{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/neural-component-analysis-for-fault-detection","title":"Neural Component Analysis for Fault Detection","arxiv_id":"1712.04118","date":"2017-12-12","proceeding":null,"authors":["Haitao Zhao"],"abstract":"Principal component analysis (PCA) is largely adopted for chemical process\nmonitoring and numerous PCA-based systems have been developed to solve various\nfault detection and diagnosis problems. Since PCA-based methods assume that the\nmonitored process is linear, nonlinear PCA models, such as autoencoder models\nand kernel principal component analysis (KPCA), has been proposed and applied\nto nonlinear process monitoring. However, KPCA-based methods need to perform\neigen-decomposition (ED) on the kernel Gram matrix whose dimensions depend on\nthe number of training data. Moreover, prefixed kernel parameters cannot be\nmost effective for different faults which may need different parameters to\nmaximize their respective detection performances. Autoencoder models lack the\nconsideration of orthogonal constraints which is crucial for PCA-based\nalgorithms. To address these problems, this paper proposes a novel nonlinear\nmethod, called neural component analysis (NCA), which intends to train a\nfeedforward neural work with orthogonal constraints such as those used in PCA.\nNCA can adaptively learn its parameters through backpropagation and the\ndimensionality of the nonlinear features has no relationship with the number of\ntraining samples. Extensive experimental results on the Tennessee Eastman (TE)\nbenchmark process show the superiority of NCA in terms of missed detection rate\n(MDR) and false alarm rate (FAR). The source code of NCA can be found in\nhttps://github.com/haitaozhao/Neural-Component-Analysis.git.","url_abs":"http://arxiv.org/abs/1712.04118v1","url_pdf":"http://arxiv.org/pdf/1712.04118v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"neural-component-analysis-for-fault-detection","repo_url":"https://github.com/haitaozhao/Neural-Component-Analysis","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"chemical-process","task_name":"Chemical Process"},{"task_slug":"fault-detection","task_name":"Fault Detection"}],"methods":[{"method_slug":"pca","method_name":"PCA"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}