Shusen Yuan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Shusen Yuan has a total of 25 co-authors. Here are the top co-authors by number of shared publications:
- Guolai Yang
National Key Laboratory of Transient Physics, Nanjing University of Science and Technology, Nanjing, China
7 shared publications - Jianyong Yao
School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
5 shared publications - Wenxiang Deng
School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
5 shared publications - Jun Guan
School of Automation, Jiangsu University of Science and Technology, Zhenjiang 212100, China
4 shared publications - Wenjun Yi
National Key Lab of Transient Physics, Nanjing University of Science and Technology, Nanjing 210094, China
3 shared publications - Yimin Wang
School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, China
3 shared publications
Latest Publications
- 12 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (7) , 12920-12933
Localization of Hostile Launch Points and Rapid Inversion of Friendly Firing Conditions Based on Deep Learning and Ensemble Optimization
In modern military operations, the localization of enemy launch points and the rapid inversion of friendly firing conditions are critical for enhancing response speed and the effectiveness of fire strikes. To improve the accuracy and efficiency of ballistic trajectory prediction and firing condition inversion, this article proposes an intelligent prediction framework that fuses deep learning with ensemble optimization. First, a temporal convolutional network (TCN)-bidirectional gated recurrent unit (BiGRU)-Attention model is designed for the localization of enemy launch points. Subsequently, a particle swarm optimization-extreme learning machine-adaptive boosting (PSO-ELM-AdaBoost) ensemble algorithm is adopted to rapidly derive the optimal firing conditions for striking the identified enemy launch points. Simulation results demonstrate that the TCN-BiGRU-Attention model reduces the comprehensive prediction error regarding range and lateral deviation to 7.89 m in the enemy launch point localization task. Furthermore, the PSO-ELM-AdaBoost algorithm exhibits superior performance in the firing condition inversion task; the prediction errors for the initial firing angle and initial velocity are only 0.006° and 0.058 m/s, respectively, representing accuracy improvements of 50% and 18% compared to the traditional ELM method. By constructing a dual-stage “trajectory prediction–parameter inversion” architecture, this study reduces the single-computation time to the millisecond level, significantly enhancing computational efficiency. These results indicate that the proposed framework possesses potential applicability under simulation validation conditions, providing methodological References and theoretical support for the design of sensing data fusion, real-time reasoning, and decision support modules in future C4ISR and Military Internet of Things (MIoT) systems.
- 12 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 13 (7) , 12920-12933
Localization of Hostile Launch Points and Rapid Inversion of Friendly Firing Conditions Based on Deep Learning and Ensemble Optimization
In modern military operations, the localization of enemy launch points and the rapid inversion of friendly firing conditions are critical for enhancing response speed and the effectiveness of fire strikes. To improve the accuracy and efficiency of ballistic trajectory prediction and firing condition inversion, this article proposes an intelligent prediction framework that fuses deep learning with ensemble optimization. First, a temporal convolutional network (TCN)-bidirectional gated recurrent unit (BiGRU)-Attention model is designed for the localization of enemy launch points. Subsequently, a particle swarm optimization-extreme learning machine-adaptive boosting (PSO-ELM-AdaBoost) ensemble algorithm is adopted to rapidly derive the optimal firing conditions for striking the identified enemy launch points. Simulation results demonstrate that the TCN-BiGRU-Attention model reduces the comprehensive prediction error regarding range and lateral deviation to 7.89 m in the enemy launch point localization task. Furthermore, the PSO-ELM-AdaBoost algorithm exhibits superior performance in the firing condition inversion task; the prediction errors for the initial firing angle and initial velocity are only 0.006° and 0.058 m/s, respectively, representing accuracy improvements of 50% and 18% compared to the traditional ELM method. By constructing a dual-stage “trajectory prediction–parameter inversion” architecture, this study reduces the single-computation time to the millisecond level, significantly enhancing computational efficiency. These results indicate that the proposed framework possesses potential applicability under simulation validation conditions, providing methodological References and theoretical support for the design of sensing data fusion, real-time reasoning, and decision support modules in future C4ISR and Military Internet of Things (MIoT) systems.
- 15 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 12 (23) , 50480-50492
The Data-Driven High-Precision Geomagnetic Matching-Assisted Navigation Algorithm Under Satellite Denial Conditions
In cases of satellite signal interference or even satellite denial, to improve the positioning accuracy of navigation systems, a sparrow search algorithm (SSA) is used to optimize the weights and biases of the extreme learning machine (ELM), and an adaptive boosting (AdaBoost) ensemble learning mechanism is introduced. This results in the SSA-ELM-AdaBoost geomagnetic matching-assisted navigation algorithm. In an ideal 2-D plane environment, the SSA-ELM-AdaBoost algorithm shows a clear advantage in matching accuracy compared to other algorithms [such as back propagation (BP), ELM, extreme gradient boosting (XGBoost), SSA-XGBoost, and SSA-ELM], achieving millimeter-level precision that meets high-accuracy positioning requirements. Despite the complexity of geomagnetic matching in 3-D space, the algorithm maintains high accuracy, especially in the height direction, where the precision can reach centimeter level with a maximum error of 10–20 m. In simulations of complex real-world environments where noise is added to the prediction data, the SSA-ELM-AdaBoost algorithm demonstrates strong robustness. The average matching error in the latitude and longitude directions remains within 10 m. Although occasional outliers of around 100 m may occur in the latitude direction, their probability is very low, and they can be prevented from affecting the positioning system through filtering algorithms. The algorithm performs well in computational real time, efficiently processing large amounts of sample data, meeting engineering application requirements, and demonstrating its potential in complex environments.
- 15 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Internet of Things Journal
- Vol. 12 (23) , 50480-50492
The Data-Driven High-Precision Geomagnetic Matching-Assisted Navigation Algorithm Under Satellite Denial Conditions
In cases of satellite signal interference or even satellite denial, to improve the positioning accuracy of navigation systems, a sparrow search algorithm (SSA) is used to optimize the weights and biases of the extreme learning machine (ELM), and an adaptive boosting (AdaBoost) ensemble learning mechanism is introduced. This results in the SSA-ELM-AdaBoost geomagnetic matching-assisted navigation algorithm. In an ideal 2-D plane environment, the SSA-ELM-AdaBoost algorithm shows a clear advantage in matching accuracy compared to other algorithms [such as back propagation (BP), ELM, extreme gradient boosting (XGBoost), SSA-XGBoost, and SSA-ELM], achieving millimeter-level precision that meets high-accuracy positioning requirements. Despite the complexity of geomagnetic matching in 3-D space, the algorithm maintains high accuracy, especially in the height direction, where the precision can reach centimeter level with a maximum error of 10–20 m. In simulations of complex real-world environments where noise is added to the prediction data, the SSA-ELM-AdaBoost algorithm demonstrates strong robustness. The average matching error in the latitude and longitude directions remains within 10 m. Although occasional outliers of around 100 m may occur in the latitude direction, their probability is very low, and they can be prevented from affecting the positioning system through filtering algorithms. The algorithm performs well in computational real time, efficiently processing large amounts of sample data, meeting engineering application requirements, and demonstrating its potential in complex environments.
- Open Access
Adaptive Robust Stable Tracking Control of Two-Axis Coupled Electromechanical Actuation System Based on Friction Compensation
The two-axis coupled electromechanical actuation system (TCEAS) is widely utilized in multiple industrial fields, but its tracking performance and stability are severely hampered by complex nonlinear friction, parameter uncertainties, and strong coupling effects. To address these issues, this paper proposes an adaptive robust stable tracking control (ARSTC) method with friction compensation. First, the friction characteristic of TCEAS is analyzed and a continuously differentiable friction moment function is introduced to accurately describe the nonlinear friction phenomenon. Then, a dynamic analysis model for the system considering friction nonlinearity and model uncertainty is established. Furthermore, the developed ARSTC algorithm leverages adaptive control to estimate and compensate unknown friction parameters (enhancing precision) and robust control to suppress disturbances (ensuring stability). Finally, the superiority is jointly verified by stability analysis and extensive comparative numerical test results. This work demonstrates a practical approach for high-precision control of TCEAS, which has important theoretical significance.
Open AccessAdaptive Robust Stable Tracking Control of Two-Axis Coupled Electromechanical Actuation System Based on Friction Compensation
The two-axis coupled electromechanical actuation system (TCEAS) is widely utilized in multiple industrial fields, but its tracking performance and stability are severely hampered by complex nonlinear friction, parameter uncertainties, and strong coupling effects. To address these issues, this paper proposes an adaptive robust stable tracking control (ARSTC) method with friction compensation. First, the friction characteristic of TCEAS is analyzed and a continuously differentiable friction moment function is introduced to accurately describe the nonlinear friction phenomenon. Then, a dynamic analysis model for the system considering friction nonlinearity and model uncertainty is established. Furthermore, the developed ARSTC algorithm leverages adaptive control to estimate and compensate unknown friction parameters (enhancing precision) and robust control to suppress disturbances (ensuring stability). Finally, the superiority is jointly verified by stability analysis and extensive comparative numerical test results. This work demonstrates a practical approach for high-precision control of TCEAS, which has important theoretical significance.
- Open Access
- 19 May 2025
- Published by Wiley in International Journal of Mechanical System Dynamics
- Vol. 6 (1) , 153-164
Predefined‐Time Dynamic Surface Control of Tank Horizontal Stabilization System With Disturbance
Modern new tanks are difficult to achieve high precision control of the horizontal stabilization system due to electromechanical coupling and road surface disturbance during high‐speed combat. A predefined‐time dynamic surface controller (DSCPT) is proposed for the fast stabilization and strong disturbance problems of the tank horizontal stabilization system. Based on backstepping control technique, the controller introduces a time‐varying tuning function and a first‐order filter to construct the dynamic surface error, where the time‐varying tuning function ensures that the tracking error is bounded within a predefined time, and the hyperbolic tangent function ensures robustness and the continuity of the control input. Using Lyapunov theory, it is rigorously demonstrated that the control method is capable of achieving tracking error convergence to a predetermined range at any predefined time in the presence of system uncertainty, and that it is able to ensure global stabilization of the closed‐loop system. In addition, a transfer model is used to link pavement roughness and major disturbance, and then added to the simulation. Finally, the effectiveness of the proposed control method is verified through a large number of comparative simulations.
- 19 May 2025
- Published by Wiley in International Journal of Mechanical System Dynamics
- Vol. 6 (1) , 153-164
Open AccessPredefined‐Time Dynamic Surface Control of Tank Horizontal Stabilization System With Disturbance
Modern new tanks are difficult to achieve high precision control of the horizontal stabilization system due to electromechanical coupling and road surface disturbance during high‐speed combat. A predefined‐time dynamic surface controller (DSCPT) is proposed for the fast stabilization and strong disturbance problems of the tank horizontal stabilization system. Based on backstepping control technique, the controller introduces a time‐varying tuning function and a first‐order filter to construct the dynamic surface error, where the time‐varying tuning function ensures that the tracking error is bounded within a predefined time, and the hyperbolic tangent function ensures robustness and the continuity of the control input. Using Lyapunov theory, it is rigorously demonstrated that the control method is capable of achieving tracking error convergence to a predetermined range at any predefined time in the presence of system uncertainty, and that it is able to ensure global stabilization of the closed‐loop system. In addition, a transfer model is used to link pavement roughness and major disturbance, and then added to the simulation. Finally, the effectiveness of the proposed control method is verified through a large number of comparative simulations.
- Open Access
Control-Oriented Real-Time Trajectory Planning for Heterogeneous UAV Formations
Aiming at the trajectory planning problem for heterogeneous UAV formations in complex environments, a trajectory prediction model combining Convolutional Neural Networks (CNNs) and Long Short-Term Memory networks (LSTM) is designed, and a real-time trajectory planning method is proposed based on this model. By pre-training trajectory prediction networks for various types of UAVs, the traditional physics-based models are replaced for flight trajectory prediction. Inspired by Model Predictive Control (MPC), in the trajectory planning stage, the method generates multi-step trajectory points using an improved artificial potential field (APF) method, estimates the actual formation trajectory using the prediction network, and optimizes the trajectory through a multi-objective particle swarm optimization (MOPSO) algorithm after evaluating the planning costs. During actual flight, the optimized parameters generate trajectory points for the formation to follow. Unlike conventional path planning based on simple constraints, the proposed method directly plans trajectory points based on trajectory tracking performance, ensuring high feasibility for the formation to follow. Experimental results show that the CNN-LSTM network outperforms other networks in both short-term and long-term trajectory prediction. The proposed trajectory planning method demonstrates significant advantages in formation maintenance, trajectory tracking, and real-time obstacle avoidance, ensuring flight stability and safety while maintaining high-speed flight.
Open AccessControl-Oriented Real-Time Trajectory Planning for Heterogeneous UAV Formations
Aiming at the trajectory planning problem for heterogeneous UAV formations in complex environments, a trajectory prediction model combining Convolutional Neural Networks (CNNs) and Long Short-Term Memory networks (LSTM) is designed, and a real-time trajectory planning method is proposed based on this model. By pre-training trajectory prediction networks for various types of UAVs, the traditional physics-based models are replaced for flight trajectory prediction. Inspired by Model Predictive Control (MPC), in the trajectory planning stage, the method generates multi-step trajectory points using an improved artificial potential field (APF) method, estimates the actual formation trajectory using the prediction network, and optimizes the trajectory through a multi-objective particle swarm optimization (MOPSO) algorithm after evaluating the planning costs. During actual flight, the optimized parameters generate trajectory points for the formation to follow. Unlike conventional path planning based on simple constraints, the proposed method directly plans trajectory points based on trajectory tracking performance, ensuring high feasibility for the formation to follow. Experimental results show that the CNN-LSTM network outperforms other networks in both short-term and long-term trajectory prediction. The proposed trajectory planning method demonstrates significant advantages in formation maintenance, trajectory tracking, and real-time obstacle avoidance, ensuring flight stability and safety while maintaining high-speed flight.
- Open Access
Three-Dimensional Event-Triggered Predefined-Time Cooperative Guidance Law
To address the problem of multiple missiles attacking a maneuvering target simultaneously in three-dimensional space, we propose a new predefined-time cooperative guidance law based on an event-triggered mechanism. The settling time of the system states under this guidance law is independent of the initial states, and the upper bound of the settling time can be directly set by the explicit parameters in the guidance law. Firstly, the time-to-go estimate is taken as a consistency variable, and the communication failure and time-delay that are easily encountered during the communication process are taken into account; the event-triggered mechanism is introduced into the guidance law along the line of sight (LOS) direction, and the event-triggered threshold is given. Then, a predefined-time extended state observer is used to accurately estimate disturbances. In addition, the stability of the proposed guidance laws along and perpendicular to the LOS direction is proven by the Lyapunov theory. Finally, the superiority of the proposed guidance law introducing the event-triggered mechanism in reducing energy consumption and its effectiveness in encountering communication failure and time-delay are verified through simulations.
Open AccessThree-Dimensional Event-Triggered Predefined-Time Cooperative Guidance Law
To address the problem of multiple missiles attacking a maneuvering target simultaneously in three-dimensional space, we propose a new predefined-time cooperative guidance law based on an event-triggered mechanism. The settling time of the system states under this guidance law is independent of the initial states, and the upper bound of the settling time can be directly set by the explicit parameters in the guidance law. Firstly, the time-to-go estimate is taken as a consistency variable, and the communication failure and time-delay that are easily encountered during the communication process are taken into account; the event-triggered mechanism is introduced into the guidance law along the line of sight (LOS) direction, and the event-triggered threshold is given. Then, a predefined-time extended state observer is used to accurately estimate disturbances. In addition, the stability of the proposed guidance laws along and perpendicular to the LOS direction is proven by the Lyapunov theory. Finally, the superiority of the proposed guidance law introducing the event-triggered mechanism in reducing energy consumption and its effectiveness in encountering communication failure and time-delay are verified through simulations.
- 7 August 2024
- Published by Elsevier in ISA Transactions
- Vol. 153, 384-403
Nonlinear adaptive robust control of tank bidirectional stabilizers with dead zone compensation based on extended state observer
- 7 August 2024
- Published by Elsevier in ISA Transactions
- Vol. 153, 384-403
Nonlinear adaptive robust control of tank bidirectional stabilizers with dead zone compensation based on extended state observer
- Open Access
- 20 May 2024
- Published by MDPI AG in Electronics
- Vol. 13 (10) , 1988
Nonlinear Robust Control of Vehicle Stabilization System with Uncertainty Based on Neural Network
To effectively suppress the effects of uncertainties including unmodeled dynamics and external disturbances in the vehicle stabilization system, a nonlinear robust control strategy based on a multilayer neural network is proposed in this paper. First, the mechanical and electrical coupling dynamics model of the vehicle stabilization system, considering model uncertainty and actuator dynamics, is refined. Second, the lumped uncertainty of the vehicle stabilization system is estimated by a multi-layer neural network and compensated by feedforward control. The high robustness of the system is ensured by constructing the sliding mode feedback control law. The proposed control method overcomes the limitations of sliding mode technology and the neural network and is naturally applied to the vehicle stabilization system, avoiding the adverse effects of high-gain feedback. Based on Lyapunov theory, it is demonstrated that the proposed controller is able to achieve the desired stability tracking performance. Finally, the effectiveness of the proposed control strategy is verified by co-simulation and comparative experiments.
- 20 May 2024
- Published by MDPI AG in Electronics
- Vol. 13 (10) , 1988
Open AccessNonlinear Robust Control of Vehicle Stabilization System with Uncertainty Based on Neural Network
To effectively suppress the effects of uncertainties including unmodeled dynamics and external disturbances in the vehicle stabilization system, a nonlinear robust control strategy based on a multilayer neural network is proposed in this paper. First, the mechanical and electrical coupling dynamics model of the vehicle stabilization system, considering model uncertainty and actuator dynamics, is refined. Second, the lumped uncertainty of the vehicle stabilization system is estimated by a multi-layer neural network and compensated by feedforward control. The high robustness of the system is ensured by constructing the sliding mode feedback control law. The proposed control method overcomes the limitations of sliding mode technology and the neural network and is naturally applied to the vehicle stabilization system, avoiding the adverse effects of high-gain feedback. Based on Lyapunov theory, it is demonstrated that the proposed controller is able to achieve the desired stability tracking performance. Finally, the effectiveness of the proposed control strategy is verified by co-simulation and comparative experiments.
- 20 March 2024
- Published by Springer Nature in International Journal of Control, Automation, and Systems
- Vol. 22 (4) , 1163-1175
Robust Control for Bidirectional Stabilization System With Time Delay Estimation
- 20 March 2024
- Published by Springer Nature in International Journal of Control, Automation, and Systems
- Vol. 22 (4) , 1163-1175
Robust Control for Bidirectional Stabilization System With Time Delay Estimation
- Open Access
- 17 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 115968-115985
Adaptive Robust Stability Control of All-Electrical Tank Gun Compensated by Radial Basis Neural Network
Stability control of the tank gun has emerged as a pivotal issue for moving tank gun control systems (TGCS). As a complex electromechanical integrated system, TGCS of moving tank inevitably possesses significant parametric uncertainties and uncertain nonlinearities. To effectively enhance the stabilization control performance of TGCS, in this study, we introduce an adaptive robust control (ARC) strategy based on radial basis function neural network (RBFNN) compensation. The adaptive technique is employed to address the parametric uncertainties, while the RBFNN is constructed to approximate the uncertain nonlinearities and realize feedforward compensation. Subsequently, to suppress the residual uncertainties, a nonlinear robust feedback control rate is devised to strengthen the robustness of the developed controller. Lyapunov analysis shows that the proposed controller achieves uniform ultimate bounded stability. Extensive simulation and electromechanical experimental results confirm the effectiveness of the proposed controller, which shows outstanding performance in handling strong parametric uncertainties and uncertain nonlinearities.
- 17 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 115968-115985
Open AccessAdaptive Robust Stability Control of All-Electrical Tank Gun Compensated by Radial Basis Neural Network
Stability control of the tank gun has emerged as a pivotal issue for moving tank gun control systems (TGCS). As a complex electromechanical integrated system, TGCS of moving tank inevitably possesses significant parametric uncertainties and uncertain nonlinearities. To effectively enhance the stabilization control performance of TGCS, in this study, we introduce an adaptive robust control (ARC) strategy based on radial basis function neural network (RBFNN) compensation. The adaptive technique is employed to address the parametric uncertainties, while the RBFNN is constructed to approximate the uncertain nonlinearities and realize feedforward compensation. Subsequently, to suppress the residual uncertainties, a nonlinear robust feedback control rate is devised to strengthen the robustness of the developed controller. Lyapunov analysis shows that the proposed controller achieves uniform ultimate bounded stability. Extensive simulation and electromechanical experimental results confirm the effectiveness of the proposed controller, which shows outstanding performance in handling strong parametric uncertainties and uncertain nonlinearities.