IEEE/CAA Journal of Automatica Sinica

Multi-Agent Systems: From Classical Paradigms to Large Foundation Model-Enabled Futures

Zixiang Wang, Mengjia Gong, Qiyu Sun, Jing Xu, Shuai Mao, Xin Jin, Qing-Long Han, Yang Tang

2026, 13(5): 1007-1023. doi: 10.1109/JAS.2026.126113

Abstract(137) HTML (58) PDF(23)

Abstract:
With the rapid advancement of artificial intelligence, multi-agent systems (MASs) are evolving from classical paradigms toward architectures built upon large foundation models (LFMs). This survey provides a systematic review and comparative analysis of classical MASs (CMASs) and LFM-based MASs (LMASs). First, within a closed-loop coordination framework, CMASs are reviewed across four fundamental dimensions: perception, communication, decision-making, and control. Beyond this framework, LMASs integrate LFMs to lift collaboration from low-level state exchanges to semantic-level reasoning, enabling more flexible coordination and improved adaptability across diverse scenarios. Then, a comparative analysis is conducted to contrast CMASs and LMASs across architecture, operating mechanism, adaptability, and application. Finally, future perspectives on MASs are presented, summarizing open challenges and potential research opportunities.

Event-Triggered Impulsive Control: A Survey

Xiaodi Li

2026, 13(5): 1024-1040. doi: 10.1109/JAS.2026.125930

Abstract(106) HTML (52) PDF(19)

Abstract:
An event-triggered impulsive control (ETIC) approach has attracted considerable attention in both theoretical and practical fields, as it ensures that control resources are deployed solely at moments of strict necessity, thereby significantly reducing the control frequency without compromising system performance. This survey aims to provide a timely, structured, and in-depth overview of recent advances in ETIC. First, a brief introduction of impulsive control and ETIC is given, analysing the advantages of ETIC over the traditional time-triggered impulsive control. Second, the representative results and methodologies reported in ETIC are generally classified from the perspective of structural characteristics of event-triggering mechanisms (ETMs), and the fundamental principles underlying each type of ETM are elaborated in detail. Third, the common methods used to exclude the Zeno behavior in recent ETIC results are presented in detail. Then, an in-depth analysis is made on several categories of theoretical results on the stability problem based on ETIC. Fourth, applications of ETIC in practical scenarios, including motor servo systems, chemical reactor systems, and wheeled robots, are provided. Finally, some challenging issues on ETIC are proposed to guide future research.

Perceiving the Battery Multi-Electrochemical States in Real-time Based on Model-Informed Neural Network

Ying Zhang, Qinghao Zhang, Bin Duan, Pingwei Gu, Changlong Li, Chenghui Zhang

2026, 13(5): 1041-1053. doi: 10.1109/JAS.2026.125945

Abstract(75) HTML (52) PDF(9)

Abstract:
Accurate estimation of electrochemical states serves as a pathway to observe internal battery behaviors, effectively bridging the gap between micro mechanism and macro performance and enabling more precise control in an advanced battery management system. Yet conventional pseudo-two-dimensional (P2D) physics methods suffer from high computational complexity and limit their online application. Thus, we develop a model-informed neural network (MINN) framework that synergistically combines deep learning with a physics-based model to accurately monitor the battery electrochemical state (such as lithium-ion concentration, plating potential). Firstly, the MINN model is constructed with the innovative loss term containing experimentally measurable parameters and governing physical laws. Secondly, a composite framework based on a convolutional neural network (CNN) architecture is integrated to automatically extract features and enforce spatial boundary conditions, which significantly reduces the number of boundary loss terms that need to be solved and alleviates the complexity of the training process. After training, the MINN model can achieve an accurate estimation of internal states and even their spatiotemporal distributions that cannot be directly measured based on limited observable data and physical laws. At last, by incorporating dynamic current input, the well-trained basic model exhibits strong robustness and can be directly transferred to other cycling protocols with high accuracy, requiring no further retraining. MINN is a novel and promising framework to realize online and accurate micro electrochemical states monitoring, achieving at least 776 times speedup compared with the P2D model. As an innovative artificial intelligence assisted modeling for electrochemical systems, this framework enables root-cause analysis of battery behavior and failure modes, while empowering the management system with more reliable and trustworthy decision-making capabilities.

A New Knowledge Mining and Root Cause Analysis Methodology for Multivariate Time Series

Xiaoliang Wang, Faming Lu, MengChu Zhou, Qingtian Zeng

2026, 13(5): 1054-1067. doi: 10.1109/JAS.2026.125837

Abstract(216) HTML (47) PDF(25)

Abstract:
Root cause analysis (RCA) aims to discover the root causes of abnormal events. Causal relations reveal the evolution process of abnormal events, which plays a crucial role in RCA. However, existing methods neither explicitly emphasize the “AND/OR” relations among causes, nor consider the synergy effects owned by non-causal variables on causal rules, thereby affecting the credibility of RCA. To address the issues, by fusing Petri nets and Bayesian networks, this study proposes a new knowledge mining and RCA methodology for multivariate time series, called synergy-incorporated Bayesian time Petri net. It integrates the advantages of Petri nets in modeling and analyzing complex temporal dependencies and Bayesian networks in evidence reasoning. It takes into account “AND/OR” relations and synergy effects in temporal knowledge mining and RCA. Two cases are employed to verify its performance in knowledge mining and RCA, including a case study of quality anomaly detection of solar panels and the Tennessee-Eastman process. The experimental results from both cases indicate that it can effectively consider “AND/OR” relations and synergy effects. When applied to the former, it outperforms the state-of-the-art RCA methods in accuracy by over 11%.

Gain-Based Neural Secure Protection Control for Feedforward Nonlinear Systems With Unknown Control Coefficients and Impulsive FDI Attacks

Debao Fan, Qingrong Liu, Rong Su, Xianfu Zhang, Wenjie Zhang

2026, 13(5): 1068-1081. doi: 10.1109/JAS.2025.125807

Abstract(588) HTML (44) PDF(40)

Abstract:
This paper proposes a gain-based neural secure protection (GBNSP) control scheme for feedforward nonlinear systems subject to unknown control coefficients and impulsive false data injection (FDI) attacks. Notably, the nonlinear functions of the systems are relaxed to any continuous functions and the control coefficients are permitted to be constants with both unknown sizes and signs, a scenario not covered in existing works. Furthermore, the uncertain abrupt changes in system states caused by impulsive FDI attacks inevitably exacerbate the challenges in control design. To this end, this paper integrates the neural network technique and the gain control method to propose a novel GBNSP control scheme. Specifically, the neural network technique effectively compensates for strong nonlinearities and uncertainties, while the gain control method quantifies the tolerable frequency of impulsive FDI attacks and avoids the tedious design procedures. It is shown that, under the designed GBNSP controller, all closed-loop signals remain bounded and the system states eventually converge to an adjustable neighborhood near the origin. Moreover, an enhanced GBNSP control scheme incorporates an improved gain scaling mechanism to withstand unknown external disturbances. In the end, the effectiveness and practicality of the proposed scheme are validated by a theoretical example and a practical example.

Toward Resilient Vehicle Platooning: A Two-Layer Secure Control Architecture Against Hybrid Cyber-Physical Threats

Jian Gong, Lei Ding, Chengfeng Jia, Yutian Liu, Jinde Cao

2026, 13(5): 1082-1096. doi: 10.1109/JAS.2026.125909

Abstract(477) HTML (50) PDF(49)

Abstract:
This paper presents a hierarchical secure control framework for resilient vehicular platooning under hybrid cyber-physical threats, including coupled false data injection (FDI) and denial-of-service (DoS) attacks, as well as actuator faults. A two-layer architecture is adopted to decouple cyber-layer disruptions from physical-layer execution, thereby enhancing system modularity and fault isolation. At the upper layer, a virtual platoon system is constructed, where a distributed resilient controller integrated with an event-triggered mechanism (ETM) is developed to ensure coordinated behavior while reducing communication overhead. At the lower layer, an adaptive fault-tolerant tracking controller is designed to compensate for actuator degradation and external disturbances, enabling each physical vehicle to follow its virtual reference independently. A layer-wise Lyapunov-based analysis is conducted to guarantee the practical exponential stability of the hierarchical control framework, where tractable LMI conditions are derived for both the cyber coordination and physical tracking components. Simulation results demonstrate that the proposed architecture effectively mitigates fault propagation, maintains robust performance under concurrent cyber and physical threats, and outperforms non-hierarchical benchmarks in terms of system stability.

Decentralized Prescribed-Time Output-Feedback Control for Interconnected Systems With Coupled Non-Identical Subsystems of Different Orders

Yan Tan, Hefu Ye, Changyun Wen, Yongduan Song

2026, 13(5): 1097-1107. doi: 10.1109/JAS.2025.125918

Abstract(81) HTML (33) PDF(9)

Abstract:
The challenge of decentralized control arises from the necessity to use only local information for constructing feedback mechanisms, with the goal of suppressing interactions among subsystems while ensuring the overall stability and performance of the entire system. The problem is significantly complicated when interconnected subsystems possess non-identical nonlinear dynamics with arbitrary relative degrees. We propose a novel time-varying decentralized control architecture that achieves global prescribed-time stabilization for the overall system through local output feedback. The developed scheme guarantees convergence to the equilibrium within any user-defined time interval, with temporal performance decoupled from initial conditions and independent of other controller parameter selection. Firstly, the decentralized prescribed-time observers for subsystems are systematically constructed to reconstruct full-state information, forming the basis for output-feedback synthesis. Secondly, a decentralized matrix pencil framework is established by incorporating symmetric positive-definite solutions of parametric Lyapunov equations, enabling concurrent management of intra-subsystem state/observer error dynamics and inter-subsystem couplings. This strategy facilitates the derivation of low-conservative design parameters while enhancing the applicability of the control algorithm. Finally, our findings demonstrate that the proposed approach is capable of accommodating a general system model characterized by unknown interaction strengths and arbitrary relative degrees. The theoretical results are validated through two simulation examples, confirming the effectiveness of the developed control scheme.

An EMS Research and Development Fast Prototyping Platform for Disaster Relief

Zhiyu Long, Mo-Yuen Chow

2026, 13(5): 1108-1121. doi: 10.1109/JAS.2025.125831

Abstract(66) HTML (40) PDF(1)

Abstract:
Power restoration after disasters is of utmost importance, due to its critical role in supporting disaster relief, medical care, transportation, communications and other essential services in disaster areas. However, post-disaster power restoration is a time-sensitive energy management problem, and it will be a challenge for conventional methods to ensure power restoration in a rapid and stable period of time for areas with limited resources and complex environmental conditions. This paper presents an energy management system (EMS) research and development (R&D) fast prototyping platform designed for disaster relief. Designed for microgrids in disaster areas, the platform enables rapid deployment of microgrid prototype and offers energy management capabilities to supply emergency power. The platform integrates both software and hardware-in-the-loop (HIL) system to form a closed-loop R&D process. The software part has a modular design of EMS components and functions, allowing for flexible combinations and convenient user configuration based on various application needs. The HIL system part provides a dedicated testing and verification solution for EMS modules through the networking of multiple devices. To verify the fast prototyping capability of the platform, simulation validation is conducted using several microgrid cases with dynamically changing environments and user requirements in the disaster scenario. The results indicate that these design features enable the platform to conveniently configure microgrid rapid prototypes for case analysis and proof-of-concept validation in a modular manner, making it well-suited for EMS applications in time-sensitive and dynamic disaster scenarios.

Current-State Opacity Verification in Petri Nets Using Fusion-Based Opacity Analysis and Graph Isomorphism Networks

Sichen Ding, Zhiwu Li

2026, 13(5): 1122-1134. doi: 10.1109/JAS.2025.125936

Abstract(76) HTML (55) PDF(3)

Abstract:
Current-state opacity is a critical security property for discrete event systems, but its verification in large-scale Petri nets is hampered by the state-space explosion problem. To address this challenge, we propose the fusion-based opacity analysis and graph isomorphism network (FOA-GIN), a novel deep learning framework. The method transforms current-state opacity verification into a graph classification task by applying a tailored graph isomorphism network to basis reachability graphs—a compact representation of the system’s dynamics. This approach integrates the theoretical strengths of basis reachability graphs with the scalability of graph neural networks to capture essential structural and behavioral features for opacity analysis. Unlike classical algorithms that require exponential state-space traversal, the proposed model’s online verification complexity is linear in the size of the input basis reachability graph. Experiments demonstrate high accuracy and robustness, establishing FOA-GIN as a powerful and practical solution for verifying current-state opacity in complex, large-scale systems.

Intelligent Safe Optimal Control Towards Koopman Operator-Driven Nonlinear Systems With Asymmetric State and Input Constraints

Yalu Su, Ding Wang, Mingming Zhao, Dan Xiong, Yiyong Huang, Wei Han

2026, 13(5): 1135-1150. doi: 10.1109/JAS.2025.125945

Abstract(122) HTML (55) PDF(14)

Abstract:
For unknown nonlinear systems subject to asymmetric state and input constraints simultaneously, this article establishes a safe value iteration paradigm to learn an optimal control policy in a data-based manner. Initially, the Koopman operator, instead of the black-box neural network, is applied to extract the inherent dynamics of the controlled systems from the measured data, thereby allowing for explicit analysis of the prediction error. To tackle the issue posed by state and input constraints, a crafted control barrier function is seamlessly incorporated into the canonical utility function, which retains the property of positive definiteness for the asymmetric case. Moreover, the value iteration algorithm with regard to the augmented utility function is adopted to attain a safe optimal controller, where the actor and critic networks are leveraged to approximate the control input and associated value function, respectively. The monotonicity, safety, and stability of the raised algorithm are further verified rigorously. Via performing three experiments on the linear system, the nonlinear system, and the manipulator plant, comparative results are obtained to substantiate the superiority and efficacy of the developed approach in achieving optimal performance and safe guarantee.

Temporal Formation Control for Multi-agent Systems Based on Reachable Sets

Yuhua Yao, Jitao Sun, Xiaoming Hu

2026, 13(5): 1151-1165. doi: 10.1109/JAS.2025.125960

Abstract(95) HTML (56) PDF(6)

Abstract:
This paper explores formation control for multi-agent systems by reformulating desired formations using signal temporal logic (STL) specifications. To achieve flexibility and efficiency, we employ sparse polynomial zonotopes (SPZs) to represent several common formations and the system’s state sets. This representation allows us to frame the formation control problem as a series of transitions between different state sets within a specific time horizon, which can be solved using optimal transport theory. By combining the optimal transport matrix with shrinking horizon model predictive control (MPC), we have developed a feasible control strategy that gradually guides the system trajectories toward the target set. This innovative approach decomposes the formation control problem into multiple temporal logic subproblems, reducing computational complexity and mitigating the impact of sampling randomness introduced by optimal transport. The effectiveness of our proposed approach is demonstrated through its application to a multi-unicycle system.

Adaptive Prescribed-Time Exact Tracking Control for Uncertain Strict-Feedback Systems With Global Prescribed-Performance

Yizhaotun Yan, Bing Mao, Ling Lei, Hui Liu, Xiaoqun Wu

2026, 13(5): 1166-1175. doi: 10.1109/JAS.2025.125966

Abstract(79) HTML (56) PDF(15)

Abstract:
For uncertain strict-feedback systems under the prescribed performance control (PPC) problem, an innovative adaptive prescribed-time tracking control method is proposed. This method combines a novel error transformation function with the prescribed-time stability theory, thereby achieving exact tracking of desired trajectories within a prescribed time while ensuring that the tracking error stays within predefined boundaries globally. By integrating a newly-designed Lyapunov-like energy function with dynamic surface control, it resolves the error surface issues that result in the semi-global boundedness of tracking error in traditional approaches. Furthermore, through a generalized Filippov solution definition, this approach overcomes the issue of non-existence of the system solution, which arises during the prescribed-time stability analysis due to the discontinuous control input. Simulation results validate the effectiveness of the proposed method.

Global Singularity-Free Prescribed Performance Control for Faulty Nonlinear Systems via Parameterized Nonmonotonic Constraints

Yu Xia, Jun He, Zsófia Lendek, Imre J. Rudas, Ramesh K. Agarwal

2026, 13(5): 1176-1183. doi: 10.1109/JAS.2026.125744

Abstract(75) HTML (51) PDF(7)

Abstract:
In this paper, we investigate potential singularity issues in prescribed performance control induced by actuator faults and propose a global nonmonotonic prescribed performance scheme for strict-feedback nonlinear systems. This scheme allows for parameterized relaxation of constraint boundaries in the presence of actuator faults. Unlike existing prescribed performance control schemes that rely on redesigning nonmonotonic rate functions to modify boundary relaxation properties, the proposed method introduces a novel adjustment module into the prescribed performance function, enabling the parametric design of nonmonotonic boundaries. Moreover, the proposed method simultaneously addresses the removal of the initial feasibility condition and the imposition of asymmetric constraints by introducing a global asymmetric design with an error correction module. This module enables flexible conversion from asymmetric to symmetric boundaries at a predetermined time instant, thereby reducing transient overshoot while maintaining steady-state tracking precision. Simulation results validate the effectiveness and superiority of the proposed scheme.

A Fast Algorithm for Matrix-Variable Triconvex Optimization With Application to Blind Image Deblurring

Songchuan Zhang, Liqing Huang, Youshen Xia, Jun Wang

2026, 13(5): 1184-1206. doi: 10.1109/JAS.2026.125711

Abstract(101) HTML (42) PDF(6)

Abstract:
Matrix-variable triconvex optimization is a significant generalization of vector-variable triconvex or biconvex optimization and has been found to have popular applications. To reduce computation time and storage requirements, this paper presents a matrix-form iterative method for quickly solving matrix-variable constrained triconvex optimization problems. The proposed method is based on a matrix-form alternating projection iteration scheme in the form of matrix state spaces, where an efficient line search strategy is adopted by exploiting the optimality conditions of the problem for a larger step length. Compared with the existing vector-form alternating projection gradient method, the proposed method reduces storage requirements and computational cost, and thus is more computationally efficient. Each sequence generated by the proposed method is guaranteed to be globally convergent to a partial optimum under mild conditions. Finally, the proposed method is effectively applied to blind image deblurring problems. Computed results show that the proposed algorithm is superior to related iterative algorithms in terms of computation time and solution quality.

A Novel Model Free Adaptive Fuzzy Control for Discrete-Time T-S Fuzzy Systems With Local Nonlinear Models

Xiaodong Bu, Shangtai Jin, Xisheng Dai, Zhongsheng Hou

2026, 13(5): 1207-1216. doi: 10.1109/JAS.2025.125810

Abstract(75) HTML (49) PDF(4)

Abstract:
This paper presents a model-free adaptive fuzzy control (MFAFC) scheme for discrete-time Takagi-Sugeno (T-S) fuzzy systems with local nonlinear models. First, the T-S fuzzy system is transformed into a linearized model using a dynamic linearization technique. Then, a model-free adaptive control scheme is developed for T-S fuzzy systems. Next, a rigorous convergence analysis of the tracking error is carried out using the contraction mapping theory. Finally, to validate the theoretical results, the scheme is tested by two numerical simulations and a mass-spring damper mechanical system. The results show that the proposed MFAFC strategy is effective in ensuring that the system output tracks the desired trajectory.

State Estimation for 2-D Markov Jump Systems With Multiple Observation Delays Over Decoded-and-Forward Relay Channels

Yufan Wang, Chunyan Han, Wei Wang, Tao Shen

2026, 13(5): 1217-1235. doi: 10.1109/JAS.2026.125753

Abstract(84) HTML (44) PDF(10)

Abstract:
This paper addresses the recursive state estimation problem for two-dimensional Markov jump systems (2-D MJSs) subject to multi-channel observation delays and packet losses under the decoded-and-forward (DaF) relay-based protocol. To enhance signal propagation distance and quality, the DaF relay-based policy is deployed to schedule data transmission over the sensor-to-estimator channels. An encoding-decoding mechanism related to the jumping parameter is introduced to transform the initial measurements into delayed decoded observations. The packet dropouts are considered to occur in the relay-to-estimator channels and are modeled as mutually uncorrelated Bernoulli random variables. Moreover, on grounds of 2-D reorganization observation technology, the delayed decoded signals transmitted through DaF relay channels are transformed into delay-free decoded measurements. The goal of this paper is to develop a recursive state estimator based on the reconstructed delay-free model, aiming to minimize the upper bounds of the filtering error variances. By applying matrix inequality theory and the inductive method, specific upper bounds on the filtering error covariances are derived. Then, the expected filter gain parameters are designed to minimize these bounds by solving a series of 2-D Riccati difference equations, and the boundedness performance of the proposed recursive filter algorithm is investigated. Finally, a simulation example is provided to demonstrate the effectiveness of the devised filter strategy.