Xin Ma
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xin Ma has a total of 57 co-authors. Here are the top co-authors by number of shared publications:
- Youqing Wang
Beijing University of Chemical Technology
27 shared publications - Mingliang Cui
College of Information Science and Technology, Beijing University of Chemical Technology, Beijing, China
9 shared publications - Tongze Hou
College of Information Science and Technology, Beijing University of Chemical Technology, Beijing, China
9 shared publications - Jipeng Guo
College of Information Science and Technology, Beijing University of Chemical Technology, Beijing, China
5 shared publications Haoqian Wang
College of Information Sciences and Technology, Beijing University of Chemical Technology , North Third Ring Road 15, 100029 Beijing
3 shared publications- Min Yin
College of Information Science and Technology, Beijing University of Chemical Technology, Beijing, China
3 shared publications
Latest Publications
- 21 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. PP (15513203) , 1-12
Kernel Recursive Ridge Regression With Varying Hyperparameter and Favorable Sample Incorporation for Slow Time-Varying Process Monitoring
Industrial process data typical exhibit nonlinear and time-varying characteristics, which limit the effectiveness of traditional process monitoring methods. Kernel-based techniques can address nonlinearity by projecting data into high-dimensional feature spaces, but their application to time-varying systems remains limited due to two key challenges: the near-singularity of the kernel matrix during updates and the high computational cost of updating implementation. This study proposes a kernel-based time-varying vector autoregressive model to characterize system dynamics. An adaptive regularization strategy is introduced to balance model bias and numerical stability by dynamically tuning the regularization parameter. To reduce computational burden in updating implementation, the eigenvalue decomposition of the kernel matrix is reformulated as that of diagonal plus rank-one matrix and arrow matrix, thereby reducing at least one-third of the computational cost. To facilitate a practical model, a favorable sample incorporation strategy is designed to suppress the influence of faulty samples and enhance model robustness. The experimental results show that the proposed method achieves superior average fault detection rates and false alarm rates on five industrial datasets compared with existing advanced time-varying process monitoring methods, demonstrating good generalization performance.
- 21 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Informatics
- Vol. PP (15513203) , 1-12
Kernel Recursive Ridge Regression With Varying Hyperparameter and Favorable Sample Incorporation for Slow Time-Varying Process Monitoring
Industrial process data typical exhibit nonlinear and time-varying characteristics, which limit the effectiveness of traditional process monitoring methods. Kernel-based techniques can address nonlinearity by projecting data into high-dimensional feature spaces, but their application to time-varying systems remains limited due to two key challenges: the near-singularity of the kernel matrix during updates and the high computational cost of updating implementation. This study proposes a kernel-based time-varying vector autoregressive model to characterize system dynamics. An adaptive regularization strategy is introduced to balance model bias and numerical stability by dynamically tuning the regularization parameter. To reduce computational burden in updating implementation, the eigenvalue decomposition of the kernel matrix is reformulated as that of diagonal plus rank-one matrix and arrow matrix, thereby reducing at least one-third of the computational cost. To facilitate a practical model, a favorable sample incorporation strategy is designed to suppress the influence of faulty samples and enhance model robustness. The experimental results show that the proposed method achieves superior average fault detection rates and false alarm rates on five industrial datasets compared with existing advanced time-varying process monitoring methods, demonstrating good generalization performance.
- 15 December 2025
- Published by Elsevier in Computers & Chemical Engineering
- Vol. 207
Intelligent fault diagnosis in hybrid chemical processes under limited samples based on multi-feature fusion learning
- 15 December 2025
- Published by Elsevier in Computers & Chemical Engineering
- Vol. 207
Intelligent fault diagnosis in hybrid chemical processes under limited samples based on multi-feature fusion learning
- Open Access
- 1 November 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (6)
BBATProt: a framework predicting biological function with enhanced feature extraction via interpretable deep learning
Accurate prediction of protein and peptide functions from amino acid sequences is essential for understanding biological processes and advancing biomolecular engineering. Due to the limitations of experimental methods, computational approaches, particularly machine learning, have gained significant attention. However, many existing tools are task-specific and lack adaptability. Here, we propose a BERT–BiLSTM–Attention–TCN Protein Function Prediction Framework (BBATProt), a versatile framework for predicting protein and peptide functions. BBATProt leverages transfer learning with a pretrained bidirectional encoder representations from transformer model to capture high-dimensional features. The custom network integrates bidirectional long short-term memory and temporal convolutional network to align with proteins’ spatial characteristics, combining local and global feature extraction via attention mechanisms to achieve more precise predictions. Evaluations demonstrate that BBATProt consistently outperforms state-of-the-art models in tasks such as hydrolytic catalysis, peptide bioactivity, and post-translational modification (PTM) site prediction. Specifically, BBATProt improves accuracy by 2.96%–41.96% in antimicrobial peptide (AMP) prediction and by 0.64%–23.54% in PTM prediction tasks. In terms of area under the receiver operating characteristic curve, improvements range from 0.71% to 40.51% for AMP prediction and 0.62%–27.82% for PTM prediction. Visualizations of feature evolution and refinement via attention mechanisms validate the framework’s interpretability, providing transparency into the feature-extraction process and offering deeper insights into the basis of property prediction.
- 1 November 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (6)
Open AccessBBATProt: a framework predicting biological function with enhanced feature extraction via interpretable deep learning
Accurate prediction of protein and peptide functions from amino acid sequences is essential for understanding biological processes and advancing biomolecular engineering. Due to the limitations of experimental methods, computational approaches, particularly machine learning, have gained significant attention. However, many existing tools are task-specific and lack adaptability. Here, we propose a BERT–BiLSTM–Attention–TCN Protein Function Prediction Framework (BBATProt), a versatile framework for predicting protein and peptide functions. BBATProt leverages transfer learning with a pretrained bidirectional encoder representations from transformer model to capture high-dimensional features. The custom network integrates bidirectional long short-term memory and temporal convolutional network to align with proteins’ spatial characteristics, combining local and global feature extraction via attention mechanisms to achieve more precise predictions. Evaluations demonstrate that BBATProt consistently outperforms state-of-the-art models in tasks such as hydrolytic catalysis, peptide bioactivity, and post-translational modification (PTM) site prediction. Specifically, BBATProt improves accuracy by 2.96%–41.96% in antimicrobial peptide (AMP) prediction and by 0.64%–23.54% in PTM prediction tasks. In terms of area under the receiver operating characteristic curve, improvements range from 0.71% to 40.51% for AMP prediction and 0.62%–27.82% for PTM prediction. Visualizations of feature evolution and refinement via attention mechanisms validate the framework’s interpretability, providing transparency into the feature-extraction process and offering deeper insights into the basis of property prediction.
- 6 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Control Systems Technology
- Vol. 34 (1) , 411-425
Monitoring and Fault Risk Analysis for Nonlinear Dynamic Processes Based on Kernel Dynamic Regression Model
Traditional multivariate statistical process monitoring algorithms focus on whether measurements are significantly shifted compared with the training data, but lack further analysis of the monitoring results. This results in frequent alarm triggering for process variations that do not require urgent operator attention, such as operating condition deviations, faults that are unrelated to key performance indicators (KPI), and faults that are compensated by the closed-loop system feedback mechanism. Machine downtime for every alarm leads to high economic losses for the plant. Therefore, it is important to perform Fault Risk analysis to identify security threats from process variations. A risk-oriented approach should be able to determine whether faults are associated with safety or quality risks, thereby reducing overhaul costs and increasing economic efficiency. In this study, a Fault Risk analysis framework is proposed for nonlinear dynamic processes based on a kernel dynamic regression (KDR) model. The framework consists of two algorithms: one is KDR for detecting process faults, and the other is KPI-related KDR (KPI-KDR) for detecting faults affecting product quality. The proposed approaches provide more reasonable and interpretable dynamic and static subspace decomposition, which facilitates further analysis of the monitoring results. First, the KDR concurrently detects normal operating condition deviations and process faults. Then, the KPI-KDR analyzes whether faults can be compensated by feedback mechanisms. Finally, a closed-loop continuous stirred tank reactor and real catalytic cracking unit data are used to validate the effective performance of the proposed algorithms.
- 6 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Control Systems Technology
- Vol. 34 (1) , 411-425
Monitoring and Fault Risk Analysis for Nonlinear Dynamic Processes Based on Kernel Dynamic Regression Model
Traditional multivariate statistical process monitoring algorithms focus on whether measurements are significantly shifted compared with the training data, but lack further analysis of the monitoring results. This results in frequent alarm triggering for process variations that do not require urgent operator attention, such as operating condition deviations, faults that are unrelated to key performance indicators (KPI), and faults that are compensated by the closed-loop system feedback mechanism. Machine downtime for every alarm leads to high economic losses for the plant. Therefore, it is important to perform Fault Risk analysis to identify security threats from process variations. A risk-oriented approach should be able to determine whether faults are associated with safety or quality risks, thereby reducing overhaul costs and increasing economic efficiency. In this study, a Fault Risk analysis framework is proposed for nonlinear dynamic processes based on a kernel dynamic regression (KDR) model. The framework consists of two algorithms: one is KDR for detecting process faults, and the other is KPI-related KDR (KPI-KDR) for detecting faults affecting product quality. The proposed approaches provide more reasonable and interpretable dynamic and static subspace decomposition, which facilitates further analysis of the monitoring results. First, the KDR concurrently detects normal operating condition deviations and process faults. Then, the KPI-KDR analyzes whether faults can be compensated by feedback mechanisms. Finally, a closed-loop continuous stirred tank reactor and real catalytic cracking unit data are used to validate the effective performance of the proposed algorithms.
- 26 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Reliability
- Vol. 74 (4) , 5439-5451
Multimode Monitoring Method Based on Adaptive Importance Coding Dictionary Learning
The emergence of new operating modes due to changes in manufacturing demands or production technologies leads to the occurrence of multimode industrial processes. To address the multimode problem, the dictionary learning method is widely utilized due to its excellent sparse representation technique capability. Despite the abundance of offline data enabling dictionary learning to model various modes, consolidating dictionaries from individual models considerably increases model complexity. Therefore, a method called adaptive importance coding dictionary learning is proposed in this study on the basis of dictionary learning. This approach utilizes dictionary learning techniques to determine the importance of dictionary atoms by encoding frequencies and selecting significant dictionary atoms to form a new dictionary, remarkably reducing the complexity of the dictionary model. In accordance with this idea, a comprehensive framework for monitoring multimode processes, including online fault detection, mode classification, fault isolation, and new mode updates, is introduced. Experimental validations are conducted through numerical examples, simulations of continuous stirred tank reactors, and application in an actual petrochemical process. The experimental results demonstrate the effectiveness and feasibility of the proposed method in monitoring multimode processes.
- 26 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Reliability
- Vol. 74 (4) , 5439-5451
Multimode Monitoring Method Based on Adaptive Importance Coding Dictionary Learning
The emergence of new operating modes due to changes in manufacturing demands or production technologies leads to the occurrence of multimode industrial processes. To address the multimode problem, the dictionary learning method is widely utilized due to its excellent sparse representation technique capability. Despite the abundance of offline data enabling dictionary learning to model various modes, consolidating dictionaries from individual models considerably increases model complexity. Therefore, a method called adaptive importance coding dictionary learning is proposed in this study on the basis of dictionary learning. This approach utilizes dictionary learning techniques to determine the importance of dictionary atoms by encoding frequencies and selecting significant dictionary atoms to form a new dictionary, remarkably reducing the complexity of the dictionary model. In accordance with this idea, a comprehensive framework for monitoring multimode processes, including online fault detection, mode classification, fault isolation, and new mode updates, is introduced. Experimental validations are conducted through numerical examples, simulations of continuous stirred tank reactors, and application in an actual petrochemical process. The experimental results demonstrate the effectiveness and feasibility of the proposed method in monitoring multimode processes.
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
Spatiotemporal Local Analysis for Nonlinear Dynamic Process Monitoring
Modern industrial processes are becoming increasingly complex, exhibiting pronounced nonlinear dynamic characteristics that pose significant challenges for traditional monitoring approaches. Moreover, existing methods for nonlinear dynamic problems often fail to preserve the spatiotemporal characteristics of process data, resulting in suboptimal feature extraction that inadequately represents the original data’s temporal dynamics and spatial correlations, ultimately degrading monitoring performance. To address these limitations, this study proposes a spatiotemporal local analysis (STLA) framework. This method builds neighborhood relationships from temporal and spatial perspectives, and uses an autoencoder to approximate the local features of original data, thus enabling the features to better retain the original information and improve monitoring performance. Finally, a numerical example, the benchmark Tennessee Eastman process, and an actual power plant process demonstrate the superiority of the proposed method. The proposed method achieves enhanced performance with improved fault detection rate (FDR) and reduced false alarm rate (FAR).
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-12
Spatiotemporal Local Analysis for Nonlinear Dynamic Process Monitoring
Modern industrial processes are becoming increasingly complex, exhibiting pronounced nonlinear dynamic characteristics that pose significant challenges for traditional monitoring approaches. Moreover, existing methods for nonlinear dynamic problems often fail to preserve the spatiotemporal characteristics of process data, resulting in suboptimal feature extraction that inadequately represents the original data’s temporal dynamics and spatial correlations, ultimately degrading monitoring performance. To address these limitations, this study proposes a spatiotemporal local analysis (STLA) framework. This method builds neighborhood relationships from temporal and spatial perspectives, and uses an autoencoder to approximate the local features of original data, thus enabling the features to better retain the original information and improve monitoring performance. Finally, a numerical example, the benchmark Tennessee Eastman process, and an actual power plant process demonstrate the superiority of the proposed method. The proposed method achieves enhanced performance with improved fault detection rate (FDR) and reduced false alarm rate (FAR).
- 13 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (14) , 27250-27264
Semi-Supervised Multitask Learning Approach Boosted by Operation Strategy Expert System for Industrial Process Fault Diagnosis
Fault diagnosis (FD) is crucial for maintaining industrial processes safe. Although there is a wealth of historical running data stored, labeled fault data is insufficient. Semi-supervised learning mechanisms can fully utilize unlabeled data to enhance model generalization and performance. Expert systems (ES) store rich expert knowledge and experience, providing additional decision information and experiential label support for neural network (NN). This study proposes a semi-supervised learning approach that combines ES with NN for industrial process FD. This approach adopts a multitask learning framework, wherein classification and metafeature learning are the main and auxiliary tasks, respectively. A rule-based ES is constructed using a mixed integer linear programming method. ES make decisions based on meta-feature learning results, and during the backpropagation process, these decisions are used alongside predictions of the main task for parameter updates. The research findings suggest that the proposed approach has several advantages, including: (1) It enhances the model’s diagnostic performance, interpretability, and interactability. (2) It lessens the demand for labeled samples in the neural network. (3) It provides visual explanations for decision results. This study was validated on a simulated three-tank water storage platform and a gasification process case in a certain factory, achieving satisfactory results.
- 13 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (14) , 27250-27264
Semi-Supervised Multitask Learning Approach Boosted by Operation Strategy Expert System for Industrial Process Fault Diagnosis
Fault diagnosis (FD) is crucial for maintaining industrial processes safe. Although there is a wealth of historical running data stored, labeled fault data is insufficient. Semi-supervised learning mechanisms can fully utilize unlabeled data to enhance model generalization and performance. Expert systems (ES) store rich expert knowledge and experience, providing additional decision information and experiential label support for neural network (NN). This study proposes a semi-supervised learning approach that combines ES with NN for industrial process FD. This approach adopts a multitask learning framework, wherein classification and metafeature learning are the main and auxiliary tasks, respectively. A rule-based ES is constructed using a mixed integer linear programming method. ES make decisions based on meta-feature learning results, and during the backpropagation process, these decisions are used alongside predictions of the main task for parameter updates. The research findings suggest that the proposed approach has several advantages, including: (1) It enhances the model’s diagnostic performance, interpretability, and interactability. (2) It lessens the demand for labeled samples in the neural network. (3) It provides visual explanations for decision results. This study was validated on a simulated three-tank water storage platform and a gasification process case in a certain factory, achieving satisfactory results.
- 12 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 55 (9) , 5912-5924
Jarque-Bera-Based Artificial Neural Correlation Analysis for Nonlinear and Non-Gaussian Process Monitoring
Nonlinear and non-Gaussian characteristics are common in industrial processes. Artificial neural correlation analysis (ANCA) is a good nonlinear process monitoring algorithm, which combines classical correlation analysis with artificial neural networks. However, its performance is not very satisfactory for industrial processes with non-Gaussian characteristics. To solve non-Gaussian problems, almost all the existing process monitoring algorithms only consider the effect of kurtosis. Nevertheless, both kurtosis and skewness affect the data distribution. To improve the limitations of existing algorithms, this study proposes a new process monitoring algorithm named Jarque-Bera-based ANCA. This new algorithm makes many improvements to ANCA scheme, and the designed loss function combines the influence of both kurtosis and skewness on the data distribution, which not only maintains the advantages of the ANCA algorithm in solving nonlinear problems, but also provides superior monitoring performance in non-Gaussian processes. Furthermore, the superior performance of the proposed new algorithm is verified through simulations using non-Gaussian and nonlinear numerical examples, the Tennessee Eastman process, and catalytic cracking units.
- 12 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Systems, Man, and Cybernetics: Systems
- Vol. 55 (9) , 5912-5924
Jarque-Bera-Based Artificial Neural Correlation Analysis for Nonlinear and Non-Gaussian Process Monitoring
Nonlinear and non-Gaussian characteristics are common in industrial processes. Artificial neural correlation analysis (ANCA) is a good nonlinear process monitoring algorithm, which combines classical correlation analysis with artificial neural networks. However, its performance is not very satisfactory for industrial processes with non-Gaussian characteristics. To solve non-Gaussian problems, almost all the existing process monitoring algorithms only consider the effect of kurtosis. Nevertheless, both kurtosis and skewness affect the data distribution. To improve the limitations of existing algorithms, this study proposes a new process monitoring algorithm named Jarque-Bera-based ANCA. This new algorithm makes many improvements to ANCA scheme, and the designed loss function combines the influence of both kurtosis and skewness on the data distribution, which not only maintains the advantages of the ANCA algorithm in solving nonlinear problems, but also provides superior monitoring performance in non-Gaussian processes. Furthermore, the superior performance of the proposed new algorithm is verified through simulations using non-Gaussian and nonlinear numerical examples, the Tennessee Eastman process, and catalytic cracking units.