IEEE/CAA Journal of Automatica Sinica

Securing the Future after PagerBombs: Lifecycle Protection of Smart Devices via Blockchain Intelligence

Fei Lin, Qinghua Ni, Jing Yang, Juanjuan Li, Nan Zheng, Levente Kovács, Radu Prodan, Mariagrazia Dotoli, Qing-Long Han, Fei-Yue Wang

2024, 11(12): 2355-2358. doi: 10.1109/JAS.2024.125031

Abstract(35) HTML (7) PDF(17)

Abstract:

Pager Explosion: Cybersecurity Insights and Afterthoughts

Chuan Sheng, Wanlun Ma, Qing-Long Han, Wei Zhou, Xiaogang Zhu, Sheng Wen, Yang Xiang, Fei-Yue Wang

2024, 11(12): 2359-2362. doi: 10.1109/JAS.2024.125034

Abstract(20) HTML (9) PDF(14)

Abstract:

Analysis and Control of Frequency Stability in Low-Inertia Power Systems: A Review

Changjun He, Hua Geng, Kaushik Rajashekara, Ambrish Chandra

2024, 11(12): 2363-2383. doi: 10.1109/JAS.2024.125013

Abstract(57) HTML (3) PDF(26)

Abstract:
Power electronic-interfaced renewable energy sources (RES) exhibit lower inertia compared to traditional synchronous generators. The large-scale integration of RES has led to a significant reduction in system inertia, posing significant challenges for maintaining frequency stability in future power systems. This issue has garnered considerable attention in recent years. However, the existing research has not yet achieved a comprehensive understanding of system inertia and frequency stability in the context of low-inertia systems. To this end, this paper provides a comprehensive review of the definition, modeling, analysis, evaluation, and control for frequency stability. It commences with an exploration of inertia and frequency characteristics in low-inertia systems, followed by a novel definition of frequency stability. A summary of frequency stability modeling, analysis, and evaluation methods is then provided, along with their respective applicability in various scenarios. Additionally, the two critical factors of frequency control—energy sources at the system level and control strategies at the device level—are examined. Finally, an outlook on future research in low-inertia power systems is discussed.

Disturbance Rejection for Systems With Uncertainties Based on Fixed-Time Equivalent-Input-Disturbance Approach

Qun Lu, Xiang Wu, Jinhua She, Fanghong Guo, Li Yu

2024, 11(12): 2384-2395. doi: 10.1109/JAS.2024.124650

Abstract(263) HTML (7) PDF(80)

Abstract:
This paper presents a fixed-time equivalent-input-disturbance (EID) approach to deal with the problem of robust output-feedback control for perturbed uncertain systems. This method uses the basic structure of the conventional EID approach and treats uncertainties and disturbances as a lumped disturbance on the control-input channel. A fixed-time state observer enables state estimation, which resolves the causality issue in an EID-based control system, is finished in a fixed time. An implicit Lyapunov function, the homogeneity with dilation, the input-to-state stability, and the small-gain theorem are used to analyze the convergence and robustness of the EID-based system with measurement noise. Numerical and experimental results are presented to demonstrate the effectiveness and superiority of the proposed method.

Distributed Fixed-Time Optimal Energy Management for Microgrids Based on a Dynamic Event-Triggered Mechanism

Feisheng Yang, Jiaming Liu, Xiaohong Guan

2024, 11(12): 2396-2407. doi: 10.1109/JAS.2024.124686

Abstract(88) HTML (7) PDF(39)

Abstract:
The article investigates the optimal energy management (OEM) problem for microgrids. To figure out the problem in fixed time and alleviate communication load with limited resources, this article devises a novel fixed-time stability lemma and an event-triggered (ET) fixed-time distributed OEM approach. Using Lyapunov stability theory, the distributed approach has been proven to converge in fixed time and the upper bound on convergence time can be derived without dependence on the initial states. The dynamic ET method is raised to dynamically adjust the triggering threshold and reduce communication redundancy. In addition, Zeno behavior is avoided. Simulations are given to show the effectiveness and advantage of the designed distributed OEM method.

Safe Q-Learning for Data-Driven Nonlinear Optimal Control With Asymmetric State Constraints

Mingming Zhao, Ding Wang, Shijie Song, Junfei Qiao

2024, 11(12): 2408-2422. doi: 10.1109/JAS.2024.124509

Abstract(137) HTML (9) PDF(56)

Abstract:
This article develops a novel data-driven safe Q-learning method to design the safe optimal controller which can guarantee constrained states of nonlinear systems always stay in the safe region while providing an optimal performance. First, we design an augmented utility function consisting of an adjustable positive definite control obstacle function and a quadratic form of the next state to ensure the safety and optimality. Second, by exploiting a pre-designed admissible policy for initialization, an off-policy stabilizing value iteration Q-learning (SVIQL) algorithm is presented to seek the safe optimal policy by using offline data within the safe region rather than the mathematical model. Third, the monotonicity, safety, and optimality of the SVIQL algorithm are theoretically proven. To obtain the initial admissible policy for SVIQL, an offline VIQL algorithm with zero initialization is constructed and a new admissibility criterion is established for immature iterative policies. Moreover, the critic and action networks with precise approximation ability are established to promote the operation of VIQL and SVIQL algorithms. Finally, three simulation experiments are conducted to demonstrate the virtue and superiority of the developed safe Q-learning method.

Cas-FNE: Cascaded Face Normal Estimation

Meng Wang, Jiawan Zhang, Jiayi Ma, Xiaojie Guo

2024, 11(12): 2423-2434. doi: 10.1109/JAS.2024.124899

Abstract(58) HTML (6) PDF(17)

Abstract:
Capturing high-fidelity normals from single face images plays a core role in numerous computer vision and graphics applications. Though significant progress has been made in recent years, how to effectively and efficiently explore normal priors remains challenging. Most existing approaches depend on the development of intricate network architectures and complex calculations for in-the-wild face images. To overcome the above issue, we propose a simple yet effective cascaded neural network, called Cas-FNE, which progressively boosts the quality of predicted normals with marginal model parameters and computational cost. Meanwhile, it can mitigate the imbalance issue between training data and real-world face images due to the progressive refinement mechanism, and thus boost the generalization ability of the model. Specifically, in the training phase, our model relies solely on a small amount of labeled data. The earlier prediction serves as guidance for following refinement. In addition, our shared-parameter cascaded block employs a recurrent mechanism, allowing it to be applied multiple times for optimization without increasing network parameters. Quantitative and qualitative evaluations on benchmark datasets are conducted to show that our Cas-FNE can faithfully maintain facial details and reveal its superiority over state-of-the-art methods. The code is available at https://github.com/AutoHDR/CasFNE.git.

Multi-View Dynamic Kernelized Evidential Clustering

Jinyi Xu, Zuowei Zhang, Ze Lin, Yixiang Chen, Weiping Ding

2024, 11(12): 2435-2450. doi: 10.1109/JAS.2024.124608

Abstract(13) HTML (4) PDF(0)

Abstract:
It is challenging to cluster multi-view data in which the clusters have overlapping areas. Existing multi-view clustering methods often misclassify the indistinguishable objects in overlapping areas by forcing them into single clusters, increasing clustering errors. Our solution, the multi-view dynamic kernelized evidential clustering method (MvDKE), addresses this by assigning these objects to meta-clusters, a union of several related singleton clusters, effectively capturing the local imprecision in overlapping areas. MvDKE offers two main advantages: firstly, it significantly reduces computational complexity through a dynamic framework for evidential clustering, and secondly, it adeptly handles non-spherical data using kernel techniques within its objective function. Experiments on various datasets confirm MvDKE’s superior ability to accurately characterize the local imprecision in multi-view non-spherical data, achieving better efficiency and outperforming existing methods in overall performance.

High-Order Fully Actuated System Models for Strict-Feedback Systems W

ith Increasing Dimensions

Xiang Xu, Guang-Ren Duan

2024, 11(12): 2451-2462. doi: 10.1109/JAS.2024.124599

Abstract(58) HTML (6) PDF(22)

Abstract:
This paper mainly addresses control problems of strict-feedback systems (SFSs) with increasing dimensions. Compared with the commonly-considered SFSs where the subsystems have the same dimension, we aim to handle more complex cases, i.e., the subsystems in the considered SFSs are assumed to have increasing dimensions. By transforming the systems into high-order fully-actuated system (HOFAS) models, the stabilizing controllers can be directly given. Besides first-order SFSs, second-order and high-order SFSs are also considered.

Human Observation-Inspired Universal Image Acquisition Paradigm Integrating Multi-Objective Motion Planning and Control for Robotics

Haotian Liu, Yuchuang Tong, Zhengtao Zhang

2024, 11(12): 2463-2475. doi: 10.1109/JAS.2024.124512

Abstract(42) HTML (5) PDF(8)

Abstract:
Image acquisition stands as a prerequisite for scrutinizing surfaces inspection in industrial high-end manufacturing. Current imaging systems often exhibit inflexibility, being confined to specific objects and encountering difficulties with diverse industrial structures lacking standardized computer-aided design (CAD) models or in instances of deformation. Inspired by the multidimensional observation of humans, our study introduces a universal image acquisition paradigm tailored for robotics, seamlessly integrating multi-objective optimization trajectory planning and control scheme to harness measured point clouds for versatile, efficient, and highly accurate image acquisition across diverse structures and scenarios. Specifically, we introduce an energy-based adaptive trajectory optimization (EBATO) method that combines deformation and deviation with dual-threshold optimization and adaptive weight adjustment to improve the smoothness and accuracy of imaging trajectory and posture. Additionally, a multi-optimization control scheme based on a meta-heuristic beetle antennal olfactory recurrent neural network (BAORNN) is proposed to track the imaging trajectory while addressing posture, obstacle avoidance, and physical constraints in industrial scenarios. Simulations, real-world experiments, and comparisons demonstrate the effectiveness and practicality of the proposed paradigm.

Controllability of Multi-Relational Networks With Heterogeneous Dynamical Nodes

Lifu Wang, Zhaofei Li, Lianqian Cao, Ge Guo, Zhi Kong

2024, 11(12): 2476-2486. doi: 10.1109/JAS.2024.124404

Abstract(57) HTML (8) PDF(13)

Abstract:
This paper studies the controllability of networked systems, in which the nodes are heterogeneous high-dimensional dynamical systems, and the links between nodes are multi-relational. Our aim is to find controllability criteria for heterogeneous networks with multi-relational links beyond those only applicable to networks with single-relational links. It is found a network with multi-relational links can be controllable even if each single-relational network topology is uncontrollable, and vice versa. Some sufficient and necessary conditions are derived for the controllability of multi-relational networks with heterogeneous dynamical nodes. For two typical multi-relational networks with star-chain topology and star-circle topology, some easily verified conditions are presented. For illustration and verification, several examples are presented. These findings provide practical insights for the analysis and control of multi-relational complex systems.

High-Order Control Barrier Function-Based Safety Control of Constrained Robotic Systems: An Augmented Dynamics Approach

Haijing Wang, Jinzhu Peng, Fangfang Zhang, Yaonan Wang

2024, 11(12): 2487-2496. doi: 10.1109/JAS.2024.124524

Abstract(132) HTML (5) PDF(49)

Abstract:
Although constraint satisfaction approaches have achieved fruitful results, system states may lose their smoothness and there may be undesired chattering of control inputs due to switching characteristics. Furthermore, it remains a challenge when there are additional constraints on control torques of robotic systems. In this article, we propose a novel high-order control barrier function (HoCBF)-based safety control method for robotic systems subject to input-output constraints, which can maintain the desired smoothness of system states and reduce undesired chattering vibration in the control torque. In our design, augmented dynamics are introduced into the HoCBF by constructing its output as the control input of the robotic system, so that the constraint satisfaction is facilitated by HoCBFs and the smoothness of system states is maintained by the augmented dynamics. This proposed scheme leads to the quadratic program (QP), which is more user-friendly in implementation since the constraint satisfaction control design is implemented as an add-on to an existing tracking control law. The proposed closed-loop control system not only achieves the requirements of real-time capability, stability, safety and compliance, but also reduces undesired chattering of control inputs. Finally, the effectiveness of the proposed control scheme is verified by simulations and experiments on robotic manipulators.

Robust Offline Actor-Critic With On-policy Regularized Policy Evaluation

Shuo Cao, Xuesong Wang, Yuhu Cheng

2024, 11(12): 2497-2511. doi: 10.1109/JAS.2024.124494

Abstract(55) HTML (6) PDF(19)

Abstract:
To alleviate the extrapolation error and instability inherent in Q-function directly learned by off-policy Q-learning (QL-style) on static datasets, this article utilizes the on-policy state-action-reward-state-action (SARSA-style) to develop an offline reinforcement learning (RL) method termed robust offline Actor-Critic with on-policy regularized policy evaluation (OPRAC). With the help of SARSA-style bootstrap actions, a conservative on-policy Q-function and a penalty term for matching the on-policy and off-policy actions are jointly constructed to regularize the optimal Q-function of off-policy QL-style. This naturally equips the off-policy QL-style policy evaluation with the intrinsic pessimistic conservatism of on-policy SARSA-style, thus facilitating the acquisition of stable estimated Q-function. Even with limited data sampling errors, the convergence of Q-function learned by OPRAC and the controllability of bias upper bound between the learned Q-function and its true Q-value can be theoretically guaranteed. In addition, the sub-optimality of learned optimal policy merely stems from sampling errors. Experiments on the well-known D4RL Gym-MuJoCo benchmark demonstrate that OPRAC can rapidly learn robust and effective task-solving policies owing to the stable estimate of Q-value, outperforming state-of-the-art offline RLs by at least 15%.