ISA transactions最新文献

This paper addresses the real-world, safety-critical planning problem for a wheeled mobile robot with an unknown model, focusing on obstacle avoidance in complex environments. Data-enabled predictive control (DeePC) is a renowned optimal control framework for unknown linear time-invariant systems, offering the ability to predict future system trajectories using only past input/output measurements, without requiring explicit system identification. To enable its application to nonlinear systems, a local linear approximation technique with online data updates is adopted. To further improve robustness, a novel adaptive Lasso-based regularization term is incorporated into the cost function. Safety is a critical challenge for data-enabled control of nonlinear systems, which is addressed by incorporating discrete-time control barrier functions into a quadratic programming-based safety filter operating alongside the DeePC. We provide a formal proof of the practical stability of the closed-loop system, ensuring bounded trajectory tracking. Real-time simulations in the ROS-Gazebo environment show that adaptive regularization reduces the trajectory tracking error by about 20.2%.

This paper presents an adaptive discrete hierarchical sliding-mode controller (AHSMC) based on a cascaded state observer (ALKF) to address inaccurate state estimation and control performance degradation caused by input saturation in industrial overhead crane anti-sway control. First, a cascaded observer integrating a Luenberger observer and an adaptive Kalman filter with an adaptive weighting mechanism is constructed to enhance the robustness and accuracy of state estimation. Subsequently, an adaptive discrete hierarchical sliding-mode controller featuring an input saturation compensation mechanism is designed. This controller suppresses the effects of input saturation and improves system robustness by dynamically adjusting the parameters of the sliding-mode reaching law, without introducing additional system complexity. The stability of the proposed control scheme is theoretically analyzed and proven. Finally, comparative experiments conducted on a 0.5-ton industrial overhead crane demonstrate superior performance in swing suppression.

The spherical mechanism inertially stabilized platform (ISP) offers significant advantages, including enhanced motion flexibility and superior space utilization efficiency. The joint space control (JSC) of such systems involves computationally intensive processes and fails to fully mitigate multisource disturbances. In order to further improve the performance, this paper proposes a double-loop disturbance observer (DLDOB) method in the task space. There are three main parts of this control method. The first portion is designed as a gravity compensation controller in order to solve the problem of unbalanced gravity distribution caused by the asymmetric structure of the spherical mechanism. The second portion is the main controller of the system, which is the DLDOB. It includes the disturbance observer (DOB) in the joint space and the adaptive sliding mode disturbance observer (ASMDO) in the task space. This method simplifies the computationally intensive joint-to-task space transformation process. It estimates and compensates for system disturbances in both joint and task spaces, thereby improving system robustness. The third portion is a velocity feedback controller, which ensures the inertial stabilization control of the ISP. To verify the effectiveness of the proposed method, the stability of the control system was proven in the paper using the Lyapunov method. Besides, simulations and experiments were carried out. It was verified that the proposed method has stronger disturbance rejection and stabilization performance compared with the conventional method.

This paper addresses the position-attitude tracking control problem of quadrotor unmanned aerial vehicles (QUAVs). Over the past decade, the attitude control of QUAVs has achieved global exponential stability (GES) by leveraging Lie group theory to address the singularities in the attitude error representation. However, the position control of QUAVs has not yet reached GES. Based on the geometric position control law and the Lie-algebra-based attitude control law, this paper proposes an improved robust position-attitude control law. For attitude control, a method for computing attitude error represented in Lie algebra so(3) is proposed to address the singularity of the logarithmic map. For position control, a novel hybrid computational scheme of the desired rotation matrix is proposed to address the body flipping problem when the desired yaw direction is parallel to the desired thrust direction. Both attitude and position errors are globally defined on their respective manifolds, ensuring singularity-free tracking performance over the entire configuration space. For stability analysis, some Lyapunov-based techniques are employed to bound the high-order coupling terms between position and attitude. In the presence of unknown disturbances, the position-attitude control system is proven to be semi-globally exponentially stable. The proposed control law is validated through numerical simulations and indoor flight experiments under wind disturbances, in comparison with the classical geometric control method.

Extracting effective anomaly features and mitigating interference from missing data is crucial for multivariate time series anomaly detection in equipment. However, existing studies tend to focus on modeling under complete data scenarios and fail to adequately account for the challenges posed by complex missing data patterns, ultimately compromising the reliability of anomaly detection. To this end, this work proposes a representation-enhanced graph temporal convolutional network (REGTCN) under a complex missing pattern for equipment anomaly detection. This method is designed as a jointly optimized framework integrating reconstruction-based and prediction-based paradigms to enhance the representation of system health status. In the reconstruction module, we develop a missing-tolerant masked graph attention (MGAT) network to mitigate the adverse effects of missing patterns. In the prediction module, we propose an adaptive multi-scale temporal convolutional interaction network (AMTCIN) to capture sufficient temporal features. Finally, Extensive experiments are conducted under various missing-data scenarios. Experimental results demonstrate that our method outperforms all baseline models.

This paper investigates the formation problem of unmanned aerial vehicles under spatiotemporal coupling constraints. Firstly, unlike existing stability theories based on predefined-time, this paper designs a predefined-time formation controller based on time-based generator and sliding mode control technology, which mitigates the initial input saturation issue. Secondly, a collision prediction mechanism is designed based on the artificial potential field method, which avoids unnecessary obstacle avoidance behavior when unmanned aerial vehicles detect non threatening obstacles and reduces redundant resource consumption. Finally, simulation results are presented to indicate that the proposed scheme enables the unmanned aerial vehicles to achieve formation and reconstruct the formation within the predefined-time after obstacle avoidance.

Uncertain delays or sinusoidal disturbances pose significant challenges for traditional active disturbance rejection control (ADRC) of time-delayed processes, leading to potential instability or performance degradation. To address these difficulties, this paper investigates the control synthesis from a unified perspective of observation reconfiguration (OR), the core concept of which involves reconfiguring the estimated state according to specific challenges. A basic ADRC with a reduced-order model-driven extended state observer is first developed, featuring a concise control structure, intuitive tuning guidance, and detailed stability analysis. To enhance delay robustness, a filter-based OR module reconstructs the estimated state by reshaping the key frequency-domain characteristics. Furthermore, the coordination between the ADRC system and the observer loop is clarified, enabling control parameters to be retuned while maintaining comparable delay robustness. For sinusoidal disturbances, the observer is augmented by incorporating resonant dynamics to generate necessary compensation signals. A delay block subsequently acts as the OR module to counteract the phase shift between the control and disturbance signals precisely. Comparative simulations demonstrate the effectiveness of the proposed method. Experimental validation on an industrial-grade platform further confirms its promising practical applicability.

This paper aims to address the consensus problem of heterogeneous nonlinear multi-agent systems (MASs) under sampled-data control with inter-agent communication delays. To tackle this problem effectively, a two-sided looped Lyapunov-Krasovskii functional is proposed to jointly capture sampling-induced holding and network-induced delays, while accounting for heterogeneous agent dynamics with Lipschitz nonlinearities. Since applying a looped functional approach to the consensus problem of multi-agent systems can lead to excessive computational complexity, the proposed framework is structured to reduce conservatism and facilitate scalable condition design by unifying heterogeneous network delays and systematically distinguishing between state sampling and control update instants. In the examples, the influence of the variables used in the looped functional on performance and computational burden is analyzed, and the practical feasibility of the proposed approach is demonstrated through simulations on the angular position consensus problem in a multi-agent system composed of brushed DC motors.

In rotating machinery condition monitoring, there is a growing demand for diagnosing bearing faults under time-varying rotational speeds using the tacholess order tracking (TLOT) technique. A critical step in TLOT is the rapid and accurate estimation of rotational frequency via time-frequency analysis (TFA). However, the current TFA methods often suffer from issues such as weak time-frequency energy concentration, uncertainty in window width selection, or high computational complexity. Moreover, achieving a high-resolution time-frequency representation (TFR) remains challenging, particularly for strong frequency modulation and amplitude modulation (FM-AM) signals. To overcome these limitations, this article introduces an improved S high-order synchroextracting transform (ISHSET). First, a novel single-parameter adaptive window function is designed to minimize the computational complexity while enhancing time-frequency energy concentration. Subsequently, an iterative estimator capable of estimating high-order instantaneous frequency (IF) is developed to accurately approximate the actual frequency values. Finally, the synchroextracting framework is incorporated to further elevate the resolution and readability of the TFR. Simulation analysis results indicate that ISHSET achieves a higher resolution, a more accurate and clearer characterization of the time-varying properties, and better noise robustness. Experimental validation through rolling bearing fault cases not only demonstrates the efficacy of the proposed method in estimating rotational frequency but also substantiates its superiority and practicality in extracting fault characteristics from rolling bearings under time-varying rotational speeds.