Wentao Xu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Wentao Xu has a total of 9 co-authors. Here are the top co-authors by number of shared publications:
- Weihao Li
College of Information Science and Engineering, Northeastern University, Shenyang 110819, China
2 shared publications - Luping Zhao
College of Information Science and Engineering, Northeastern University, Shenyang 110819, China
1 shared publication - Peiyuan Guan
Department of Informatics, University of Oslo, Oslo, Norway
1 shared publication - Tianyi Li
Aalborg University
1 shared publication - Yumeng Song
Aalborg University
1 shared publication - Yushuai Li
Computer Science, Aalbersityorg University, 9220 Aalborg, Denmark
1 shared publication
Latest Publications
- Open Access
TCRL: Temporal-Coupled Adversarial Training for Robust Constrained Reinforcement Learning in Worst-Case Scenarios
Constrained Reinforcement Learning (CRL) aims to optimize decision-making policies under constraint conditions, making it highly applicable to safety-critical domains such as autonomous driving, robotics, and power grid management. However, existing robust CRL approaches predominantly focus on single-step perturbations and temporally independent adversarial models, lacking explicit modeling of robustness against temporally coupled perturbations. To tackle these challenges, we propose TCRL, a novel temporal-coupled adversarial training framework for robust constrained reinforcement learning (TCRL) in worst-case scenarios. First, TCRL introduces a worst-case-perceived cost constraint function that estimates safety costs under temporally coupled perturbations without the need to explicitly model adversarial attackers. Second, TCRL establishes a dual-constraint defense mechanism on the reward to counter temporally coupled adversaries while maintaining reward unpredictability. Experimental results demonstrate that TCRL consistently outperforms existing methods in terms of robustness against temporally coupled perturbation attacks across a variety of CRL tasks.
Open AccessTCRL: Temporal-Coupled Adversarial Training for Robust Constrained Reinforcement Learning in Worst-Case Scenarios
Constrained Reinforcement Learning (CRL) aims to optimize decision-making policies under constraint conditions, making it highly applicable to safety-critical domains such as autonomous driving, robotics, and power grid management. However, existing robust CRL approaches predominantly focus on single-step perturbations and temporally independent adversarial models, lacking explicit modeling of robustness against temporally coupled perturbations. To tackle these challenges, we propose TCRL, a novel temporal-coupled adversarial training framework for robust constrained reinforcement learning (TCRL) in worst-case scenarios. First, TCRL introduces a worst-case-perceived cost constraint function that estimates safety costs under temporally coupled perturbations without the need to explicitly model adversarial attackers. Second, TCRL establishes a dual-constraint defense mechanism on the reward to counter temporally coupled adversaries while maintaining reward unpredictability. Experimental results demonstrate that TCRL consistently outperforms existing methods in terms of robustness against temporally coupled perturbation attacks across a variety of CRL tasks.
- Open Access
A Multi-Mode Dynamic Fusion Mach Number Prediction Framework
The precise control of Mach numbers in supersonic and hypersonic compressor wind tunnel systems is a critical challenge in aerodynamic research. Although existing studies have improved prediction accuracy to some extent through machine learning methods, they generally neglect the multi-mode characteristics of complex wind tunnel systems, limiting the generalizability of the models. To address this issue, the present study proposes a multi-mode dynamic fusion Mach number prediction framework that integrates strategies of segmented modeling and cross-modal information fusion. First, single-mode segmented prediction models are constructed on the basis of Multi-output Support Vector Regression (MSVR), with hyperparameters optimized to capture the characteristics of individual modes. Second, the Partial Least Squares (PLS) method is employed to explore the correlations between historical and new modes, dynamically selecting the optimal prediction model and updating the historical mode repository. Experimental results demonstrate that the multi-mode dynamic fusion framework reduces the Root Mean Square Error (RMSE) by 70.57%, 56.4%, and 63.64% compared to Support Vector Regression (SVR), PLS, and Long Short-term Memory (LSTM) networks across six operating conditions. The framework proposed in this paper enhances Mach number prediction accuracy while improving model generalizability.
Open AccessA Multi-Mode Dynamic Fusion Mach Number Prediction Framework
The precise control of Mach numbers in supersonic and hypersonic compressor wind tunnel systems is a critical challenge in aerodynamic research. Although existing studies have improved prediction accuracy to some extent through machine learning methods, they generally neglect the multi-mode characteristics of complex wind tunnel systems, limiting the generalizability of the models. To address this issue, the present study proposes a multi-mode dynamic fusion Mach number prediction framework that integrates strategies of segmented modeling and cross-modal information fusion. First, single-mode segmented prediction models are constructed on the basis of Multi-output Support Vector Regression (MSVR), with hyperparameters optimized to capture the characteristics of individual modes. Second, the Partial Least Squares (PLS) method is employed to explore the correlations between historical and new modes, dynamically selecting the optimal prediction model and updating the historical mode repository. Experimental results demonstrate that the multi-mode dynamic fusion framework reduces the Root Mean Square Error (RMSE) by 70.57%, 56.4%, and 63.64% compared to Support Vector Regression (SVR), PLS, and Long Short-term Memory (LSTM) networks across six operating conditions. The framework proposed in this paper enhances Mach number prediction accuracy while improving model generalizability.
- Open Access
- 31 December 2024
- Published by MDPI AG in Mathematics
- Vol. 13 (1) , 132
Fast Frequency Control Strategy Based on Worst-Case Network Attack Perception
A fast frequency control (FFC) strategy using the proximal policy optimization based on worst-case network attack perception (worst-case PPO) algorithm is proposed to address the complexity of fast frequency control in power systems and the risks posed by network attacks. This strategy focuses on frequency stability in power systems with a high penetration of renewable energy, and utilizes a reinforcement learning agent to intelligently adjust the power setpoint of voltage source converters (VSCs), ensuring that both the frequency and the rate of change of the frequency remain within permissible limits. Considering the potential for network attacks, this strategy adopts the robust worst-case PPO algorithm, which ensures system stability even under the worst-case attack scenarios. The experimental results demonstrate that the proposed strategy effectively prevents frequency degradation under various disturbances, exhibiting a stronger disturbance resistance and robustness compared to traditional reinforcement learning methods. Furthermore, the strategy is easy to implement, highly adaptable, and suitable for the complex and dynamic operational environment of power systems, providing strong support for the secure and stable operation of smart grids.
- 31 December 2024
- Published by MDPI AG in Mathematics
- Vol. 13 (1) , 132
Open AccessFast Frequency Control Strategy Based on Worst-Case Network Attack Perception
A fast frequency control (FFC) strategy using the proximal policy optimization based on worst-case network attack perception (worst-case PPO) algorithm is proposed to address the complexity of fast frequency control in power systems and the risks posed by network attacks. This strategy focuses on frequency stability in power systems with a high penetration of renewable energy, and utilizes a reinforcement learning agent to intelligently adjust the power setpoint of voltage source converters (VSCs), ensuring that both the frequency and the rate of change of the frequency remain within permissible limits. Considering the potential for network attacks, this strategy adopts the robust worst-case PPO algorithm, which ensures system stability even under the worst-case attack scenarios. The experimental results demonstrate that the proposed strategy effectively prevents frequency degradation under various disturbances, exhibiting a stronger disturbance resistance and robustness compared to traditional reinforcement learning methods. Furthermore, the strategy is easy to implement, highly adaptable, and suitable for the complex and dynamic operational environment of power systems, providing strong support for the secure and stable operation of smart grids.