Causal Mechanism Transfer Network for Time Series Domain Adaptation in Mechanical Systems

Data-driven models are becoming essential parts in modern mechanical systems, commonly used to capture the behavior of various equipment and varying environmental characteristics. Despite the advantages of these data-driven models on excellent adaptivity to high dynamics and aging equipment, they ar...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:ACM transactions on intelligent systems and technology 2021-03, Vol.12 (2), p.1-21
Hauptverfasser: Li, Zijian, Cai, Ruichu, Ng, Hong Wei, Winslett, Marianne, Fu, Tom Z. J., Xu, Boyan, Yang, Xiaoyan, Zhang, Zhenjie
Format: Artikel
Sprache:eng
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
container_end_page 21
container_issue 2
container_start_page 1
container_title ACM transactions on intelligent systems and technology
container_volume 12
creator Li, Zijian
Cai, Ruichu
Ng, Hong Wei
Winslett, Marianne
Fu, Tom Z. J.
Xu, Boyan
Yang, Xiaoyan
Zhang, Zhenjie
description Data-driven models are becoming essential parts in modern mechanical systems, commonly used to capture the behavior of various equipment and varying environmental characteristics. Despite the advantages of these data-driven models on excellent adaptivity to high dynamics and aging equipment, they are usually hungry for massive labels, mostly contributed by human engineers at a high cost. Fortunately, domain adaptation enhances the model generalization by utilizing the labeled source data and the unlabeled target data. However, the mainstream domain adaptation methods cannot achieve ideal performance on time series data, since they assume that the conditional distributions are equal. This assumption works well in the static data but is inapplicable for the time series data. Even the first-order Markov dependence assumption requires the dependence between any two consecutive time steps. In this article, we assume that the causal mechanism is invariant and present our Causal Mechanism Transfer Network (CMTN) for time series domain adaptation. By capturing causal mechanisms of time series data, CMTN allows the data-driven models to exploit existing data and labels from similar systems, such that the resulting model on a new system is highly reliable even with limited data. We report our empirical results and lessons learned from two real-world case studies, on chiller plant energy optimization and boiler fault detection, which outperform the existing state-of-the-art method.
doi_str_mv 10.1145/3445033
format Article
fullrecord <record><control><sourceid>crossref</sourceid><recordid>TN_cdi_crossref_primary_10_1145_3445033</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>10_1145_3445033</sourcerecordid><originalsourceid>FETCH-LOGICAL-c225t-a207ba4352920479098b945d9509517fd6f1bf58955aa9cdbf623f5886e5704c3</originalsourceid><addsrcrecordid>eNo9kD1PwzAYhC0EElWp-AvemAL-epN4rMJHkQoMDQNT9MaxhaGJKzsI9d8TRMQtd89wNxwhl5xdc67gRioFTMoTshAciizXXJz-Z6bOySqlDzZJaaF5uSBvFX4l3NMna95x8KmndcQhORvpsx2_Q_ykLkRa-97SnY3eJnobevQDXXd4GHH0YaATzX0zTe2OabR9uiBnDvfJrmZfktf7u7raZNuXh8dqvc2MEDBmKFjRopIgtGCq0EyXrVbQaWAaeOG63PHWQakBELXpWpcLOXGZWyiYMnJJrv52TQwpReuaQ_Q9xmPDWfN7SjOfIn8AJ3JSnA</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Causal Mechanism Transfer Network for Time Series Domain Adaptation in Mechanical Systems</title><source>ACM Digital Library Complete</source><creator>Li, Zijian ; Cai, Ruichu ; Ng, Hong Wei ; Winslett, Marianne ; Fu, Tom Z. J. ; Xu, Boyan ; Yang, Xiaoyan ; Zhang, Zhenjie</creator><creatorcontrib>Li, Zijian ; Cai, Ruichu ; Ng, Hong Wei ; Winslett, Marianne ; Fu, Tom Z. J. ; Xu, Boyan ; Yang, Xiaoyan ; Zhang, Zhenjie</creatorcontrib><description>Data-driven models are becoming essential parts in modern mechanical systems, commonly used to capture the behavior of various equipment and varying environmental characteristics. Despite the advantages of these data-driven models on excellent adaptivity to high dynamics and aging equipment, they are usually hungry for massive labels, mostly contributed by human engineers at a high cost. Fortunately, domain adaptation enhances the model generalization by utilizing the labeled source data and the unlabeled target data. However, the mainstream domain adaptation methods cannot achieve ideal performance on time series data, since they assume that the conditional distributions are equal. This assumption works well in the static data but is inapplicable for the time series data. Even the first-order Markov dependence assumption requires the dependence between any two consecutive time steps. In this article, we assume that the causal mechanism is invariant and present our Causal Mechanism Transfer Network (CMTN) for time series domain adaptation. By capturing causal mechanisms of time series data, CMTN allows the data-driven models to exploit existing data and labels from similar systems, such that the resulting model on a new system is highly reliable even with limited data. We report our empirical results and lessons learned from two real-world case studies, on chiller plant energy optimization and boiler fault detection, which outperform the existing state-of-the-art method.</description><identifier>ISSN: 2157-6904</identifier><identifier>EISSN: 2157-6912</identifier><identifier>DOI: 10.1145/3445033</identifier><language>eng</language><ispartof>ACM transactions on intelligent systems and technology, 2021-03, Vol.12 (2), p.1-21</ispartof><lds50>peer_reviewed</lds50><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c225t-a207ba4352920479098b945d9509517fd6f1bf58955aa9cdbf623f5886e5704c3</citedby><cites>FETCH-LOGICAL-c225t-a207ba4352920479098b945d9509517fd6f1bf58955aa9cdbf623f5886e5704c3</cites><orcidid>0000-0001-8972-167X</orcidid></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>314,780,784,27924,27925</link.rule.ids></links><search><creatorcontrib>Li, Zijian</creatorcontrib><creatorcontrib>Cai, Ruichu</creatorcontrib><creatorcontrib>Ng, Hong Wei</creatorcontrib><creatorcontrib>Winslett, Marianne</creatorcontrib><creatorcontrib>Fu, Tom Z. J.</creatorcontrib><creatorcontrib>Xu, Boyan</creatorcontrib><creatorcontrib>Yang, Xiaoyan</creatorcontrib><creatorcontrib>Zhang, Zhenjie</creatorcontrib><title>Causal Mechanism Transfer Network for Time Series Domain Adaptation in Mechanical Systems</title><title>ACM transactions on intelligent systems and technology</title><description>Data-driven models are becoming essential parts in modern mechanical systems, commonly used to capture the behavior of various equipment and varying environmental characteristics. Despite the advantages of these data-driven models on excellent adaptivity to high dynamics and aging equipment, they are usually hungry for massive labels, mostly contributed by human engineers at a high cost. Fortunately, domain adaptation enhances the model generalization by utilizing the labeled source data and the unlabeled target data. However, the mainstream domain adaptation methods cannot achieve ideal performance on time series data, since they assume that the conditional distributions are equal. This assumption works well in the static data but is inapplicable for the time series data. Even the first-order Markov dependence assumption requires the dependence between any two consecutive time steps. In this article, we assume that the causal mechanism is invariant and present our Causal Mechanism Transfer Network (CMTN) for time series domain adaptation. By capturing causal mechanisms of time series data, CMTN allows the data-driven models to exploit existing data and labels from similar systems, such that the resulting model on a new system is highly reliable even with limited data. We report our empirical results and lessons learned from two real-world case studies, on chiller plant energy optimization and boiler fault detection, which outperform the existing state-of-the-art method.</description><issn>2157-6904</issn><issn>2157-6912</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2021</creationdate><recordtype>article</recordtype><recordid>eNo9kD1PwzAYhC0EElWp-AvemAL-epN4rMJHkQoMDQNT9MaxhaGJKzsI9d8TRMQtd89wNxwhl5xdc67gRioFTMoTshAciizXXJz-Z6bOySqlDzZJaaF5uSBvFX4l3NMna95x8KmndcQhORvpsx2_Q_ykLkRa-97SnY3eJnobevQDXXd4GHH0YaATzX0zTe2OabR9uiBnDvfJrmZfktf7u7raZNuXh8dqvc2MEDBmKFjRopIgtGCq0EyXrVbQaWAaeOG63PHWQakBELXpWpcLOXGZWyiYMnJJrv52TQwpReuaQ_Q9xmPDWfN7SjOfIn8AJ3JSnA</recordid><startdate>20210301</startdate><enddate>20210301</enddate><creator>Li, Zijian</creator><creator>Cai, Ruichu</creator><creator>Ng, Hong Wei</creator><creator>Winslett, Marianne</creator><creator>Fu, Tom Z. J.</creator><creator>Xu, Boyan</creator><creator>Yang, Xiaoyan</creator><creator>Zhang, Zhenjie</creator><scope>AAYXX</scope><scope>CITATION</scope><orcidid>https://orcid.org/0000-0001-8972-167X</orcidid></search><sort><creationdate>20210301</creationdate><title>Causal Mechanism Transfer Network for Time Series Domain Adaptation in Mechanical Systems</title><author>Li, Zijian ; Cai, Ruichu ; Ng, Hong Wei ; Winslett, Marianne ; Fu, Tom Z. J. ; Xu, Boyan ; Yang, Xiaoyan ; Zhang, Zhenjie</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c225t-a207ba4352920479098b945d9509517fd6f1bf58955aa9cdbf623f5886e5704c3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2021</creationdate><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Li, Zijian</creatorcontrib><creatorcontrib>Cai, Ruichu</creatorcontrib><creatorcontrib>Ng, Hong Wei</creatorcontrib><creatorcontrib>Winslett, Marianne</creatorcontrib><creatorcontrib>Fu, Tom Z. J.</creatorcontrib><creatorcontrib>Xu, Boyan</creatorcontrib><creatorcontrib>Yang, Xiaoyan</creatorcontrib><creatorcontrib>Zhang, Zhenjie</creatorcontrib><collection>CrossRef</collection><jtitle>ACM transactions on intelligent systems and technology</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Li, Zijian</au><au>Cai, Ruichu</au><au>Ng, Hong Wei</au><au>Winslett, Marianne</au><au>Fu, Tom Z. J.</au><au>Xu, Boyan</au><au>Yang, Xiaoyan</au><au>Zhang, Zhenjie</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Causal Mechanism Transfer Network for Time Series Domain Adaptation in Mechanical Systems</atitle><jtitle>ACM transactions on intelligent systems and technology</jtitle><date>2021-03-01</date><risdate>2021</risdate><volume>12</volume><issue>2</issue><spage>1</spage><epage>21</epage><pages>1-21</pages><issn>2157-6904</issn><eissn>2157-6912</eissn><abstract>Data-driven models are becoming essential parts in modern mechanical systems, commonly used to capture the behavior of various equipment and varying environmental characteristics. Despite the advantages of these data-driven models on excellent adaptivity to high dynamics and aging equipment, they are usually hungry for massive labels, mostly contributed by human engineers at a high cost. Fortunately, domain adaptation enhances the model generalization by utilizing the labeled source data and the unlabeled target data. However, the mainstream domain adaptation methods cannot achieve ideal performance on time series data, since they assume that the conditional distributions are equal. This assumption works well in the static data but is inapplicable for the time series data. Even the first-order Markov dependence assumption requires the dependence between any two consecutive time steps. In this article, we assume that the causal mechanism is invariant and present our Causal Mechanism Transfer Network (CMTN) for time series domain adaptation. By capturing causal mechanisms of time series data, CMTN allows the data-driven models to exploit existing data and labels from similar systems, such that the resulting model on a new system is highly reliable even with limited data. We report our empirical results and lessons learned from two real-world case studies, on chiller plant energy optimization and boiler fault detection, which outperform the existing state-of-the-art method.</abstract><doi>10.1145/3445033</doi><tpages>21</tpages><orcidid>https://orcid.org/0000-0001-8972-167X</orcidid></addata></record>
fulltext fulltext
identifier ISSN: 2157-6904
ispartof ACM transactions on intelligent systems and technology, 2021-03, Vol.12 (2), p.1-21
issn 2157-6904
2157-6912
language eng
recordid cdi_crossref_primary_10_1145_3445033
source ACM Digital Library Complete
title Causal Mechanism Transfer Network for Time Series Domain Adaptation in Mechanical Systems
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-23T05%3A59%3A13IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-crossref&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Causal%20Mechanism%20Transfer%20Network%20for%20Time%20Series%20Domain%20Adaptation%20in%20Mechanical%20Systems&rft.jtitle=ACM%20transactions%20on%20intelligent%20systems%20and%20technology&rft.au=Li,%20Zijian&rft.date=2021-03-01&rft.volume=12&rft.issue=2&rft.spage=1&rft.epage=21&rft.pages=1-21&rft.issn=2157-6904&rft.eissn=2157-6912&rft_id=info:doi/10.1145/3445033&rft_dat=%3Ccrossref%3E10_1145_3445033%3C/crossref%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true