Li Chai

Li Chai

Auto-generated profile, based on publicly available information

Li Chai

Zhejiang University

Research Focus & Metrics

Academic Career Length

19Years (2008-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
16 August 2024 in IEEE Transactions on Smart Grid
Verified
Journal Article
Closed access
SPSha Peng
Sha Peng
MLMengxiang Liu
Mengxiang Liu
LCLi Chai
Li Chai
RDRuilong Deng
Ruilong Deng
The increasing deployment of solar photovoltaic (PV) systems in the electric grid, aimed at addressing the energy crisis and surging power demands, has expanded the potential vulnerability to cyberattacks due to the inter-networking of the grid-connected power electronics converters. In this paper, we propose a dynamic spatiotemporal graph neural network (DST-GNN) for cyberattack detection in grid-tied PV systems. Specifically, to exploit the inherent graph topology of the grid-tied PV system, we start by employing a GNN with a dynamic weighted adjacency matrix to capture the latent spatial correlations within signal data. Then, a one-dimensional convolution neural network (1D-CNN) is utilized to extract the underlying temporal patterns. Notably, we leverage the system dynamics to determine the dynamic graph weights and the number of graph convolution layers, while the hyper-parameters of 1D-CNN are designed based on the periodicity of input signals. Finally, the integration of the priori physical system knowledge further enhances the interpretability and improves the detection performance of DST-GNN. To the best of our knowledge, this is the first work that embeds the grid-tied PV system into a graph structure for cyberattack detection. The effectiveness of DST-GNN is evaluated through comprehensive case studies on a hardware-in-the-loop (HIL) grid-tied PV testbed, and numerical results demonstrate its superiority over baseline methods.

23 February 2022 in IEEE Transactions on Industrial Informatics
Verified
Journal Article
Closed access
HWHan Wu
Han Wu
LCLi Chai
Li Chai
YTYu-Chu Tian
Yu-Chu Tian
With the increasing integration of intermittent energy sources into the smart grid, distributed battery energy storage systems (DBESSs) are employed to balance power generation and demand. Power allocation among DBESSs plays an important role in maintaining the stability of energy systems. So far, the control of DBESSs has focused on either continuous-time control for continuous-time battery dynamics or discrete-time control for discretized battery dynamics. However, in realistic industrial applications, DBESSs have continuous-time dynamics in nature, and their control is implemented on digital controllers. To tackle this issue, a distributed multirate control system is designed in this article for continuous-time DBESSs. It allocates power by keeping the same relative State-of-Charge (SoC) variation rate for all DBESSs. For the accurate computation of the output/input power without global information, the control strategy consists of distributed multirate estimators each for a DBESS. The operating rate of the estimation algorithm is designed multiple times higher than the sampling rate of the measurement of battery and the sampling rate of the battery controller. With the proposed multirate estimator, a smaller estimation error is achieved. The same relative SoC variation rate and the stability of closed-loop system can be guaranteed. Simulation studies are given to demonstrate the effectiveness of the proposed control strategy.

11 November 2024 in IEEE Transactions on Systems, Man, and Cybernetics: Systems
Verified
Journal Article
Closed access
Siwei LouSiwei Lou
Siwei Lou
Chunjie YangChunjie Yang
Chunjie Yang
LKLiyuan Kong
Liyuan Kong
TLTao Lin
Tao Lin
HZHanwen Zhang
Hanwen Zhang
LCLi Chai
Li Chai
Broad learning system (BLS), a tri-layer feedforward neural network, has gained widespread recognition for its exceptional scalability and computational efficiency. However, BLS and its derivatives encounter several challenges: 1) overlooking the uncertainty introduced by numerous nonlinear random mappings; 2) failing to cope with the misalignment of model inputs with the output sampling rate; 3) lack of attention to nonstationary scenarios; and 4) absence of theoretical optimization for incremental learning. To overcome these obstacles, we propose a regression modeling and anomaly detection scheme rooted in a temporal kernel stationary BLS (TKS-BLS). We first create a nonlinear kernel broad representation (NKBR) extraction strategy, providing a robust nonlinear foundation for random feature mapping via kernel technology. Following this, we probe the mechanism of temporal matching between model inputs and outputs through a temporal alignment parameter, interpretable under a latent variable relationship. In the integration phase, we establish a Kullback-Leibler divergence objective function to facilitate the capture of stationary relationships within time-series data, in conjunction with the regression error. Subsequently, a double-loop parameter optimization algorithm and an independent incremental learning mechanism are put forth, both backed by comprehensive theoretical analyses. Our method’s superiority is thoroughly confirmed by experimental outcomes from extensive case studies across seven real ironmaking process datasets.

27 June 2024 in IEEE Transactions on Industrial Informatics
Verified
Journal Article
Closed access
HWHan Wu
Han Wu
LCLi Chai
Li Chai
ZZZhen-Hua Zhu
Zhen-Hua Zhu
YTYu-Chu Tian
Yu-Chu Tian
DC microgrids with battery energy storage systems are being widely implemented for integrating renewable energy. The convergence performance of the battery controller is an important index in the evaluation of the microgrids performance. However, the convergence time of existing finite-time control, fixed-time control, and predefined-time control cannot be preset explicitly. Moreover, existing state-of-charge (SoC)-equalization-based accelerating control algorithms will make the batteries suffer excessive voltage and current. To deal with these problems, this article presents a distributed prescribed-time control scheme embedded with a prescribed-time dynamic average consensus (DAC) algorithm for both discharging mode and charging mode. In discharging mode, the droop coefficient is designed such that all batteries keep the same relative SoC variation rate. With the proposed secondary control input, theoretical analysis shows that the voltage regulation and accurate current sharing can be obtained within any physically allowable user-preassigned time, which is independent of any other control parameters and initial states. SoC balancing is also achieved within the preassigned time. All batteries can keep the same relative SoC variation rate, which is more reasonable than simple SoC equalization. In charging mode, an SoC-based virtual resistance and a prescribed-time virtual voltage compensation control are proposed, with which the current sharing and the same relative SoC variation rate of each battery is achieved within the preassigned time. Simulation studies are conducted to demonstrate the effectiveness of the proposed control scheme

28 August 2023 in IEEE Transactions on Image Processing
Verified
Journal Article
Closed access
XSXiaoying Song
Xiaoying Song
KWKe Wu
Ke Wu
LCLi Chai
Li Chai
Since high-order relationships among multiple brain regions-of-interests (ROIs) are helpful to explore the pathogenesis of neurological diseases more deeply, hypergraph-based brain networks are more suitable for brain science research. Unlike the existing hypergraph based brain network (brain hypernetwork), where hyperedges containing the same number of ROIs are assumed to have equal weights (to some extent, the network is unweighted), and the underlying structure is described only by an incidence/adjacency matrix, in this paper, we propose a framework for constructing a truly weighted brain hypernetwork described by an adjacency tensor. Considering the relationships among vertices within a hyperedge, we propose a novel hyperedge weight estimation method and convert the incidence matrix into a weighted adjacency tensor. On the basis of tensor decomposition, we apply hypergraph signal processing tools, such as hypergraph Fourier transform, to analyze and compare the spectrum between schizophrenia patients and normal controls. It is found that there are more high frequency components in the spectrum of patients than controls, and the average amplitude is significantly greater than that of controls. Instead of extracting some simple topological features from brain hypernetworks for classification, we innovatively use the hypergraph spectrum and the spectral signal as classification features, and the classification results on two public datasets demonstrate the effectiveness of our proposed method.

10 March 2023 in IEEE Transactions on Automatic Control
Verified
Journal Article
Closed access
JYJing-Wen Yi
Jing-Wen Yi
LCLi Chai
Li Chai
JZJingxin Zhang
Jingxin Zhang
Previous researches have shown that adding local memory can accelerate the consensus. It is natural to question what is the fastest rate achievable by the $M$ -tap memory acceleration, and what are the corresponding control parameters. This paper introduces a set of effective and previously unused techniques to analyze the convergence rate of accelerated consensus with $M$ -tap memory of local nodes and to design the control protocols. These effective techniques, including the Kharitonov stability theorem, the Routh stability criterion and the robust stability margin, have led to the following new results: 1) the direct link between the convergence rate and the control parameters; 2) explicit formulas of the optimal convergence rate and the corresponding optimal control parameters for $M \leq 2$ on a given graph; 3) analytic formulas of the optimal worst-case convergence rate and the corresponding optimal control parameters for the memory $M \geq 1$ on a set of uncertain graphs. We show that the acceleration with the memory $M=1$ provides the optimal convergence rate in the sense of worst-case performance. Several numerical examples are given to demonstrate the validity and performance of the theoretical results.