搜索结果：ISA transactions

Mode identification for hybrid systems with unknown inputs has important applications to multiple-model approaches for health monitoring of cyber-physical systems. Active mode discerning applies a certain auxiliary control signal or reconfigures the controller to excite the hybrid system such that modes of the system are guaranteed to be distinguished, which is associated with the idea of active fault diagnosis in fault detection and isolation (FDI). The mode discerning auxiliary control design for nonlinear hybrid systems subject to unknown-but-constrained inputs with applications to active fault diagnosis has rarely been considered in existing works, and is far from being thoroughly addressed yet due to its nonlinear, nonconvex, bi-level, and multi-objective nature. To tackle the problem, a novel optimization scheme presenting a bi-level structure is proposed in this paper for the formulated active mode discerning problem for smooth nonlinear hybrid systems with unknown inputs. The proposed scheme can solve the local optimization problem of active mode discerning in a numerically stable manner without conservative convex relaxations. The effectiveness of the proposed scheme is demonstrated using an example of active fault detection for an aircraft control surface driven by an electro-mechanical actuator.

To address multi-distribution perception and temporal generalization challenges in cross-aircraft aero-engine monitoring, the Lyapunov-Schmidt Multi-distributed Perception Network (LSMPNet) is developed as a continuous-time domain generalization framework grounded in improved Lyapunov-Schmidt reduction (LSR). LSMPNet reformulates state monitoring as a continuous nonlinear system, leveraging a T-LSR Decomposer to perform structured decomposition and employing stacked T-LSR Blocks for hierarchical learning of dynamic features. A time-difference-based operator enhances sensitivity to continuous distribution shifts, while complementary manifold topologies strengthen global distribution perception and noise suppression. The linearized operator design ensures low computational complexity and interpretability. Extensive experiments on public benchmarks and real-world cross-aircraft datasets demonstrate superior performance in degradation modeling and water-wash early warning tasks.

This paper addresses cooperative transportation of multi-mobile robots (MMRs) under input delays, frictional forces, external disturbances and parameter perturbations by proposing a distributed leader-follower cooperative control strategy based on a Pade-augmented artificial potential field (P-APF) to enhance stability and robustness. Firstly, a dynamic model of the MMRs system (consisting of the MMRs formations and the workpieces) is established, which is further transformed into an approximate delay-free form via the first-order Pade approximation to handle input delays. Subsequently, a fixed-time nonlinear disturbance observer (NDO) is designed to accurately estimate and compensate for the lumped disturbances composed of frictional forces, external disturbances and parameter perturbations, thereby significantly improving the system's resistance to uncertainties. Then, the equivalent delay state variables are established and incorporated into the artificial potential field (APF) function to develop a cooperative controller based on the P-APF method, which ensures accurate formation convergence and maintains inter-robot connectivity. The asymptotic convergence of the nominal error system (residual-free) is proved by LaSalle's invariance theorem and Barbalat's lemma. For the actual closed-loop system, the formation and velocity errors are uniformly ultimately bounded under the bounded residual term. Finally, simulation results demonstrate the effectiveness of the proposed approach in achieving stable cooperative transportation of the MMRs system.

In this paper, the output-feedback control problem under adaptive neural network (NN) hybrid dynamic event-triggered control (HDETC) for nonlinear heterogeneous vehicular platoon systems (HVPSs) is investigated using extended state observers (ESOs) and high-order feedback filters. NNs are utilized to approximate the lumped uncertainties of HVPSs, and ESOs are adopted to estimate unknown internal states and external disturbances. Then, based on the limited network communication bandwidth and resources, to reduce the data transmission frequency, which encompasses inter-vehicle communication, as well as the intra-vehicle communication within each vehicle including the uplink from sensors and the downlink to actuators, a novel hybrid dynamic event-triggered control (HDETC) communication mechanism is designed. Furthermore, in order to produce smooth output estimates with the required differentiability, high-order feedback filters are developed, thus resolving the non-differentiability problem of virtual control signals in the backstepping control design. By formulating a practical spacing policy and carrying out Lyapunov stability analysis, an adaptive NN-HDETC control scheme is further presented, which can ensure the boundedness of all signals and string stability of the HVPSs. Therefore, both the individual and string stability of HVPSs can be achieved. Finally, simulation analyses are carried out to confirm the efficacy of the developed heterogeneous vehicular platoon control (HVPC) scheme.

This paper investigates the distributed resource allocation problem in second-order nonlinear multi-agent systems (MASs) and proposes an integrated adaptive dynamic programming (ADP) framework with a dynamic event-triggered mechanism. A coupled performance index based on resource allocation errors is first formulated, embedding a positive-definite quadratic form to transform the resource allocation control problem into the design of optimal control policies for all agents. Within this framework, a fixed-time single-critic learning law is developed to approximate the Hamilton-Jacobi-Bellman (HJB) equation online. The dynamic event-triggered mechanism updates the control input only when the adaptive triggering condition is violated. Theoretical analysis guarantees the uniform ultimate boundedness of all closed-loop signals. Numerical simulations verify the effectiveness of the proposed framework in achieving accurate resource allocation while balancing allocation errors and control efforts.

The morphing glide aircraft (MGA) can adapt to complex environments and mission requirements by altering its aerodynamic configuration through the jettisoned wings, which face the challenges in dynamic and aerodynamic characteristics variation, as well as lumped uncertainty. To deal with the above issues, a neural network (NN) parameter identification-based prescribed-time adaptive control for MGA is proposed. Firstly, a time-scale function is proposed, which avoids the singularity and the unrealistic issue of unbounded growth in control effort. Further, a fractional-power prescribed-time Lyapunov stability theorem is established, which overcomes the limitation of conventional theorems in analyzing robust sliding mode control with non-smooth control terms, providing a theoretical foundation for the design and stability analysis of prescribed-time fractional-power sliding mode controllers. Then, for the issue of aerodynamic parameter uncertainty, an NN parameter identification method is proposed. Based on this, a prescribed-time adaptive sliding mode control of MGA is proposed to guarantee the controller error convergence in the prescribed-time, independent of initial conditions and control parameters. Finally, the proposed algorithm is employed to achieve attitude tracking for the MGA, and the effectiveness is illustrated.

Cloud-based intelligent connected vehicles (CICVs) are vulnerable to denial of service (DoS) and false data injection (FDI) attacks through wireless control channels, which can significantly degrade trajectory tracking performance and compromise driving safety. To improve control resilience and safety under hybrid cyber attacks, this paper proposes a safety-guided reinforcement learning control framework for CICVs. A unified discrete-time hybrid attack model is established to characterize the temporal, frequency, and amplitude constraints of hybrid cyber attacks, and reachable and controllable set analyses are performed to quantify the resulting degradation of system stability and safety margins. On this basis, a robust control Lyapunov function and control barrier function (CLBF)-based safety controller is developed by explicitly incorporating attack bounds into tightened stability and safety constraints. Furthermore, the tightened CLBF mechanism is embedded into the reinforcement learning framework through expert-guided policy learning, safety-aware reward shaping, and online action projection, thereby combining model-based safety assurance with data-driven performance optimization. Finally, simulations and cloud-based hardware-in-the-loop (C-HiL) experiments are conducted to validate the proposed method. The results show that, compared with the CLBF-based controller, the proposed method reduces the average lateral displacement error and average heading angle error by 87.42% and 94.57%, respectively, while achieving smoother control behavior and a larger safe controllable region under hybrid cyber attacks.

The purpose of this paper is to address the parameter identification problem of Hammerstein systems with dead-zone nonlinearity interfered with by colored noise. By introducing a switching function, the linear parameter expression of the piecewise nonlinear subsystem is obtained. On this basis, combined with the maximum likelihood principle, the maximum likelihood weighted recursive least squares (ML-WRLS) identification algorithm is constructed. A two-stage recursive least squares algorithm is proposed to improve estimation accuracy while reducing the computational burden. Compared with the ML-WRLS algorithm, the two-stage maximum likelihood weighted recursive least squares algorithm can provide better steady-state performance. The effectiveness of the proposed algorithms is demonstrated via examples.

To address the challenges of parameter perturbations and load disturbances in dual permanent magnet synchronous motor systems, this paper proposes a control scheme integrating prescribed performance cross-coupling control (PPCCC), adaptive extended state observer (AESO), and novel sliding mode control (NSMC). Key innovations: (1) A hybrid sliding mode reaching law is designed, which dynamically adjusts the reaching speed through a combination of exponential and power terms. A sigmoid function is incorporated to effectively suppress chattering. (2) PPCCC is introduced to constrain the synchronization error's convergence rate and overshoot through a logarithmic barrier function transformation. (3) AESO with adaptive bandwidth estimates lumped disturbances for real-time compensation. Simulations and experiments verify that the scheme confines the synchronization error within preset bounds, suppresses disturbances effectively, and outperforms existing methods.

This paper investigates the trajectory-tracking problem of an uncertain robotic system subject to coupled output constraints. A decoupling approach is introduced to convert the coupled constraints into independent time-varying constraints on the transformed output coordinates. Based on this decoupling, a partial controller is developed using a time-varying barrier Lyapunov function (TVBLF) to enforce constraint satisfaction. In addition, fuzzy logic systems (FLSs) are employed to approximate the uncertain dynamics caused by unknown system parameters. To reduce unnecessary state transmissions and controller updates, a state-based event-triggered mechanism (ETM) is incorporated without significantly degrading tracking performance. A Lyapunov-based stability analysis is further provided to establish the feasibility of the proposed method. Finally, simulation results validate the effectiveness of the developed control scheme.

Improving computational efficiency and reducing the impact of parameter mismatch are important for the model predictive control (MPC) of permanent magnet synchronous motors (PMSM). In this article, an MPC algorithm based on reference voltage vector optimization is proposed. The proposed MPC is based on the discrete space vector modulation (DSVM) to synthesize 38 voltage vectors. This method reduces the number of candidate voltage vectors from 38 to 7 through the reference voltage vector. To overcome the problem of poor parameter robustness, a data prediction model based on behavioral system theory is constructed. This method does not require accurate motor parameters, and only needs input/output data. Then, a reference voltage vector solving method based on the data prediction model is designed. Finally, the experimental results indicate that the method used can reduce computational burden relative to exhaustive DSVM search and exhibit stronger robustness in the case of parameter mismatch.

High-precision and fast-response trajectory tracking control with high energy efficiency for unmanned underwater vehicles (UUVs) plays an increasingly important role in numerous marine tasks. Although the predefined-time control offers fast convergence of the system trajectories, it tends to overestimate the actual settling time and produce excessive control effort at the initial stage which may result in actuator saturation. This paper presents an energy-efficient trajectory tracking control strategy with exact-time convergence for UUVs under disturbances. Firstly, by employing the trigonometric-type function, an exact-time recursive sliding mode control (ETRSMC) scheme is presented, where the tracking errors reach zero at exact reference time for arbitrary initial conditions and the initial excessive control inputs are avoided. Then, an exact-time disturbance observer (ETDO) is developed for accurately estimating the external disturbance at the exact time independent of the initial estimation error, and then integrated to attenuate external disturbances while keeping the small discontinuous control gain of the designed ETRSMC scheme. Comparative numerical simulations are performed for two trajectory tracking scenarios. Compared with the predefined-time counterpart, the proposed ETDO-ETRSMC scheme guarantees exact-time convergence at any user-defined settling time, reduces peak control effort by 88.2%-98.7%, and lowers average energy consumption by 35.8%-96.9% across extensive simulation scenarios, while maintaining comparable steady-state accuracy and effectively suppressing chattering.

Accurate fault diagnosis is critical to ensuring the reliable operation of equipment. However, traditional deep learning methods typically rely on a predefined set of known fault types, limiting their ability to identify unknown faults. Thus, this paper proposes an open set fault diagnosis method based on Hyper-opinion Dual Branch Evidential Deep Learning with Distance Perception (HDEDL-DP). Firstly, a dual-branch framework is proposed to capture sharp evidence supporting single proposition and vague evidence supporting composite propositions via hyper-opinion modeling. A projection mechanism converts the vague evidence into sharp evidence for model inference. Secondly, mahalanobis distance is used to provide distance perception. Finally, an OOD score integrating uncertainty estimation and distance perception is designed to identify unknown fault classes. Two case studies validate the effectiveness of HDEDL-DP.

This paper explores the hierarchical optimal control strategy for high-order nonlinear systems with unknown dynamics, with a focus on two-player scenarios. The Stackelberg differential game theory provides a leader-follower framework to investigate the optimal control problem for players within a predefined sequential decision order. The optimized controllers for the corresponding subsystems of each player are designed using the command filtered backstepping method. In each subsystem, an adaptive dynamic programming architecture is employed to derive the Stackelberg equilibrium solution for the player. To this end, actor-critic functions are introduced to approximate the cost function and execute the control behavior for all players. An identifier based on fuzzy logic systems is utilized to estimate the unknown nonlinear dynamics arising from modeling inaccuracies. To avoid repeated differentiation of the virtual control law, a command filter is adopted to reduce the computational burden by directly obtaining virtual control without differentiation. Finally, simulation results demonstrate that the proposed optimal control scheme can achieve the desired control objective under the hierarchical performance index.

In engineering practice, intelligent fault diagnosis for high-end equipment often involves small-sample, multi-class imbalanced data with noise and complex intra- and inter-class distributions. Existing sampling methods may generate low-quality samples, amplify noise, and depend heavily on hyperparameters. To address these issues, this paper proposes an interpretable fault diagnosis framework, termed Adapted Oversampling-based Multi-layer Support Vector Machines (AM-SVMs). The framework embeds a Multi-mechanism Adaptive Oversampling Technique (MAOTE) into a multi-layer LSSVM architecture. MAOTE adaptively determines sampling strategies according to data characteristics and employs a Newton-Raphson-inspired evolutionary mechanism to search for high-quality candidate solutions guided by multi-class classification performance. The optimized solutions are reorganized into diverse and representative fault samples, and a classifier-feedback-based evaluation mechanism improves distributional consistency and interpretability between generated and real samples. Finally, balanced feature samples are used to train a multi-class LSSVM classifier, yielding a robust and generalizable diagnostic model. Experiments on public bearing datasets and self-collected data show that AM-SVMs outperforms ten data augmentation methods and eight multi-class classifiers in diagnostic accuracy and robustness, demonstrating its effectiveness for imbalanced fault diagnosis in intelligent manufacturing.

Networked control systems are widely adopted in modern applications. Nevertheless, learning optimal control policies from data under communication attacks remains challenging. In this paper, we propose a bias-compensated Q-learning algorithm for knowledge-informed optimal tracking control. The method considers discrete-time linear time-invariant systems subject to control-channel denial-of-service attacks. Through a bias compensation mechanism that addresses control-input losses, the proposed method eliminates the need for explicit expectation computation in Q-function updates and enables direct data-driven learning under DoS attacks. Moreover, the proposed method requires no knowledge of system dynamics beyond the known system structure and supports data collection under arbitrary behavior policies. This design preserves high data efficiency, computational simplicity, and robustness within a model-free framework. Numerical simulations on a planar quadrotor position-tracking task show that a Riccati-equivalent optimal policy can still be learned from data under 40% DoS attacks.

Rebar node binding on construction sites is predominantly carried out manually, and existing mobile robots struggle to navigate complex rebar mesh due to desynchronization and kinematic instability caused by the unstructured grid-like holes and uneven nodes. To address this, we propose a dual-motor cross-coupled synchronous control strategy based on impedance control, applied to a wheel-legged rebar binding robot. This innovative strategy is accomplished by integrating synchronization error terms into the impedance model to establish a "virtual mechanical coupling" between bilateral motors, allowing for real-time torque redistribution when the robot encounters uneven environmental resistance. We conduct a dynamic analysis of the robot's lateral movement mechanism, deriving the theoretical output torque, followed by comprehensive validation. Quantitative results demonstrate that under conditions with obstacle interference, the proposed method reduces the peak displacement difference by 29.5% and the displacement variance by 75.9% compared to traditional decoupled impedance control. Notably, in high-friction threaded rebar mesh, the system successfully neutralizes internal force asymmetries, achieving nearly identical cumulative torque profiles for both legs. Furthermore, even at a velocity of 1600 mm/s, the longitudinal error is limited to 2.75 mm, effectively suppressing cumulative deviations stemming from off-center of gravity and mesh deformation. By harmonizing dynamic load distribution, this strategy ensures superior robustness and kinematic consistency for robotic platforms in complex construction environments.

To address the problem of reduced effective flux linkage, degraded dynamic performance, and poor robustness against load variations in permanent magnet synchronous motors (PMSMs) under demagnetization faults, this study presents a composite fault-tolerant predictive control strategy. Firstly, mathematical models of the motor under both normal and demagnetized operating conditions are established. Subsequently, within the deadbeat predictive current control (DPCC) framework, a fast integral terminal sliding mode observer (FITSMO) is designed to enable real-time flux linkage estimation and achieve demagnetization compensation. Simultaneously, to enhance dynamic accuracy and resilience to load disturbances, an adaptive fast terminal sliding mode controller (AFTSMC) and an improved super-twisting algorithm-based load disturbance observer are developed. The observed disturbance values are fed into the speed loop for active compensation. The stability and finite-time convergence of the developed composite fault-tolerant control system incorporating disturbance compensation are proven using Lyapunov theory. Finally, the efficacy of the developed control strategy is validated by experiments. Experimental results demonstrate that the proposed framework has superior fault tolerance, enhanced dynamic response, and significantly improved robustness compared with conventional methods.

Industrial bearing fault diagnosis is constrained by the scarcity of labeled data. Although abundant data is available from diverse equipment, its heterogeneity in sampling parameters, operational conditions, and label granularity hinders direct use by existing methods. To bridge this gap, we propose a Multi-source Domain Adaptive Siamese Network (MDASN) to exploit such heterogeneous data effectively. It integrates multi-domain sample alignment for unified preprocessing, a Bidirectional Importance Attention mechanism to dynamically localize discriminative fault characteristics, and a detail division strategy for fine-grained feature learning. Evaluations on five bearing datasets show that MDASN significantly improves diagnostic accuracy, with gains of 26% on laboratory data and 24% on industrial data. Using only one sample per class, it achieves 86.74% accuracy in a challenging wind turbine bearing task, demonstrating strong potential for robust, data-efficient bearing health monitoring.