed state feedback control is provided. Next, the necessary and sufficient conditions for the mean-square stability of this large system are given, and the scalability and design method of the distributed state feedback strategy are implemented based on a recursive algorithm. Finally, quantitative simulation results based on a DC microgrid example demonstrate that system state trajectories decay to zero under allowable composite switching, confirming the theoretical mean-square stability and demonstrating the practical feasibility of the control framework.

Aircraft fuel consumption modeling is essential for enhancing aviation energy efficiency and security, especially within the global context of achieving carbon neutrality. This paper proposes a novel Local-Segment Dual Attention (LSDA) Network for dynamic modeling and prediction of aircraft fuel consumption. Specifically, we employ adaptive time-series decomposition to extract trend and seasonal components, followed by a lightweight dual-path attention mechanism that captures both local condition dependencies and global temporal information simultaneously. Experiments conducted on extensive flight operation data demonstrate that the LSDA Network outperforms seven state-of-the-art methods. The proposed model achieves superior predictive performance with lower mean absolute error (MAE), mean square error (MSE) and Mean Absolute Percentage Error (MAPE). The results validate the potential of LSDA for precise and reliable fuel consumption forecasting, contributing to sustainable and optimized aviation operations.