In this article, the security issue of remote state estimation in cyber-physical systems (CPSs) for multihop networks is investigated. A smart sensor transmits its local estimate to a remote estimator through some relay nodes to save limited energy while facing the threat of a malicious Denial-of-Service (DoS) attacker. Due to the energy constraints, the defender allocates transmitting energy on each channel to minimize the estimation error, while the attacker selects the attacking energy implemented on each channel to maximize the degradation of the estimation performance. Supposing that the attacker cannot obtain the acknowledgment, and considering the interaction between the defender and the attacker, a dynamic game with asymmetric information is formulated. To overcome the computational difficulty, this game is transformed into a belief-based stochastic game (BBSG) with symmetric information such that the Nash equilibrium of the original game can be determined by solving the BBSG. Furthermore, a multiagent Q-learning algorithm is proposed to obtain the optimal strategy. A numerical example is finally provided to demonstrate the effectiveness of the results.

Jamming attacks pose a significant threat to the security of air-ground communications, where the challenge becomes more severe when involving multiple unmanned aerial vehicles (UAVs) incurring complex interference. To address this issue, this paper proposes a graph attention-based reinforcement learning strategy for anti-jamming UAV communications. Specifically, we consider the multi-UAV transmission and deployment in the presence of jamming attacks. Then, we formulate a zero-sum game with the legitimate side and adversary to maximize and minimize the overall transmission rate, respectively. Given the complicated structure of the game, we decompose it into two layers, tackled in a hierarchical learning framework. Particularly, the inner layer addresses the legitimate beamforming, for which we establish the graph attention network (GAT) to track the complicated interference and jamming relationship based on the graph representation of the UAV network. The outer layer address the legitimate UAV deployment and adversarial jamming policy, which is reinterpreted in a multi-agent deep reinforcement learning framework to obtain the strategies of both sides. The inner GAT is then nested within the outer multi-agent learning framework in a hierarchical manner to approximate the equilibrium of the original game model. Simulation results demonstrate the convergence and the performance superiority of the proposed learning scheme in terms of anti-jamming transmission rate. Also, the results exhibit significant generalization capability to cover different network configurations and parameters with reliable communication performance.

Information disclosure and cyber attacks pose significant challenges to practical implementation of distributed control in microgrids (MGs). This paper investigates the distributed privacy-preserving resilient secondary control problem for an isolated AC MG under hybrid attacks including false data injection (FDI) and denial-of-service (DoS) attacks at low communication costs. Specifically, using state decomposition strategy and event-triggered mechanism, a distributed privacy-preserving event-triggered resilient secondary controller is designed, which reduces communication while safeguarding system privacy. Additionally, a long short-term memory (LSTM)-based estimator is designed to estimate the aggregated communication signals from neighboring DGs. The proposed resilient control method can further mitigate the impact of cyber attacks by using the estimated aggregated communication signals as the reference value, and the stability of the MG system is proven by Lyapunov theory. Compared to existing data-driven methods for attack detection, the proposed LSTM-based estimator uses communication signals as the database, thus avoiding the leakage of complete system data during offline training. To validate the effectiveness of the proposed method, a hardware-in-the-loop experiment is conducted under various hybrid attacks in OPAL-RT.

For heterogeneous multi-agent systems (MASs), unknown follower dynamics and unmeasurable noise disturbances are common challenges. Such systems are found in many Internet of Things (IoT) applications, including intelligent transportation, smart manufacturing, and environmental monitoring, where reliable formation control is essential. This paper proposes a new data-driven leader-follower output formation control scheme to address these issues.A distributed adaptive observer is designed using only collected data to estimate both the leader’s system matrix and its state information simultaneously, thereby eliminating the need for centralized communication. For the followers, an internally stabilizing controller is directly constructed from noisy input-state data via the Slater condition, with the noise characterized by new nonconservative cross-covariance bounds. This completely bypasses explicit identification of the follower system matrices. A data-driven adaptive regulation equation is subsequently constructed from the collected data. To mitigate the computational burden induced by large matrix dimensions when abundant data are available, the Lagrange multiplier method with adaptive step size is adopted to compute the controller gain efficiently, enabling scalability to high-dimensional systems. Using this gain, a leader-follower controller is developed to achieve the desired output formation. Finally, simulation studies on nonholonomic mobile robots validate the feasibility and effectiveness of the proposed distributed control method. Comparative results demonstrate the superiority of using cross-covariance bounds for noise characterization in controller design.

This article focuses on the challenging problems for robust load frequency control (RLFC) in multi-area power systems considering unknown parameter uncertainties in both load frequency control (LFC) operation conditions of nominal power systems and coupling inputs under dynamically changing reconfigurable communication networks by developing an autonomous gain scheduling scheme. We consider a class of coupled smart grids (e.g., multi-energy coupling microgrids as a modern multi-area multi-source power system that can realize multi-energy complementarity and comprehensive utilization improving energy efficiency) in which the process dynamics are composed of time-varying vector functions of scalar combinations of the states and dynamically changing networks. By incorporating the impact of generation-rate constraints (GRC), we propose a performance estimation index to evaluate and determine the optimal configuration of communication networks in multi-area power systems. In the event of link failures due to interference or when partial limits exceed predefined security redundancy, the automatic decision function is activated, scheduling reconfigurable communication modes and adjusting controller gains accordingly. Based on this, we first develop an RLFC strategy via a distributed framework that is driven by a dynamically reconfigurable communication network and a time-varying switching scheme. The proposed strategy can handle the system coupling dynamics to achieve global exponential stability for multi-area power systems with multiple aggregated uncertainties and GRC limiters. Furthermore, an RLFC strategy via an exponentially distributed adaptive framework is proposed to schedule the control gains and manage time-varying adaptive coupling dynamics for the multi-area power systems with multiple GRC limiters and aggregated system uncertainties. These uncertainties consist of aggregated parameter uncertainties, coexisting matched, and mismatched parameter uncertainties that can be unknown, where the continuous excitation conditions are equivalent to matrix inequality conditions to ensure exponential stability at the origin. The effectiveness of the proposed strategies is verified via a three-area power system with dynamically changing configurable communication networks.

In this paper, we consider the resilient fully distributed control problem for multi-agent systems (MASs) under denial-of-service (DoS) attacks. In contrast to the existing resilient control results on DoS attacks, our method is developed without using any information about the Laplacian matrix through a two-step design procedure. Firstly, a couple of weight estimators are designed, which consist of an impulsive system to update the state of the estimators and an event-based algorithm to generate the jump set in the impulsive system. Then, a switching-based fully distributed consensus protocol, consisting of a locally prescribed time controller to regulate the state of each agent to zero at the state reset interval and a fully distributed controller to achieve the consensus objective at the consensus interval, is proposed. It is shown that there is only a finite number of elements in the jump set and the consensus errors converge exponentially. Secondly, the developed method is extended to solve the resilient fully distributed consensus problem for MASs with DoS attacks. Finally, a simulation example is presented to show the effectiveness.

This work presents an event-triggered (ET) output feedback Lyapunov-based distributed model predictive control (DMPC) approach for large-scale nonlinear systems. In practical applications, incomplete state information, process disturbances, and measurement noise may degrade control performance. To address these issues, a distributed extended Kalman filter (DEKF) estimator is designed to reconstruct the system states for output feedback controller design. The convergence of the designed DEKF estimator is theoretically established. Based on the estimated states, an output feedback Lyapunov-based DMPC algorithm is developed by explicitly considering the influences of the coupling subsystems to reduce the scale of the control problem. Furthermore, an event-triggering condition is derived to reduce unnecessary online optimization and communication, forming a DEKF-based ET-DMPC framework. The recursive feasibility of the proposed ET-DMPC and the stability of the closed-loop system are rigorously proved. Finally, the proposed DEKF-based ET-DMPC algorithm is applied to a nonlinear continuous stirred-tank reactor (CSTR) system. The simulation results demonstrate that the proposed method reduces the computational burden while maintaining satisfactory control performance.

The assisted driving system is a key strategy for promoting the growth of science and technology since it improves the driving experience, ensures stable vehicle operation, and protects the lives of drivers. Object detection algorithms, which are the basic technology of the assisted driving system, significantly influence its stability and sensitivity. By fusing visible and infrared pictures, multispectral object detection (MOD) methods have been suggested to improve detection accuracy. Nonetheless, the current approaches for feature-level fusion detection exhibit low detection efficiency. To solve this issue, we present YOLO-MSLite, a lightweight multispectral object recognition technique based on feature-channel-wise knowledge distillation. The technique improves the Conv and C3 modules of the YOLOv5 backbone by introducing group convolution, which decreases the number of parameters while allowing the one-stream network to interact with features. To increase the information selection capabilities of YOLO-MSLite, two-stream, and one-stream models are employed as the teacher and student models, respectively. Experiment findings on several datasets show that YOLO-MSLite achieves the same degree of accuracy as existing state-of-the-art approaches while being lighter in structure and more efficient in detection. The validation findings of the algorithm installed on an embedded platform further reveal that the model gets good detection results and can reach the level of real-time detection.

This research focuses on gain-scheduling control for minecart active suspension systems (ASSs). Addressing practical scenarios in complex mining environments where transition probability information may be imprecisely accessible or completely unavailable, we establish a stochastic nonlinear suspension model featuring incomplete transition probability matrix (TPM) information. To ensure reliable controller design under such demanding conditions, a novel biharmonic polynomial framework is proposed. At the structural level, this framework reconstructs partially unknown TPMs into weighted convex combinations of precisely known TPMs using polytopic techniques. Simultaneously considering the challenges of suspension state acquisition and potential packet losses, an observer-assisted polynomial gain-scheduling controller is developed. By further integrating homogeneous polynomial techniques into both controller and observer synthesis, this framework achieves significantly expanded feasible solution domains. Hardware-in-the-loop (HIL) validation demonstrates 64.2% conservatism reduction in control design constraints and 18.2% root mean square (rms) reduction in body acceleration, confirming superior design generality and enhanced ride comfort performance.

This article concentrates on the attack detection and active attack defense strategies for discrete-time linear cyber-physical systems (CPSs) with unknown but bounded (UBB) disturbance and noise in the presence of both actuator and sensor attacks. First, a novel zonotopic observer is constructed to estimate the set-valued state and actuator attack by introducing augmentation techniques. To mitigate the effects of uncertainty and enhance estimation accuracy, the $H_{\infty }$ technique is introduced to construct the observer. Unlike most existing works, the constructed observer simultaneously estimates the system state and actuator attacks. Then, by combining the designed observer with reachability analysis, a set-valued abnormal detector and a residual-based abnormal detector are designed to detect actuator and sensor attacks, respectively. In addition, by incorporating the obtained state reachable sets and the $H_{\infty }$ technique, an active attack defense mechanism is designed to mitigate the impact of attacks on system performance. The proposed defense strategy does not introduce any performance loss in the absence of attacks. Finally, the superiority of the developed method is demonstrated by its application to a numerical simulation and an autonomous aircraft system.

In this paper, we are interested in how to achieve consensus among normal agents for a class of multi-agent systems (MASs) with malicious agents, in which the states of agents are determined by minimizing their cost functions. We first equip each agent with two states: desired state and behavior state, in which the behavior state is used to interact with the neighbors. Then, using the multi-round information filtering technique, we design a security fault-tolerant algorithm, and develop a sufficient topological condition that can ensure the consensus of the desired states among the normal agents. Finally, we provide a numerical example to demonstrate the effectiveness of the theoretical results.

This paper focuses on the multi-agent synchronization problem with an open-loop unstable leader and followers under the switching topologies. For this issue, the typical approach is intermittent communication (including a spanning tree intermittently) or fast switching strategy. We here consider a more general scenario where each communication link between two agents is randomly disconnected and reconnected, and the durations of both the disconnected intervals and the connected intervals follow negative exponential distributions. To handle this issue, we propose a network transformation mapping method that divides the communication network into reachable and unreachable parts at any time. A node can access the leader's information and synchronize only when it lies in the reachable part; otherwise, it cannot. For each node, the synchronization speed is designed such that its convergence amplitude in the reachable part exceeds its divergence amplitude in the unreachable part. Hence, all nodes achieve asymptotic synchronization over the entire time domain. We further develop adaptive strategies for coupling gains to reduce the computational complexity introduced by the network transformation mapping. The proposed method is also applicable to jointly connected switching topologies and distributed observers under topologies that lack a spanning tree at any time. Finally, two numerical simulations – synchronous control of multi-one-link manipulator systems and multi-motor systems based on Fieldbus – are provided to demonstrate the effectiveness of our approach.

This article focuses on the challenging problems for robust load frequency control (RLFC) in multi-area power systems considering unknown parameter uncertainties in both LFC operation conditions of nominal power systems and coupling inputs under reconfigurable communication networks. A class of coupled smart grids is modeled in which the process dynamics are composed of time-varying vector functions of scalar combinations of the states and communication networks that are not interconnected. In part I, the networks can be any connected static configuration but not dynamically changing network configurations. To overcome these issues, a distributed RLFC strategy is developed that uses a configurable communication network. This strategy can handle the system coupling dynamics and obtain global asymptotic stability for multi-area power systems with multiple aggregated uncertainties. Furthermore, a distributed adaptive RLFC protocol is proposed to schedule the control gains and time-varying adaptive coupling dynamics for the multi-area power systems with aggregated system uncertainties consisting of aggregated parameter uncertainties, coexisting matched, and mismatched parameter uncertainties that can be unknown. The proposed strategies are verified through the modeling of a three-area power system with configurable communication networks. The results show that the proposed methods are efficient. Specifically, for the case of unknown coexisting matched and mismatched parameter uncertainties, the LFC of closed-loop power systems is asymptotic stability without relying on any bounded-state set.

The dual processing paradigm is a powerful design structure for an autonomous vehicle, in which the driver can take over the vehicle control from an autonomous driving mode to a personal driving mode when the driver thinks the current environment is too complicated for autonomous driving. However, the perception of the motion of a preceding vehicle often affects a follower driver’s decision, which inevitably affects the cruising speed of a concerned platoon. With the advancement of information and communication technologies (ICT), the ability of drivers to perceive information can be significantly improved. This paper addresses the platoon control problem for connected automated vehicles (CAVs) with a car-following model and a reaction-time delay under different inter-vehicle communication environments (such as the vehicle-to-vehicle (V2V) mode with pairwise communications and vehicle-to-everything (V2X) mode with simultaneous broadcast). By modeling the reaction-time delay, a nonlinear integrated parameter model is introduced, in which the vehicles’ behavior is assumed to be affected by the “optimal velocity model” (OVM) and connected environments. To capture our daily experience that a driver typically immediately follows the speed change of the preceding vehicle, a combination of the integrated parameter model and velocity difference models is presented, and the string stability of traffic flow and the platoon dynamics for CAVs with a car-following model and the reaction-time delay under V2V and V2X modes are developed and analyzed, respectively. In addition, a delayed feedback design for platoon tracking is developed to improve the tolerable upper bound of reaction-time delay. Our analysis and simulation results show that when the leader vehicle slows down suddenly, (1) the tolerable upper bound of reaction-time delay of the CAVs in the V2X mode is significantly greater than that in the V2V mode; (2) increasing a small reaction-time delay may improve the traveling speed of some rear follower vehicles; and (3) the delayed feedback control can provide additional benefits in terms of traveling speed adjustment for the last few follower vehicles.

Information disclosure and cyber attacks present significant challenges to the practical implementation of distributed control in microgrids (MGs). Existing works have not been well-equipped to address these two challenges simultaneously, primarily because privacy protection mechanisms increase the complexity of system modeling, which in turn makes attack detection more difficult. This article investigates the distributed privacy-preserving resilient secondary control problem for hybrid ac/dc MGs under hybrid false data injection (FDI) and denial-of-service (DoS) attacks. Specifically, a deep neural network (DNN)-based estimator is first designed to estimate the aggregated remote signals from neighboring nodes by using real-time local signals (e.g., active/reactive power) as input. Subsequently, based on the estimation of aggregated remote signals and data encryption strategy, a privacy protection resilient secondary controller is designed to mitigate the impact of cyber attacks while safeguarding the system privacy simultaneously. Finally, the effectiveness of the proposed method under hybrid attacks is confirmed through a real-time experiment in OPAL-RT.

In this article, the resilient distributed cooperative optimization problem is investigated for cyber-physical vehicle systems (CPVSs) in the presence of denial-of-service (DoS) attacks. Unlike existing results on distributed cooperative optimization for CPVSs, a novel layered design approach is introduced, which consists of a resilient distributed optimization algorithm, low-pass filters, and adaptive fuzzy controller. In particular, a novel resilient distributed optimization algorithm is first designed to ensure convergence to the optimal solution in the presence of DoS attacks. Then, a novel third-differentiable variable is generated through the low-pass filters, which ensures the existence of the third-order time derivatives of the signal even under DoS attacks. In order to deal with situations where the states of vehicles involve both dynamic uncertainties and unknown nonlinearities, an adaptive controller is proposed based on the adaptive technique that enables the states of the vehicles to converge toward the optimal trajectory. A numerical simulation is finally presented to demonstrate the practical feasibility and performance of the proposed approach.

This paper proposes a method that combines phase-space reconstruction and bidirectional temporal convolution networks to accurately predict the degree of ash fouling on the heating surface of the boiler reheater. First, the phase-space reconstruction method maps the original one-dimensional chaotic time series into a high-dimensional phase space to analyze its intrinsic nonlinear dynamics. Then, the bidirectional temporal convolution network uses the reconstructed sequence for time series prediction. Finally, the prediction results are evaluated by evaluation indicators such as the root mean square error. The results show that the PSR-BiTCN model not only improves prediction accuracy by 14.3668% compared to the traditional single neural network model; but also reduces prediction error by 6.18226%. While verifying the rationality of the model, it also lays a theoretical foundation for the subsequent transition from time-based soot-blowing to state-based soot-blowing models.

This study focuses on mission planning for repairing multiple damaged spacecraft in geosynchronous orbit (GEO). A mixed repair strategy is proposed, including using a single service spacecraft (SSc), simultaneous repair by two SScs, and repair within a space station (SS). We employ hybrid propulsion to reduce fuel consumption. The optimization objective is to minimize mission duration, aiming for the fastest possible repair to restore the damaged spacecraft's functionality. The SSc-SS-target spacecraft repair mission planning model is established, considering practical constraints such as spacecraft maneuverability, fuel capacity, service sequence conflict, etc. A two-level solution framework is presented: the upper level applies a genetic algorithm (GA) to solve for the allocation method, service sequence, and repair strategy, with enhanced optimization efficiency through a local search (LS) strategy. In addition, graph theory is applied to resolve sequence conflicts and ensure the strategy's feasibility. The lower level uses variable neighborhood search (VNS) to optimize the maneuver trajectory. Finally, representative simulations and comparative examples validate the proposed solution framework.

This article proposes a fast multispacecraft rendezvous sequence mission planning algorithm based on neural combinatorial optimization and deep reinforcement learning (DRL). The algorithm can address rendezvous requirements for various geostationary Earth orbit (GEO) spacecraft on-orbit servicing missions, such as on-orbit maintenance and on-orbit refueling. In order to meet the fast decision-making needs of multispacecraft rendezvous missions and ensure the successful completion of the missions, we take advantage of the fast calculation speed of the DRL algorithm and make some unique designs to address the challenges of missions, and finally propose the data augmented REINFORCE algorithm (DARA). The algorithm can rapidly plan the optimal rendezvous sequence for multiple GEO targets and minimize fuel consumption while completing all rendezvous missions. It is implemented based on the REINFORCE algorithm using a multihead attention neural network. In order to solve the problem that DRL training is difficult due to the time-varying characteristics of the space targets' positions, a data augmentation method is designed to strengthen the learning weights of different parameters. In order to solve the problem of trivial objective function gradient and slow iteration caused by the dense distribution of orbital planes, a loss function is designed to improve the earning efficiency of the neural network. Through multiple comparative experiments with existing algorithms, the calculation speed of DARA significantly exceeds that of commonly used heuristic mission planning algorithms. The results accuracy is better than that of various meta-heuristic algorithms and typical reinforcement learning algorithm, which proves the effectiveness and superiority of the proposed algorithm.

This brief leverages a novel resilient reinforcement learning (RL) technique to tackle the model-free optimal transient problem for virtual synchronous generators (VSGs) under denial-of-service (DoS) attacks. An optimal control strategy is established from real-time data of the VSGs under DoS attacks using the iterative algebraic Riccati equation (ARE). Furthermore, with a theoretical upper limit for DoS attack durations, system resilience is analyzed using the Lyapunov method. The resilient hybrid iteration (RHI) algorithm introduced in this brief eliminates the assumption of initial stable control in previous approaches, removes the need for precise system dynamics, and reduces computational requirements. Finally, the effectiveness of this approach is demonstrated through simulation experiments.