Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Jian Sun has a total of 53 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Crack Structure Preserving Environmental Appearance Augmentation for Robust Surface Crack Segmentation in Complex Environments

    by Wenfeng Li, Bo Zhang, Zhiyu Luo, Yuhan Zhang, Guangrui Wen, Zhiqin Zhu, Jian Sun

    Crack segmentation is an important task in infrastructure maintenance. Engineering structures require regular crack inspection and maintenance to reduce the risk of further deterioration. For crack segmentation under different environmental conditions, sufficient and representative training data are needed to train neural network models effectively. However, variations in illumination, visibility, surface texture, image resolution, and environmental noise during practical image acquisition can alter crack appearance and reduce segmentation stability. To address these problems, this study develops a sequential crack data augmentation framework using GAN, CNN, and VGG19. The GAN first generates diverse crack images. A mask prediction CNN then produces the corresponding binary pseudo-label masks, after which VGG19 style transfer introduces the target environmental appearance. During style transfer, the content constraint limits changes to crack-related structure, while the previously predicted mask remains unchanged and is paired with the transferred image. The resulting image and mask pairs are used to expand the training data for subsequent crack segmentation. The framework is evaluated using crack datasets representing different environmental conditions. Experimental results show that the proposed augmentation procedure supports crack segmentation under complex environmental interference and improves the model’s adaptability to variations in environmental appearance. The framework provides a practical approach for expanding crack training data when representative images from complex environments are limited.

  • Open Access

    A Spatio-temporal Load Forecasting Method for Electric Vehicles Based on Multi-Graph Dynamic Fusion

    by Ziqi Huang, Mengting Xie, Haixiang Zhang, Wenjing Huang, Chao Li, Jian Sun

    Large-scale charging of electric vehicles poses challenges to safety of power grid operations, and accurately predicting their spatio-temporal load is key to achieving vehicle-grid coordination. Aiming at the limited expression ability of inter-regional correlation relationships in existing load forecasting models, this paper proposes a spatio-temporal load forecasting framework based on multi-graph dynamic fusion to capture more precise spatial dependencies. Experiments using real charging load data from Shenzhen demonstrate that the proposed model outperforms baseline models across primary performance metrics, achieving a 10.9% reduction in MAE compared to the optimal baseline model. This validates its effectiveness and generalization capability in complex spatio-temporal load prediction tasks.

  • A DoS Attack-Aware Adaptive Switching Reinforcement Learning Control for Resilient V2G-Based Grid Frequency Regulation

    by Jian Sun, Lingli Dong, Shuiqing Xu, Ruixiao Lv, Jingren Tang, Huaqing Li, Yi Chai, Josep M. Guerrero

    Vehicle-to-grid (V2G) technology enables electric vehicles (EVs) to provide regulation services, mitigating grid frequency deviations caused by renewable energy sources. However, V2G chargers integrating with communications can be disrupted by denial-of-service (DoS) attacks, thus degrading the performance of V2G-based frequency regulation (V2G-FR). This paper proposes an attack-aware adaptive switching learning control scheme to improve the resilience of the V2G-FR system under time-varying DoS attacks. A switching structure framework is proposed, consisting of distributed cooperative consensus and local controls with integral reinforcement learning that relies on remote measurements in DoS rest time and local measurements in DoS time, respectively. An adaptive mechanism with attack awareness is integrated into the reinforcement learning to adapt the controls to the time-varying attack intensity. In addition, the distributed corporative consensus control achieves fair V2G power usage among EV agents and reduces the communication burden caused by centralized V2G power dispatch. The proposed scheme is verified on an IEEE 39- bus system. The results show the proposed scheme can mitigate the impact of DoS attacks through the proposed switching and adaptation of the time-varying attack intensity. Meanwhile, EV agents can also reach a consensus on fair V2G power usage in V2G-FR.


  • A V2G Capacity-Maximized Dispatch Control Strategy via Nested Consensus Control Learning for Grid Frequency Regulation

    by Jian Sun, Huaqing Li, Huiwei Wang, Sen Tan, Yi Chai, Josep M. Guerrero

    Optimizing vehicle-to-grid (V2G)-based frequency regulation (FR) is challenging due to the need to coordinate dispatch control under diverse electric vehicle (EV) owner behaviors while avoiding excessive charge/discharge. This article proposes an optimized V2G dispatch control strategy for FR using nested consensus reinforcement learning (NCRL) to improve coordination among V2G resources. The strategy applies an integral reinforcement learning framework to optimally dispatch FR signals for EV clusters, while a novel nested consensus mechanism synchronizes energy consumption to prevent premature cluster withdrawal and sustain maximum FR capacity. An innovative exponential metric is developed for consensus-based power dispatch, which, by incorporating energy consumption consensus, preserves the maximum aggregated V2G FR power and enhances performance during critical grid disturbances. Compared with conventional proportional metrics, the proposed approach maximizes the available V2G FR capacity, i.e., the deliverable FR power. Simulations on the IEEE 39-bus system demonstrate that the strategy achieves intercluster consensus, maintains maximum FR capacity, and improves V2G FR performance.


  • Sparse-Promoting Generative Adversarial Learning for Microgrid Control With Structural Adaptive Optimization Against FDI Attacks

    by Jian Sun, Ruixiao Lv, Shuiqing Xu, Sen Tan, Josep M. Guerrero

    With the rapid development of power electronics and renewable energy, microgrids have evolved into a flexible and complex networked control system. However, it also increases the likelihood of false data injection (FDI) attacks, which damages the performance of microgrids. To mitigate the negative impact of FDI attacks and enhance the reliability of microgrids, we optimize the control performance by addressing both the control system structure and the control laws. In this study, we propose a sparse-promoting generative adversarial learning scheme to simultaneously design the control structure and the control laws, without relying on a linearized analytical model. The proposed scheme estimates the attacked probability of each distributed generator (DG) nodes by conditional generative adversarial network (cGAN), while considering the impact of attacked nodes selection and adversarial detection on system performance. Subsequently, the solution of the optimal structure and control laws is conbined with the attack probability to realize the adaptive regulation against FDI attcks. In addition, we designed a security confidence function to ensure the maximum power output of attacked nodes under secure control when the control structure is optimized by sparse neural network (SNN). Simulations are conducted in MATLAB/Simulink to verify the effectiveness and advantages of the proposed scheme. The results demonstrate that the proposed scheme can effectively mitigate feeder power degradation during FDI attacks.


  • Open Access

    Load Forecasting for Commercial Buildings Using BiLSTM–Transformer Network and Cyber–Physical Cognitive Control Systems

    by Xiong Xiong, Zicheng Huang, Yilin Chen, Jian Sun

    With the widespread adoption of electric vehicles (EVs), their charging and discharging schedules pose new challenges for real-time load forecasting in commercial buildings. This study proposes a prediction model based on the integration of bidirectional long short-term memory (BiLSTM) networks and Transformer architecture, along with the introduction of a cognitive control system and cyber–physical systems (CPS) to address issues such as data loss and excessive computation time during the forecasting process. The BiLSTM–Transformer model significantly improves load-forecasting accuracy and real-time performance by combining time-series modeling with global feature extraction capabilities. Additionally, the cognitive control system includes user-aware cognitive control (UACC) and Microgrid Control Center Cognitive Control (MACC). UACC quantifies information gaps in real time and adaptively adjusts strategies during communication instability, while MACC employs Q-learning algorithms to evaluate the impact of data loss on scheduling and optimize power resource allocation. The synergy between these mechanisms ensures system stability and predictive performance in scenarios involving data loss or communication disruptions. Experimental results demonstrate that the model achieves outstanding predictive accuracy under complete data conditions and significantly reduces errors in scenarios with data loss, validating its superior accuracy and robustness. This provides reliable support for load forecasting in commercial buildings.


  • Prescribed-Time Optimal Consensus for Switched Stochastic Multiagent Systems: Reinforcement Learning Strategy

    by Weiwei Guang, Xin Wang, Lihua Tan, Jian Sun, Tingwen Huang

    This paper focuses on the event-triggered-based prescribed-time optimal consensus control issue for switched stochastic nonlinear multi–agent systems under switching topologies. Notably, the system stability may be affected owing to the change in information transmission channels between agents. To surmount this obstacle, this paper presents a reconstruction mechanism to rebuild the consensus error at the switching topology instant. Combining optimal control theory and reinforcement learning strategy, the identifier neural network is utilized to approximate the unknown function, with its corresponding updating law being independent of the switching duration of system dynamics. In addition, an event-triggered mechanism is adopted to enhance the efficiency of resource utilization. With the assistance of the Lyapunov stability principle, sufficient conditions are established to ensure that all signals in the closed-loop system are cooperatively semi-globally uniformly ultimately bounded in probability and the consensus error is capable of converging to the specified interval in a prescribed time. At last, a simulation example is carried out to validate the feasibility of the presented control scheme.