ISA transactions最新文献

To address uncertainty and inequality constraints in dual-arm collaborative robot systems, this paper proposes an adaptive robust control strategy based on the diffeomorphism technique. Specifically, system uncertainties are characterized using fuzzy set theory, while inequality constraints are systematically incorporated into the constraint-following control framework via diffeomorphism. Moreover, an adaptive robust control scheme is developed to ensure practical stability in the presence of uncertainties. In addition, the control parameters are optimized based on the fuzzy description of uncertainty, striking a balance between system performance and control cost. Finally, numerical simulations are conducted to validate the effectiveness of the proposed scheme.

This paper proposes a novel dynamic event-triggered approximate optimal consensus control scheme based on adaptive dynamic programming (ADP) for nonlinear multi-agent systems (MASs) with unknown dynamics, aiming to bridge the gap between theoretical control design and practical physical system applications. Firstly, a neural network (NN) state observer is developed to address the common challenge in physical systems where direct measurement of key states is either costly or technically infeasible due to sensor limitations. Typical examples include joint velocities of manipulators and angular positions of unmanned aerial vehicles. To enhance robustness against real-world disturbances, a disturbance-aware term is incorporated into the cost function, ensuring the scheme's adaptability to complex operating environments of physical systems. Secondly, a dynamic event-triggered mechanism (DETM) is integrated to significantly reduce communication and computational overhead. This reduction is critical for resource-constrained physical systems; a representative example is distributed robotic arms. Meanwhile, the DETM rigorously eliminates Zeno behavior to guarantee practical implementability. Additionally, a critic-only NN architecture is designed to approximate the solution of the Hamilton-Jacobi-Bellman (HJB) equation, which not only relaxes the restrictive persistent excitation (PE) condition but also reduces network complexity and computational load, making it suitable for real-time control of physical systems with limited on-board computing resources. Finally, the effectiveness and practicality of the proposed scheme are validated through two physics-relevant case studies: a nonlinear affine system mimicking industrial process dynamics and a multiple manipulator system. Simulation results demonstrate that the scheme achieves stable consensus tracking, robust disturbance rejection, and efficient resource utilization, providing a control solution for MASs.

Deep reinforcement learning (DRL) algorithms are increasingly applied to robotic manipulator planning. However, conventional DRL approaches suffer from slow learning convergence and low success rates in industrial trajectory planning tasks. To address these issues, this paper proposes an improved deep deterministic policy gradient (DDPG) algorithm that more effectively achieves time-optimal trajectory planning for robotic manipulators. Firstly, a radial basis function neural network is introduced to calculate nonlinear function values during parameter training to improve the learning convergence speed of the algorithm. The gradient descent algorithm is used to update the weights of the neural network. Meanwhile, the SumTree sample pool is used to screen high-quality samples and improve the utilization rate of the algorithm. The simulation experimental results show that compared with the traditional DDPG algorithm, the improved DDPG algorithm proposed in this paper has the torque and angle of each joint of the robotic manipulator change steadily, which improves the utilization rate of the algorithm in trajectory planning and the learning efficiency of the strategy.

Accurate real-time estimation of the instantaneous vehicle state plays a crucial role in modern automotive research, both in the state diagnostics and anomaly detection and in the design and development of advanced control systems and onboard monitoring strategies. In particular, accurate knowledge of chassis motion and wheel dynamics in response to road disturbances is essential for advanced control strategies aimed at simultaneously enhancing ride quality and handling. However, the road profile represents an unmeasured and highly variable input, often requiring complex and costly sensors such as LiDAR for direct observation: this motivates the development of virtual sensing approaches capable of inferring road irregularities from standard onboard sensors. This work presents a novel state observer based on an Extended Kalman Filter (EKF) architecture for the online estimation of road-induced excitations and key vehicle dynamic quantities, including chassis out-of-plane motions, suspension displacements, and tyre-loaded radii. The observer relies on a computationally efficient 7-degree-of-freedom vehicle model, analytically derived through a streamlined multibody formulation, and validated against a high-fidelity multibody reference model under two sensor configurations, both limited to signals typically available in mass-produced vehicles. The results achieved, even when using high-noise measurements, are encouraging for further applications in real-world virtual sensing scenarios.

In order to address the tracking accuracy degradation of the tank gun control system (TGCS) with inherent structural nonlinearity and feedback hysteresis under complex disturbances, a Disturbance Modeling Compensated Linear Active Disturbance Rejection Predictive Control (DMC-LADRPC) method is proposed. Firstly, the mathematical model of the TGCS transmission mechanism is derived, from which the disturbance modeling compensation (DMC) term is inversely deduced and incorporated as an improvement module of LADRC. Adjusted by a gun tracking error related threshold function, the DMC term compensates the controller's output quantity to directly enhance disturbance rejection performance. Furthermore, a synergistic network consisting of a prediction module and an optimization module is introduced within the LADRC framework. The prediction module combines a bidirectional long short-term memory (Bi-LSTM) network with a multi-head self-attention mechanism to predict short-term gun future servo instructions or motion trajectories based on historical data. The optimization module adopts an Actor-Critic framework with a reward function designed according to gun tracking error and its pitch rate, enabling the controller to dynamically select the optimal pre-control instruction from the predicted sequences through interactive iteration with the TGCS to achieve minimized tracking error and prevent vibration of the servo system induced by instruction oscillations, thereby effectively compensating for feedback loop hysteresis. The results of the multi-body dynamics co-simulations and experiments demonstrate that, compared with several high-performance controllers in this field, the proposed method significantly improves the response speed and reduces the tracking error of the TGCS under various typical working conditions.

Iterative learning control (ILC) is a control strategy specifically devised for finite-length batch processes that can be repeatedly executed. By iteratively refining the input signal across successive system trials, ILC enables accurate tracking of a predefined reference trajectory. Since its inception, this control methodology has evolved over four decades into a relatively mature and comprehensive theoretical framework. Nevertheless, in the past decade, the field has lacked systematic review and in-depth discussion on the overall progress of the field, with only a handful of studies offering limited retrospectives within specific subdomains. To provide a holistic understanding of the current state of the art and to identify promising directions for future investigation, this paper presents a literature review of recent key developments from five essential dimensions: system dynamics and settings, signal acquisition and transmission, reference trajectory, algorithm design and analysis, and implementations and applications. For each dimension, we summarize the major advancements and representative contributions, followed by critical discussions and forward-looking perspectives. This review aims to help researchers and practitioners in grasping the prevailing research trends and to inspire further theoretical and applied developments in ILC.

Variable impedance control (VIC) improves robotic performance in complex tasks, but time-varying impedance parameters may destabilize the system. This paper investigates stability constraints for the desired variable impedance model (DVIM). By leveraging an invariance-like theorem for non-autonomous systems, two relaxed state-independent global uniform exponential stability (GUES) constraints are derived for the DVIM under zero external force. Compared with existing results, the proposed constraints are less conservative and are applicable to general impedance matrices. The first constraint allows non-strict inequalities, whereas the second admits damping that is neither positive definite nor differentiable and does not impose an upper bound on the stiffness rate of variation. Based on an input-to-state stability (ISS) framework, the proposed constraints guarantee bounded DVIM states under bounded external forces. Robustness of the VIC framework enforcing state-independent stability constraints is further established in the presence of bounded trajectory tracking errors and bounded external force measurement errors. Simulation and experimental results validate the proposed findings.

The inherent difficulty in acquiring high-frequency process data and sparse information representation in complex industrial environments often leads to small-data problems, which degrade the generalization ability and prediction reliability of data-driven soft sensors. To address this challenge, a Hierarchical Temporal Refinement Diffusion Network is proposed for high-fidelity soft sensing data augmentation. By systematically decomposing industrial time-series patterns into multi-scale temporal dependencies, this method realizes coarse-to-fine generation of virtual samples through progressive diffusion refinement. The hierarchical structure effectively preserves both the macro-scale process trends and micro-scale fluctuation characteristics of the generated data. This architecture also integrates a dedicated noise prediction network, enabling simultaneous global correlation modeling and local feature extraction. Furthermore, a dynamic weighting strategy is developed for the joint training of hybrid datasets, which adaptively coordinates the learning of real-sample characteristics and virtual-sample regularities. The effectiveness of the proposed method is verified through two industrial application cases.

This study proposes a novel neural network-based active disturbance rejection control (ADRC) scheme, enhanced with the adaptive moment estimation (ADAM) optimizer for precise trajectory tracking of quadrotor UAVs under disturbances. A cascaded ADRC architecture is designed for attitude control, featuring an inner loop that enables rapid attitude response and an outer loop that handles low-frequency disturbances and uncertainties. This study implements the online adaptive tuning of extended state observer (ESO) parameters via a radial basis function neural network (RBFNN), which dynamically adjusts observer gains as disturbances evolve. The integration of the ADAM optimizer accelerates RBFNN training compared to traditional backpropagation by leveraging gradient moment estimations for adaptive learning rates. This approach enables real-time rejection of time-varying disturbances and eliminates the need for manual parameter recalibration. The theoretical stability of the proposed system is rigorously proven using Lyapunov analysis. Hardware-in-the-loop experiments validate the superior performance of the proposed scheme in three scenarios.

Chatter remains a significant challenge in milling operations, resulting in deteriorated surface quality, accelerated tool wear, and excessive noise-induced environmental impacts. Timely chatter detection is therefore essential for maintaining sustainable manufacturing processes. The primary contribution of this work is the introduction of a smart chatter identification approach designed for different tool-workpiece systems in milling. First, chatter and stable frequency were automatically identified based on the frequency distribution characteristics of milling chatter using frequency search methods. Then, an adaptive signal reconstruction approach integrating Wavelet Packet Decomposition (WPD) was developed to automatically extract chatter and stable signal components. Based on these developments, a novel energy indicator, Chatter to Stable Energy Ratio after Signal Reconstruction (CSERSR), was constructed through signal energy calculations to realize different degrees of chatter detection in real-time. A simulation study was conducted to preliminarily validate the effectiveness of the proposed method, and experimental validation further demonstrates its capability to accurately identify different degrees of chatter across varying tool-workpiece systems. Comparative analyses with existing methodologies further substantiate the reliability and adaptability of the presented approach.