Article
Automation & Control Systems
Yang Tao, Hongbo Shi, Bing Song, Shuai Tan
Summary: In this article, a hierarchical latent variable extraction and multisegment probability density analysis method is proposed to detect the incipient fault. The method achieves high detection rates by constructing data subspaces and evaluating distribution distance. The effectiveness of the proposed method is demonstrated in a real-world application.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2022)
Article
Engineering, Multidisciplinary
Liangliang Shang, Yinbo Gu, Yiming Tang, Huailiang Fu, Liang Hua
Summary: In this paper, a recursive ensemble canonical variate analysis (RECVA) approach is proposed for detecting incipient faults in dynamic processes. Multiple incipient fault models are obtained based on ensemble learning using bootstrap sampling. Two sensitive detection metrics are created using the maximum eigenvalue of a sliding window over each row of two matrices, and the eigenvalue is recursively updated. Experimental results demonstrate the effectiveness of RECVA for detecting early-stage faults that cannot be detected by traditional statistics.
Article
Engineering, Chemical
Miao Mou, Xiaoqiang Zhao
Summary: This paper proposes a method for detecting and diagnosing incipient nonlinear faults with missing data in industrial processes. The method uses low rank matrix decomposition to recover missing data and builds a mixed kernel function model in the recovered data to extract both local information and global characteristics. The dissimilarity statistic is introduced for fault detection. Numerical examples and simulation verification demonstrate the method's good detection and diagnosis capabilities.
JOURNAL OF THE TAIWAN INSTITUTE OF CHEMICAL ENGINEERS
(2022)
Article
Computer Science, Artificial Intelligence
Min Wang, Min Xie, Yanwen Wang, Maoyin Chen
Summary: In this article, a deep quality monitoring network (DQMNet) is developed for the detection of quality-related incipient faults. DQMNet uses feature extraction and Bayesian inference to extract hidden information and construct statistics, demonstrating its superiority through numerical simulation and benchmark data.
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
(2023)
Article
Automation & Control Systems
Wanke Yu, Chunhui Zhao, Biao Huang, Min Wu
Summary: In this study, a robust dissimilarity distribution analytics (RDDA) method is proposed for incipient industrial fault detection. The probabilistic model of the RDDA method is formulated with Laplace distribution, which provides better robustness compared to Gaussian distribution based models. By using variational inference, maximum likelihood estimations of latent variables and model parameters can be derived. A monitoring strategy is established based on static and dynamic statistics, utilizing dissimilarity between distributions of datasets. The proposed RDDA method is more suitable for practical industrial applications due to its consideration of missing data problems. Experimental results demonstrate the method's ability to accurately detect incipient faults using historical data with missing values.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2023)
Article
Engineering, Electrical & Electronic
Junjie Yang, Claude Delpha
Summary: Incipient fault detection is a challenging and popular topic that focuses on detecting early subtle changes to prevent severe security issues. The proposed methodology combines a Local Mahalanobis Distance algorithm and an Empirical Probability Density estimation technique to improve detection sensitivity for non-Gaussian data conditions. Performance analysis using simulation data and a case study on the Continuous-flow Stirred Tank Reactor demonstrate the effectiveness and benefits of the proposed approach.
Article
Automation & Control Systems
Lei Zeng, Xin Zhang, Ke Yu, Qiwen Jin, Yingchun Wu, Linghong Chen, Xuecheng Wu
Summary: In this paper, a deep representation learning fault detection scheme based on stacked sparse denoising autoencoder (SSDAE) is proposed to tackle the challenges in processing data in contemporary thermal power plants. The proposed method combines sparse denoising autoencoder (SDAE) and a deep learning architecture to achieve a highly nonlinear representation capability. Three monitoring indicators, RE2, MD2, and ZD(2), are designed based on the low-dimensional representation and residual distance of SSDAE. The effectiveness of the proposed method is validated through experiments on a nonlinear numerical case and a practical power plant pulverizing system, showing superior performance in detecting incipient and slight faults that traditional methods struggle with.
TRANSACTIONS OF THE INSTITUTE OF MEASUREMENT AND CONTROL
(2023)
Article
Automation & Control Systems
Bo Chen, Xiong-Lin Luo
Summary: The study introduces a process variation driven voting fusion strategy to detect incipient faults in complex industrial processes, utilizing varying control limits to track actual system dynamics and combining process information of each variable using a voting fusion strategy to monitor process variations. When tested on a real industrial process, the method shows sensitivity to incipient faults under varying operating conditions.
CHEMOMETRICS AND INTELLIGENT LABORATORY SYSTEMS
(2021)
Article
Automation & Control Systems
Huihui Gao, Wenjie Huang, Xuejin Gao, Honggui Han
Summary: To improve the performance of incipient fault detection for large-scale nonlinear industrial processes, a decentralized adaptively weighted stacked autoencoder (DAWSAE) -based fault detection method is proposed. It divides the industrial process into sub-blocks and establishes local adaptively weighted stacked autoencoders (AWSAE) to mine local information. It then constructs local and global statistics based on adaptively weighted feature vectors and residual vectors for fault detection.
Article
Computer Science, Artificial Intelligence
Imen Hamrouni, Hajer Lahdhiri, Khaoula Ben Abdellafou, Ahamed Aljuhani, Okba Taouali
Summary: This article discusses the application of multivariate static analysis in diagnosing uncertain systems. A fault detection strategy based on interval PCA and interval KPCA is proposed, which captures the variability of observations using interval variables. The proposed methods are validated and compared using synthetic data and real scenarios.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Automation & Control Systems
Li Cai, Hongpeng Yin, Jingdong Lin, Han Zhou, Dandan Zhao
Summary: This study addresses the issue of sample value imbalance in process monitoring and proposes a fault detection method based on variable selection and support vector data description. The method effectively highlights relevant information and handles high-dimensional and nonlinear variables. Experimental results show that it outperforms counterparts in terms of fault detection rate.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2022)
Article
Energy & Fuels
Xiaohui Wang, Yanjiang Wang, Xiaogang Deng, Zheng Zhang
Summary: An improved SVDD method, called DPSVDD, is proposed in this study by integrating convolutional autoencoder and probability-related monitoring indices, aiming to provide better monitoring performance on incipient faults, especially in batch processes.
Article
Engineering, Electrical & Electronic
Hongquan Ji
Summary: The air brake system of high-speed trains is essential for ensuring passenger safety. A new combined index has been proposed for detecting potential faults, and a fault-recovery-based optimization strategy has been presented for fault isolation.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Computer Science, Artificial Intelligence
Jiayang Liu, Qiang Zhang, Fuqi Xie, Xiaosun Wang, Shijing Wu
Summary: This paper proposes a new method based on improved octave convolution for detecting incipient faults in gearboxes under steady and variable conditions. By converting vibration signals into images and enhancing image detail learning through the addition of convolution kernels and residual connections in the high-frequency component of the octave convolution, as well as introducing self-attention units in the information interaction branches, the proposed method effectively identifies weak fault features. The combination of improved octave convolution and ResNet50 backbone network (IOC-ResNet50) is used for deep fault feature mining, and the results indicate superior performance compared to other published methods in both time-varying conditions and steady state.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Automation & Control Systems
Mohammad Amin Jarrahi, Haidar Samet, Teymoor Ghanbari
Summary: This article proposes a fault detection method for dc microgrids (DCMGs) based on the transient monitoring function (TMF). The method utilizes modal components of the measured currents and estimates the modal current using TMF concept. A residual signal is defined as the absolute difference between the modal and estimated currents, which exhibits significant variations after a fault occurrence. The Teager-Kaiser energy operator is then applied to amplify the changes, followed by a fault detection index (FDI) to reveal the fault and distinguish faulty conditions in different operation modes of DCMGs.
IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS
(2023)
Article
Automation & Control Systems
Xingding Zhao, Youqing Wang
Summary: This article studies and solves the problem of point-to-point iterative learning control (P2PILC), and proposes a learning law to compensate the initial state error and validates its effectiveness.
Article
Automation & Control Systems
Xin Ma, Yabin Si, Yihao Qin, Youqing Wang
Summary: Fault detection has always been a hot research topic in the industry. This paper proposes a novel algorithm called recursive innovational component statistical analysis (RICSA), which accurately estimates the dynamic structure of the data and divides the data space into dynamic components and innovational components, reducing the false alarm rate. Comparative experiments, especially on a practical coal pulverizing system in a 1000-MW ultra-supercritical thermal power plant, verify the superiority of RICSA in terms of accuracy rate, false alarm rate, and detection delay. The reduced computational complexity associated with RICSA is also discussed.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)
Article
Automation & Control Systems
Jian Zhang, Xianghua Wang, Youqing Wang
Summary: This paper focuses on the dynamic output feedback H infinity (DOFH) control for continuous-time Takagi-Sugeno (T-S) fuzzy systems under imperfect premise matching (IPM). A DOFH switching controller with membership functions (MFs) distinct from the fuzzy systems is designed. The controller is based on a non-quadratic Lyapunov function (NQLF) using MFs, and the time derivatives of MFs are addressed by a switching strategy. The proposed method incorporates more boundary information of MFs into the stability conditions to reduce conservatism.
Article
Automation & Control Systems
Shoulin Hao, Tao Liu, Xinpeng Geng, Youqing Wang
Summary: To tackle the asymmetric input constraint typically involved with industrial temperature control systems, an anti-windup active disturbance rejection control (ADRC) is proposed, which can also handle output delay. An anti-windup extended state observer (AESO) is constructed to estimate the delay-free system state and external disturbance, providing anti-windup compensation when input saturation constraint occurs. A feedback controller is derived with set-point prefilter design, ensuring no overshoot and output tracking deviation. The proposed control method is superior to existing anti-windup control designs, as demonstrated by benchmark examples and real applications.
Article
Automation & Control Systems
Yukun Shi, Youqing Wang, Jianyong Tuo
Summary: This paper investigates the problem of distributed secure state estimation for multi-agent systems under homologous sensor attacks. Two types of secure Luenberger-like distributed observers are proposed to estimate the system state and attack signal simultaneously. The proposed observers can handle both cases with and without time delays during network communication. It is shown that the attack estimations from different agents asymptotically converge to the same value. Sufficient conditions for ensuring the asymptotic convergence of the estimation errors are derived. Simulation examples are provided to validate the effectiveness of the proposed results.
IEEE-CAA JOURNAL OF AUTOMATICA SINICA
(2023)
Article
Engineering, Chemical
Yongming Han, Zilan Du, Zhiqiang Geng, Jinzhen Fan, Youqing Wang
Summary: This paper proposes a novel production prediction and energy structure optimization model (MC-LSTM) based on the long short-term memory neural network (LSTM) combining the Monte Carlo (MC) method. The model expands real production data using the MC method for prediction and optimizes the energy efficiency of inefficient samples based on the analysis of the MC-LSTM model's results. The experiment shows that the prediction accuracy of the ethylene production process based on the proposed model is about 96.57%, better than other prediction models, with an energy-saving potential of approximately 13.22%.
CHEMICAL ENGINEERING SCIENCE
(2023)
Article
Engineering, Electrical & Electronic
Xunlong Yin, Zonglei Mou, Youqing Wang
Summary: In the fault diagnosis of wind turbine planetary, a fault feature extraction method based on multiscale residual features (MRFs) is proposed, which amplifies signal dimensions and enriches fault information. The MRFs are obtained and placed in a classifier to train the diagnostic model. The proposed method improves the accuracy of gearbox fault diagnosis and has engineering application values.
IEEE SENSORS JOURNAL
(2023)
Article
Computer Science, Hardware & Architecture
Youqing Wang, Kun Zhao, Zhengqi Zheng
Summary: In this paper, a 3D indoor multipath signal model is built and 3D indoor positioning with a single base station is achieved by utilizing multipath channels. The proposed algorithm accurately matches the angles of arrival (AoAs) and time of arrival (ToAs) of each signal, and obtains linear equations of multipath channels with ray tracing modeling for 3D indoor positioning. An angle compensation algorithm is proposed by evaluating the probability weight of each multipath AoA and compensating the AoA according to the weights. Simulation results show that the proposed algorithm achieves positioning accuracy within 0.64 m in 90% of the cases, and within 0.25 m in 50% of the cases.
MOBILE NETWORKS & APPLICATIONS
(2023)
Article
Automation & Control Systems
Rongrong Sun, Youqing Wang, Zonglei Mou, Kaixun He
Summary: This study proposes a robust multiblock global orthogonal projections to latent structures (MBGOPLS) method for intelligently diagnosing faults in thermal power plants. The method achieves robustness to outliers while retaining diagnostic properties through double hierarchical clustering and optimization of the block regression coefficient matrix.
IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING
(2023)
Article
Engineering, Electrical & Electronic
Xianghua Wang, Youqing Wang, Ziye Zhang, Xiangrong Wang, Ron Patton
Summary: This paper proposes a novel active fault tolerant control scheme for a 3-degree-of-freedom (3-DOF) helicopter with sensor faults. A new interval observer (IO) with adaptive parameters is formulated to estimate disturbances and unmeasurable states. The IO acts as a state estimator and a fault detection and isolation observer. Fault estimation schemes are developed based on fault detectability, and a fault tolerant controller is constructed to ensure acceptable performance. Experimental verification on a 3-DOF helicopter platform is conducted.
IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS I-REGULAR PAPERS
(2023)
Article
Automation & Control Systems
Xingding Zhao, Jianyong Tuo, Youqing Wang
Summary: This article proposes a collective point-to-point iterative learning controller designed through collective intelligence for independent point-to-point control tasks. The article also introduces switched reference and corresponding iterative learning control switching strategies, and verifies the effectiveness of the algorithm through a simulation example.
JOURNAL OF THE FRANKLIN INSTITUTE-ENGINEERING AND APPLIED MATHEMATICS
(2023)
Article
Automation & Control Systems
Xin Ma, Dehao Wu, Shaoxu Gao, Tongze Hou, Youqing Wang
Summary: This article proposes an innovative dynamic process monitoring algorithm called autocorrelation feature analysis (AFA) which mines the dynamic information of continuous samples by calculating the correlation between current and past time features. The AFA algorithm has extremely low online computational complexity and has been verified on a CSTR and real data from a 1000-MW ultrasupercritical thermal power plant, showing its superiority.
IEEE TRANSACTIONS ON CYBERNETICS
(2023)
Proceedings Paper
Automation & Control Systems
Qiankun Li, Mingliang Cui, Youqing Wang
Summary: In this study, a lightweight network model is proposed to address the issues of low parameter count and high accuracy through experimental comparison.
2023 IEEE 12TH DATA DRIVEN CONTROL AND LEARNING SYSTEMS CONFERENCE, DDCLS
(2023)
Article
Engineering, Electrical & Electronic
Qiankun Li, Xin Ma, Mingliang Cui, Yu Hu, Jingfeng Zhao, Youqing Wang
Summary: Significant progress has been made in fault diagnosis algorithms, but they do not consider computational resources and require expensive equipment. To address this issue, this article proposes a CNN-based architecture that uses fewer computational resources and achieves higher accuracy. A loss function is also introduced to improve the accuracy without consuming too many resources. Experimental comparisons demonstrate the clear advantages of the proposed technique in terms of accuracy, resources, training time, and stability.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Proceedings Paper
Automation & Control Systems
Tongze Hou, Xin Ma, Qing Chen, Mingliang Cui, Youqing Wang
Summary: This study proposes a change-rate principal component analysis (CR-PCA) method for selecting features that are most conducive to process monitoring. The method reduces information loss by considering the fluctuation of score vectors and takes into account historical fault information. A sliding window algorithm is introduced to adaptively change the window width, which helps in self-adjusting the update speed of features and reducing the influence of random errors.
2022 41ST CHINESE CONTROL CONFERENCE (CCC)
(2022)
