Li Xu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Li Xu has a total of 82 co-authors. Here are the top co-authors by number of shared publications:
- Guan Gui
Nanjing University of Posts and Telecommunications
12 shared publications - Dongdong Zhao
Lanzhou University
11 shared publications - Shi Yan
School of Information Science and Engineering, Lanzhou University, Lanzhou 730000, China
8 shared publications Zhiping Lin
Nanyang Technological University
5 shared publications- Shin-Ya Matsushita
Akita Prefectural University
4 shared publications Huiling Xu
School of Science, Nanjing University of Science and Technology, Nanjing, People's Republic of China
3 shared publications
Latest Publications
- 6 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (2) , 3059-3070
Kolmogorov–Arnold Transformer Koopman Modeling Approach to Unknown Nonlinear Robotic Systems
This article presents a Kolmogorov–Arnold transformer Koopman modeling (KATKM) approach for unknown nonlinear robotic systems. Specifically, a Kolmogorov–Arnold transformer (KAT) is developed by introducing the Kolmogorov–Arnold network (KAN) into the Transformer encoder to efficiently capture spatial dynamic dependencies among system states. In addition, a KAN Koopman predictor is developed to directly learn Koopman observable functions by parameterizing B-spline activation functions. Experimental evaluations on robotic manipulators show that KATKM outperforms state-of-the-art Koopman-based methods in prediction accuracy, trajectory tracking, and robustness against measurement noise.
- 6 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (2) , 3059-3070
Kolmogorov–Arnold Transformer Koopman Modeling Approach to Unknown Nonlinear Robotic Systems
This article presents a Kolmogorov–Arnold transformer Koopman modeling (KATKM) approach for unknown nonlinear robotic systems. Specifically, a Kolmogorov–Arnold transformer (KAT) is developed by introducing the Kolmogorov–Arnold network (KAN) into the Transformer encoder to efficiently capture spatial dynamic dependencies among system states. In addition, a KAN Koopman predictor is developed to directly learn Koopman observable functions by parameterizing B-spline activation functions. Experimental evaluations on robotic manipulators show that KATKM outperforms state-of-the-art Koopman-based methods in prediction accuracy, trajectory tracking, and robustness against measurement noise.
- 19 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 71 (12) , 16047-16056
A Kalman-Koopman LQR Control Approach to Robotic Systems
This article presents a Kalman–Koopman linear quadratic regulator (KKLQR) control approach to robotic systems. In the proposed approach, an optimal Koopman modeling method based on neural networks, in which continuous Koopman eigenfunctions are constructed without requiring any predefined dictionary, is proposed to obtain approximated linear models with high precision for robotic systems. Specifically, the linear model is constructed through a multistep prediction error minimization, which enables a long-term prediction capability. Furthermore, the Kalman filter is employed to alleviate the effects of disturbances in the KKLQR control approach. Experimental results show that the proposed KKLQR control approach achieves better prediction and control performance than other existing representative methods.
- 19 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 71 (12) , 16047-16056
A Kalman-Koopman LQR Control Approach to Robotic Systems
This article presents a Kalman–Koopman linear quadratic regulator (KKLQR) control approach to robotic systems. In the proposed approach, an optimal Koopman modeling method based on neural networks, in which continuous Koopman eigenfunctions are constructed without requiring any predefined dictionary, is proposed to obtain approximated linear models with high precision for robotic systems. Specifically, the linear model is constructed through a multistep prediction error minimization, which enables a long-term prediction capability. Furthermore, the Kalman filter is employed to alleviate the effects of disturbances in the KKLQR control approach. Experimental results show that the proposed KKLQR control approach achieves better prediction and control performance than other existing representative methods.
- 18 April 2024
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 361 (8)
A vector decomposition approach to F–M model realization and its application to robust observed-state feedback control
- 18 April 2024
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 361 (8)
A vector decomposition approach to F–M model realization and its application to robust observed-state feedback control
- 20 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automatic Control
- Vol. 69 (10) , 6821-6835
Constrained Common Invariant Subspace and Its Application
The notion of constrained common invariant subspaces (CCISs) is proposed in this paper as a generalization of the well-known invariant subspace to study the structural properties of multiple matrices. Specifically, some necessary and sufficient conditions for the existence of a CCIS are established to provide a methodology to compute such a CCIS. Then, the properties of CCISs and their relation to common eigenvectors are revealed. The existence of common eigenvectors leads to the existence of CCIS, but not vice versa, so the established CCIS can reveal the structural properties of multiple matrices better than common eigenvectors can. The established CCIS is applied to the reducibility of Fornasini-Marchesini (F-M) state-space models, i.e., the necessary and sufficient conditions and the related algorithm for reducibility of F-M models are developed. Finally, a gain-scheduled state-feedback control is proposed for a rational parameter system to further demonstrate the superiority of the established CCIS.
- 20 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Automatic Control
- Vol. 69 (10) , 6821-6835
Constrained Common Invariant Subspace and Its Application
The notion of constrained common invariant subspaces (CCISs) is proposed in this paper as a generalization of the well-known invariant subspace to study the structural properties of multiple matrices. Specifically, some necessary and sufficient conditions for the existence of a CCIS are established to provide a methodology to compute such a CCIS. Then, the properties of CCISs and their relation to common eigenvectors are revealed. The existence of common eigenvectors leads to the existence of CCIS, but not vice versa, so the established CCIS can reveal the structural properties of multiple matrices better than common eigenvectors can. The established CCIS is applied to the reducibility of Fornasini-Marchesini (F-M) state-space models, i.e., the necessary and sufficient conditions and the related algorithm for reducibility of F-M models are developed. Finally, a gain-scheduled state-feedback control is proposed for a rational parameter system to further demonstrate the superiority of the established CCIS.
- 20 October 2023
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 360 (18) , 14261-14278
State-space model realization for non-commensurate fractional-order systems based on Gleason’s problem
- 20 October 2023
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 360 (18) , 14261-14278
State-space model realization for non-commensurate fractional-order systems based on Gleason’s problem
- 11 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 70 (9) , 3639-3652
FPGA Implementation of Non-Commensurate Fractional-Order State-Space Models
In order to obtain a general-purpose real-time simulator that can handle actual analog input and output for a linear time-invariant (LTI) fractional-order system, and also be suitable for MIMO and non-zero initial-state start-up LTI fractional-order systems, this paper presents a method for designing a real-time hardware simulator of non-commensurate fractional-order state-space model (NCFO-SSM) based on FPGA. The numerical approximation of NCFO-SSM for FPGA implementation is given to enhance computing efficiency and reduce the computational complexity of the NCFO-SSM systems. The FPGA implementation of NCFO-SSM is then established to achieve the hardware simulator design of a general purpose for LTI fractional-order systems. To further balance the relationship between the accuracy and the occupancy of FPGA logic resources, a fractional-order Gr $\ddot{\text{u}}$ nwald-Letnikov (GL) operator is proposed based on the piecewise quadratic equation fitting (PWQEF) method. Applications of the proposed PWQEF operator in the NCFO-SSM hardware simulator show that the new hardware simulator can achieve the closest results to theoretical computation. The proposed NCFO-SSM hardware simulator is used to implement fractional-order systems, including a fractional-order PID controller system, an MIMO system, and an H-bridge current source PWM regulator system, to further illustrate its application scope and practicability in LTI fractional-order systems.
- 11 July 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems I: Regular Papers
- Vol. 70 (9) , 3639-3652
FPGA Implementation of Non-Commensurate Fractional-Order State-Space Models
In order to obtain a general-purpose real-time simulator that can handle actual analog input and output for a linear time-invariant (LTI) fractional-order system, and also be suitable for MIMO and non-zero initial-state start-up LTI fractional-order systems, this paper presents a method for designing a real-time hardware simulator of non-commensurate fractional-order state-space model (NCFO-SSM) based on FPGA. The numerical approximation of NCFO-SSM for FPGA implementation is given to enhance computing efficiency and reduce the computational complexity of the NCFO-SSM systems. The FPGA implementation of NCFO-SSM is then established to achieve the hardware simulator design of a general purpose for LTI fractional-order systems. To further balance the relationship between the accuracy and the occupancy of FPGA logic resources, a fractional-order Gr $\ddot{\text{u}}$ nwald-Letnikov (GL) operator is proposed based on the piecewise quadratic equation fitting (PWQEF) method. Applications of the proposed PWQEF operator in the NCFO-SSM hardware simulator show that the new hardware simulator can achieve the closest results to theoretical computation. The proposed NCFO-SSM hardware simulator is used to implement fractional-order systems, including a fractional-order PID controller system, an MIMO system, and an H-bridge current source PWM regulator system, to further illustrate its application scope and practicability in LTI fractional-order systems.
- 13 May 2022
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 359 (10) , 5014-5035
A digraph approach to the state-space model realization of MIMO non-commensurate fractional order systems
- 13 May 2022
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 359 (10) , 5014-5035
A digraph approach to the state-space model realization of MIMO non-commensurate fractional order systems
- 1 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 69 (12) , 16183-16187
DOA Estimation With Small Snapshots Using Weighted Mixed Norm Based on Spatial Filter
- 1 December 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 69 (12) , 16183-16187
DOA Estimation With Small Snapshots Using Weighted Mixed Norm Based on Spatial Filter
- 16 November 2020
- Published by Wiley in Asian Journal of Control
- Vol. 24 (1) , 282-296
Distributed finite‐time control for systems interconnected over an undirected graph
In this paper, the distributed finite‐time control problem for the class of systems interconnected over an undirected graph is considered. First, the concepts of well‐posedness, finite‐time boundedness, and contractiveness for the considered systems are given. Then, a sufficient condition is developed to ensure that the plant is well‐posed, finite‐time bounded, and contractive, where the results are stated by terms of linear matrix inequalities (LMIs). For the control synthesis, we construct a distributed dynamic output‐feedback controller inheriting the structure of the plant such that the resulting closed‐loop system is finite‐time bounded and contractive. Since the analysis condition of the plant used for the closed‐loop system is presented by some nonlinear inequalities in given finite‐time interval, we further derive a sufficient condition in terms of LMIs with some fixed parameter for the existence of an appropriate output‐feedback controller such that the closed‐loop system is contractive. Finally, numerical simulations are provided to show the validity of the proposed results.
- 16 November 2020
- Published by Wiley in Asian Journal of Control
- Vol. 24 (1) , 282-296
Distributed finite‐time control for systems interconnected over an undirected graph
In this paper, the distributed finite‐time control problem for the class of systems interconnected over an undirected graph is considered. First, the concepts of well‐posedness, finite‐time boundedness, and contractiveness for the considered systems are given. Then, a sufficient condition is developed to ensure that the plant is well‐posed, finite‐time bounded, and contractive, where the results are stated by terms of linear matrix inequalities (LMIs). For the control synthesis, we construct a distributed dynamic output‐feedback controller inheriting the structure of the plant such that the resulting closed‐loop system is finite‐time bounded and contractive. Since the analysis condition of the plant used for the closed‐loop system is presented by some nonlinear inequalities in given finite‐time interval, we further derive a sufficient condition in terms of LMIs with some fixed parameter for the existence of an appropriate output‐feedback controller such that the closed‐loop system is contractive. Finally, numerical simulations are provided to show the validity of the proposed results.
- 1 February 2020
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 357 (3) , 1764-1789
A novel constructive procedure to low-order Fornasini–Marchesini model realization
- 1 February 2020
- Published by Elsevier in Journal of the Franklin Institute
- Vol. 357 (3) , 1764-1789
A novel constructive procedure to low-order Fornasini–Marchesini model realization