Article
Engineering, Multidisciplinary
H. Safaeipour, M. Forouzanfar, A. Ramezani
Summary: This paper addresses the challenging issue of incipient fault detection in real-time nonlinear closed-loop systems in mixed Gaussian and non-Gaussian environments. An online incipient fault detection method with acceptable computational efforts and an adaptive-robust residual scheme is proposed, based on the autocorrelation of the windowed residual signal and reasonable assumptions in nonlinear systems. The effectiveness of the proposed solution is demonstrated through the design and simulation of a closed-loop form of the three-tank system (DTS200).
Article
Engineering, Chemical
Zhijiang Lou, Youqing Wang
Summary: The proposed Neural Component Analysis (NCA) combines artificial neural networks (ANN) with principal component analysis (PCA) to address the common nonlinearity in industrial processes. NCA, with a similar network structure as ANN and utilizing the gradient descent method for training, successfully extracts uncorrelated components from the process data with PCA's dimension reduction strategy, and constructs statistical indices for process monitoring, showing superior performance compared to other nonlinear approaches in simulation tests.
INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH
(2021)
Article
Automation & Control Systems
Zhichao Li, Li Tian, Qingchao Jiang, Xuefeng Yan
Summary: A novel dynamic nonlinear process monitoring method based on dynamic nonlinear feature selection and kernel principal component regression (KPCR) is proposed in this study. The method considers the autocorrelation, cross-correlation, and nonlinear relationships among variables in time series data and achieves monitoring performance through variable selection and model establishment. The proposed method outperforms other advanced dynamic process monitoring methods.
JOURNAL OF THE FRANKLIN INSTITUTE-ENGINEERING AND APPLIED MATHEMATICS
(2022)
Article
Chemistry, Multidisciplinary
Zhe Zhou, Jian Wang, Chunjie Yang, Chenglin Wen, Zuxin Li
Summary: In this paper, a novel process monitoring method using SPA and the k-nearest neighbor algorithm is proposed. The method can effectively monitor the statistics of process variables and overcome the problems caused by non-Gaussianity and nonlinearity.
Article
Computer Science, Artificial Intelligence
Rudrasis Chakraborty, Liu Yang, Soren Hauberg, Baba C. Vemuri
Summary: Principal component analysis (PCA) and Kernel principal component analysis (KPCA) are fundamental methods in machine learning for dimensionality reduction. A geometric framework for computing principal linear subspaces is presented, showing consistency with principal components from Gaussian distribution. Efficient algorithm for computing projection onto the average subspace is provided, along with a faster method akin to KPCA and a novel online version of KPCA. Competitive performance of all algorithms is demonstrated on various real and synthetic data sets.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2021)
Article
Engineering, Environmental
Mohammed Tahar Habib Kaib, Abdelmalek Kouadri, Mohamed Faouzi Harkat, Abderazak Bensmail, Majdi Mansouri
Summary: Principal Component Analysis (PCA) is a commonly used technique for fault detection and diagnosis, but for datasets with nonlinear characteristics, Kernel PCA (KPCA) can be used as an alternative solution. However, KPCA has disadvantages such as a large number of observations and longer execution time. Reduced KPCA (RKPCA) is a solution to the limitations of KPCA, reducing storage space and execution time while maintaining acceptable monitoring performance by reducing the number of observations.
PROCESS SAFETY AND ENVIRONMENTAL PROTECTION
(2023)
Article
Engineering, Multidisciplinary
Hajer Lahdhiri, Okba Taouali
Summary: This paper introduces a proposed fault detection method for nonlinear uncertain systems, aiming to establish a guaranteed failure detection process by considering uncertainties.
Article
Chemistry, Multidisciplinary
Xuemei Wang, Ping Wu
Summary: This paper proposes an improved process monitoring method by considering autocorrelation among process data, integrating ensemble learning and kernel canonical variate analysis. The method achieves significantly enhanced fault detection performance.
Article
Engineering, Multidisciplinary
Hajer Lahdhiri, Okba Taouali
Summary: The paper introduces a new defect detection method RR-KGLRT, which leverages RRKPCA to build a reduced reference model and combines with GLRT for detection, improving detection performance. Additionally, the proposed RR-KGLRT based EWMA method significantly reduces false alarm rates and enhances detection performance.
Article
Engineering, Aerospace
Gunner S. Fritsch, Kyle J. DeMars
Summary: Traditional spaceflight navigation methods prioritize computational efficiency over model accuracy. This research introduces a new fault-tolerant approach that incorporates faulty sensor data into the measurement model to achieve intrinsic fault resistance. The proposed method not only outperforms residual editing, but also demonstrates strong robustness to unknown and approximated faulty measurement distributions.
JOURNAL OF GUIDANCE CONTROL AND DYNAMICS
(2021)
Article
Engineering, Chemical
Simin Li, Shuang-hua Yang, Yi Cao
Summary: Most industrial systems today are nonlinear and dynamic, and traditional fault detection techniques are limited in simultaneously extracting both nonlinear and dynamic features. This work proposes a novel nonlinear dynamic process monitoring method called canonical variate kernel analysis (CVKA), which combines the CVA method for linear dynamic feature extraction and kernel principal component analysis for nonlinear feature extraction. Experimental results on a TE process case study demonstrate the excellent performance of CVKA compared to other common approaches in dynamic nonlinear process monitoring for TE-like processes.
Article
Engineering, Marine
Yifan Xue, Gang Chen, Zhitong Li, Gang Xue, Wei Wang, Yanjun Liu
Summary: This study introduces an online identification scheme, namely the online fast noisy input Gaussian process, to accurately identify ship response models. The scheme is capable of reducing computational complexity while incorporating new noisy measurements online and making fast predictions. Through simulation and experimental verification, the developed scheme is proven to be a powerful online identification tool for ship maneuvering systems.
Article
Mathematics, Interdisciplinary Applications
Zhaojing Wang, Weidong Yang, Hong Zhang, Ying Zheng
Summary: This paper introduces a data-driven method for monitoring multimode processes, using statistics pattern analysis to extract statistical information from process data. A support vector data description method is proposed to address nonlinear and non-Gaussian problems, with a modified local reachability density ratio introduced as a weight factor to improve monitoring performance. The effectiveness of the method is demonstrated through example applications.
Article
Automation & Control Systems
Zhiyong Hu, Chao Wang
Summary: This article proposes a nonlinear online multioutput Gaussian process framework for the processing and analysis of multistream data. By developing a tailored covariance structure and applying Bayesian analysis and marginalized particle filter, this method is able to achieve nonlinear correlation modeling and online processing with superior modeling and prediction capabilities. It is particularly suitable for industrial informatics with complex relationships and large-scale datasets.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2022)
Article
Automation & Control Systems
Ali Asghar Sheydaeian Arani, Mahdi Aliyari Shoorehdeli, Ali Moarefianpour, Mohammad Teshnehlab
Summary: This paper introduces a method for fault estimation in a nonlinear system using the unscented Kalman filter, augmented by a fault signal as a state variable. A filter combining Gaussian mixture model and augmented ensemble unscented Kalman filter is designed for estimating faults in nonlinear systems, with suitable conditions and assumptions for convergence. The proposed method is evaluated in simulating a bioreactor system, demonstrating better performance compared to traditional methods in the presence of non-Gaussian noise.
INTERNATIONAL JOURNAL OF SYSTEMS SCIENCE
(2021)
Article
Engineering, Marine
Mohammad Yazdi, Faisal Khan, Rouzbeh Abbassi
Summary: Microbiologically Influenced Corrosion (MIC) poses a significant challenge to offshore oil and gas facilities, leading to pinholes and leaks. Preventive (coatings, cathodic protection) and mitigative (inhibitor and biocide treatment) actions are crucial for pipeline integrity management. A multi-objective functional methodology involving dynamic continuous Bayesian network modeling and a genetic algorithm is proposed to minimize operational risks associated with MIC.
Article
Engineering, Environmental
Mohammad Zaid Kamil, Faisal Khan, S. Zohra Halim, Paul Amyotte, Salim Ahmed
Summary: This study aims to develop a framework and tools to extract data from an accident database and establish a generalized accident causation model. By integrating Natural Language Processing, Interpretive Structural Model, and probabilistic methods, the model provides insights into accident factors, interactions, and pathways, and can be used for accident prevention strategies.
PROCESS SAFETY AND ENVIRONMENTAL PROTECTION
(2023)
Article
Engineering, Chemical
James Daley, Faisal Khan, Md. Tanjin Amin
Summary: This study proposes a method to assess the probability of process system failure based on operational data and system knowledge using principal component analysis and Bayesian network. The method can determine the probability of system failure once a fault is detected and provides valuable information for process safety.
PROCESS SAFETY PROGRESS
(2023)
Article
Engineering, Industrial
Vindex Domeh, Francis Obeng, Faisal Khan, Neil Bose, Elizabeth Sanli
Summary: Probabilistic safety assessment using the Bayesian network (BN) is a popular method for developing risk analysis tools. However, the subjectivity introduced by using subject-matter experts in eliciting probabilities for conditional probability tables (CPTs) decreases the reliability of the resulting tool. To address this issue, a probability-scoring scale was proposed to assign probabilities to CPTs, ensuring consistent results among different experts. The scale was applied to a BN-based risk -awareness (RAw) tool for monitoring safety aboard small fishing vessels (SFV), benefiting SFV owners and operators, the commercial fishing industry, and maritime administrations.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2023)
Article
Public, Environmental & Occupational Health
Mohammad Zaid Kamil, Mohammed Taleb-Berrouane, Faisal Khan, Paul Amyotte, Salim Ahmed
Summary: Underlying information about failure in free text can be used to determine causation in risk assessment. Advanced methodology is needed to identify the features in natural language expression. This study addresses the knowledge gap by extracting relevant features from textual data to develop cause-effect scenarios. The proposed methodology applies natural language processing and text-mining techniques to extract features and utilizes them in Bayesian networks for risk assessment.
Article
Computer Science, Interdisciplinary Applications
He Wen, Faisal Khan, Salim Ahmed, Syed Imtiaz, Stratos Pistikopoulos
Summary: Human-automation conflict is a frontier subject that needs to be vigilant against, especially under cyberattacks. This study transforms common attacks into understandable representations and explores the conflict under five generalized attacks using game theory. The results show that cyberattacks can significantly cause conflicts, and the control actions can buffer the impact of attacks within a limited range. The conflict risk can be used to distinguish faults and attacks, and appropriate measures can be taken accordingly.
COMPUTERS & CHEMICAL ENGINEERING
(2023)
Article
Engineering, Chemical
He Wen, Md. Tanjin Amin, Faisal Khan, Salim Ahmed, Syed Imtiaz, Efstratios Pistikopoulos
Summary: The conflict between human and artificial intelligence is a critical issue in Process System Engineering. This study proposes a novel methodology to quantify interpretation conflict probability and risk. The results show that interpretation conflict is often hidden or mixed with traditional faults and noises, which can easily be triggered by sensor faults, logic errors, cyberattacks, human mistakes, and misunderstandings.
INDUSTRIAL & ENGINEERING CHEMISTRY RESEARCH
(2023)
Article
Engineering, Marine
Adhitya Ramadhani, Faisal Khan, Bruce Colbourne, Salim Ahmed, Mohammed Taleb-Berrouane
Summary: Offshore structures such as oil platforms are affected by significant environmental loads. The complexity of the offshore environment requires robust models to capture the dependencies among environmental variables. Vine copula models, using both symmetric and asymmetric copula functions, are shown to provide a better estimation of the total environmental load on offshore structures compared to traditional methods. The results have implications for probabilistic structural analysis and design of offshore structures.
Article
Engineering, Environmental
Sumit Kumar, Ehsan Arzaghi, Til Baalisampang, Vikram Garaniya, Rouzbeh Abbassi
Summary: Green hydrogen has the potential to play a crucial role in a resilient and sustainable future, but challenges such as high costs and safety concerns need to be addressed. Offshore renewables offer a promising solution, with their decreasing costs and technological advances allowing for cost-competitive green hydrogen production. The offshore industrial sector, including oil and gas, aquaculture, and shipping, could serve as a substantial market and support the production, storage, and transmission of green hydrogen. This study provides critical insights into the key factors governing decision-making in offshore green hydrogen systems and offers guidance for site selection in uncertain and hazardous ocean environments.
PROCESS SAFETY AND ENVIRONMENTAL PROTECTION
(2023)
Article
Engineering, Chemical
He Wen, Faisal Khan
Summary: In recent years, the increase in cyber-connected industrial control systems (ICS) has raised cyber and process risks, highlighting the need for an integrated study on cybersecurity and process safety. This study analyzes cyber incidents related to ICS since 1990 and connects them with process accidents using the Bowtie method and the ATT&CK framework. It develops a Bayesian network to account for insignificant probabilities and confirms the vulnerability of the process industry to cyberattacks, with field controllers being the main targets. The study emphasizes the criticality of the safety instrument system (SIS) and the importance of dynamic threat assessment and neutralizing strategies.
CANADIAN JOURNAL OF CHEMICAL ENGINEERING
(2023)
Article
Engineering, Chemical
Stewart W. Behie, Hans J. Pasman, Syeda Zohra Halim, Kathy Shell, Ahmed Hamdy El-Kady, Faisal Khan
Summary: This article discusses the importance of safety systems in high-risk businesses and the necessity of long-term impact for cross-functional programs. The article also emphasizes the core principles and beliefs needed to make necessary adjustments and ensure continued success in all circumstances.
JOURNAL OF LOSS PREVENTION IN THE PROCESS INDUSTRIES
(2023)
Article
Engineering, Marine
Vindex Domeh, Francis Obeng, Faisal Khan, Neil Bose, Elizabeth Sanli
Summary: A quantitative risk analysis approach is proposed to develop a tool for studying loss of stability aboard small fishing vessels. The tool, which incorporates Bayesian network and De Morgan gates, provides a probabilistic assessment of the risk factors responsible for loss of stability occurrence. It is an innovative tool that can proactively ensure the stability of small fishing vessels.
Article
Engineering, Industrial
Uyen Dao, Zaman Sajid, Faisal Khan, Yahui Zhang
Summary: This paper develops a dynamic model to study the impact of corrosion on pipeline equipment failure. It uses a Bayesian network model to understand different risk factors and their interdependencies, transforming the corrosion mechanism into a probabilistic framework. The results are consistent with industrial practices and are crucial for the inspection and maintenance schedule of corroded pipelines.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2023)
Article
Transportation Science & Technology
A. Sardar, M. Anantharaman, V. Garaniya, F. Khan
Summary: The maritime industry is essential to the global economy, handling about 90% of worldwide trade and employing over a million seafarers. However, the industry faces a high number of casualties due to human errors in decision-making. This study proposes an AI-based approach, using an Ant Colony Optimization algorithm, to design and validate standardized instructions for daily operational tasks. This solution can optimize task paths, provide clear instructions, and reduce human errors, contributing to improved efficiency in the maritime industry.
TRANSNAV-INTERNATIONAL JOURNAL ON MARINE NAVIGATION AND SAFETY OF SEA TRANSPORTATION
(2023)
Article
Engineering, Industrial
Md. Tanjin Amin, Giordano Emrys Scarponi, Valerio Cozzani, Faisal Khan
Summary: This article examines the effectiveness and accuracy of threshold-based and probit-based methods in assessing domino effects. The results indicate that threshold-based methods are not suitable for quantitative assessment, and there are limitations in probit-based methods for time-dependent domino effect assessment. By utilizing site-specific structural response data and data analytics, a new improved time to failure prediction model is proposed, which demonstrates better performance compared to existing models.
RELIABILITY ENGINEERING & SYSTEM SAFETY
(2024)