Deep Graph Convolutional Autoencoder With Conditional Normalizing Flow for Power Distribution Systems Fault Classification and Location

Accurate fault classification and location are critical to ensure the reliability and resilience of large-scale power distribution systems (PDSs). The existing data-driven works in this area struggle to capture essential space-time correlations of PDS measurements and often rely on deterministic and...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Veröffentlicht in:	IEEE transactions on artificial intelligence Jg. 6; H. 9; S. 2448 - 2463
Hauptverfasser:	Saffari, Mohsen, Khodayar, Mahdi, Khodayar, Mohammad E., Fazlhashemi, Seyed Saeed
Format:	Journal Article
Sprache:	Englisch
Veröffentlicht:	IEEE 01.09.2025
Schlagworte:	Accuracy Artificial intelligence Conditional normalizing flow (CNF) Convolution Convolutional neural networks Current measurement Decoding deep convolution network deep spars architectures fault classification Fault location Feature extraction Logic gates Transmission line measurements
ISSN:	2691-4581, 2691-4581
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Abstract	Accurate fault classification and location are critical to ensure the reliability and resilience of large-scale power distribution systems (PDSs). The existing data-driven works in this area struggle to capture essential space-time correlations of PDS measurements and often rely on deterministic and shallow neural architectures. Furthermore, they encounter challenges such as over-smoothing and the inability to capture deep correlations. To overcome these limitations, a novel deep space-time generative graph convolutional autoencoder (SGGCA) is proposed. First, the PDS is modeled as a space-time graph where the nodes and edges show the bus measurements and line impedance values, respectively. The proposed SGGCA's encoder captures deep correlations of the space-time graph using a new graph convolution with early connections and identity transformations to mitigate the over-smoothing. Our encoder encompasses a new recurrent method to adjust graph convolution parameters without relying on node embeddings on the temporal dimension. Additionally, it incorporates generative modeling by capturing the probability distribution function of the latent representation through a conditional normalizing flow model. The extracted generative space-time features are enhanced by a multi-head attention mechanism to better capture task-relevant characteristics of the PDS measurements. The extracted features are fed to sparse decoders to classify and locate the faults in the PDS. The feature sparsity of decoders ensures a high generalization capacity and avoids overfitting. The proposed method is evaluated on the IEEE 69-bus and 123-bus systems. It achieves substantial improvements in fault classification accuracy by 3.33% and 6.26% and enhances fault location accuracy by 6.33% and 5.73% for the respective PDSs compared with state-of-the-art models.
AbstractList	Accurate fault classification and location are critical to ensure the reliability and resilience of large-scale power distribution systems (PDSs). The existing data-driven works in this area struggle to capture essential space-time correlations of PDS measurements and often rely on deterministic and shallow neural architectures. Furthermore, they encounter challenges such as over-smoothing and the inability to capture deep correlations. To overcome these limitations, a novel deep space-time generative graph convolutional autoencoder (SGGCA) is proposed. First, the PDS is modeled as a space-time graph where the nodes and edges show the bus measurements and line impedance values, respectively. The proposed SGGCA's encoder captures deep correlations of the space-time graph using a new graph convolution with early connections and identity transformations to mitigate the over-smoothing. Our encoder encompasses a new recurrent method to adjust graph convolution parameters without relying on node embeddings on the temporal dimension. Additionally, it incorporates generative modeling by capturing the probability distribution function of the latent representation through a conditional normalizing flow model. The extracted generative space-time features are enhanced by a multi-head attention mechanism to better capture task-relevant characteristics of the PDS measurements. The extracted features are fed to sparse decoders to classify and locate the faults in the PDS. The feature sparsity of decoders ensures a high generalization capacity and avoids overfitting. The proposed method is evaluated on the IEEE 69-bus and 123-bus systems. It achieves substantial improvements in fault classification accuracy by 3.33% and 6.26% and enhances fault location accuracy by 6.33% and 5.73% for the respective PDSs compared with state-of-the-art models.
Author	Saffari, Mohsen Khodayar, Mahdi Khodayar, Mohammad E. Fazlhashemi, Seyed Saeed
Author_xml	– sequence: 1 givenname: Mohsen orcidid: 0000-0002-8336-8542 surname: Saffari fullname: Saffari, Mohsen email: msaffari@pnw.edu organization: Department of Electrical and Computer Engineering, Purdue University Northwest, Hammond, IN, USA – sequence: 2 givenname: Mahdi orcidid: 0000-0003-4683-7810 surname: Khodayar fullname: Khodayar, Mahdi email: mahdi-khodayar@utulsa.edu organization: Department of Computer Science, University of Tulsa, Tulsa, OK, USA – sequence: 3 givenname: Mohammad E. orcidid: 0000-0003-3856-5704 surname: Khodayar fullname: Khodayar, Mohammad E. email: mkhodayar@smu.edu organization: Department of Electrical and Computer Engineering, Southern Methodist University, Dallas, TX, USA – sequence: 4 givenname: Seyed Saeed orcidid: 0000-0003-4333-859X surname: Fazlhashemi fullname: Fazlhashemi, Seyed Saeed email: sfazlhashemi@smu.edu organization: Department of Electrical and Computer Engineering, Southern Methodist University, Dallas, TX, USA
BookMark	eNpNkMtOwkAUhidGExHZu3AxL1A8M23pdEmKIAlREzEum9PpqY4pHTJTJPgCvrblsmB1_pP_svhu2GVjG2LsTsBQCEgfluP5UIKMh2EcJSpRF6wnR6kIoliJyzN9zQbefwN0USGlTHrsb0K05jOH6y-e2ebH1pvW2AZrPt60lhptS3L8w7QHuzQn89m6Fdbm1zSffFrbLa-s469222UnxrfOFIcZ_rbzLa08n-KmbnlWo_emMhoPJjYlX9jjc8uuKqw9DU63z96nj8vsKVi8zObZeBFoEco2IK2TCkoA0gkSCY1S4YgUQggQRzKKRFEUMcSlKhCoUhFK0AhYUBoLUYV9Bsdd7az3jqp87cwK3S4XkO9Z5h3LfM8yP7HsKvfHiiGis3gK6Uio8B-PJ3W-
CODEN	ITAICB
Cites_doi	10.1109/ONCON56984.2022.10126859 10.1109/TPWRD.2023.3268201 10.1109/TIA.2021.3083645 10.5555/3454287.3455008 10.3390/pr11082494 10.1109/ACCESS.2014.2323353 10.1016/j.tej.2022.107137 10.1002/ese3.1573 10.1016/j.energy.2018.06.111 10.1109/BDICN58493.2023.00039 10.1002/eng2.12950 10.1109/ISGT50606.2022.9817473 10.1016/j.epsr.2022.108085 10.1109/JSAC.2019.2951964 10.1016/j.epsr.2023.109998 10.1016/j.ijepes.2013.09.011 10.1017/ATSIP.2020.13 10.3390/app122311903 10.1561/2200000056 10.1109/PTC.2017.7980907 10.1007/s00202-021-01223-7 10.1016/j.ijepes.2021.107102 10.5121/ijdkp.2015.5201 10.1016/j.rser.2017.03.021 10.1007/s00521-021-06541-2 10.1109/ISGT59692.2024.10454243 10.1109/SEST50973.2021.9543326 10.1016/j.apenergy.2023.120932 10.1109/ICEEE2.2018.8391345 10.1016/j.asej.2023.102427 10.1016/B978-0-12-809633-8.20349-X 10.1109/TIM.2023.3238059 10.35833/MPCE.2022.000204 10.1016/j.measurement.2021.109330 10.1109/ACCESS.2020.2971582 10.1109/TNNLS.2023.3280078 10.1109/TPWRD.2018.2875598 10.1016/j.epsr.2020.106914 10.1109/TIA.2018.2885045
ContentType	Journal Article
DBID	97E RIA RIE AAYXX CITATION
DOI	10.1109/TAI.2025.3547878
DatabaseName	IEEE All-Society Periodicals Package (ASPP) 2005–Present IEEE All-Society Periodicals Package (ASPP) 1998–Present IEEE Electronic Library (IEL) CrossRef
DatabaseTitle	CrossRef
DatabaseTitleList
Database_xml	– sequence: 1 dbid: RIE name: IEEE Electronic Library (IEL) url: https://ieeexplore.ieee.org/ sourceTypes: Publisher
DeliveryMethod	fulltext_linktorsrc
Discipline	Computer Science
EISSN	2691-4581
EndPage	2463
ExternalDocumentID	10_1109_TAI_2025_3547878 10909618
Genre	orig-research
GrantInformation_xml	– fundername: National Science Foundation grantid: ECCS-2223628; ECCS-2223629 funderid: 10.13039/100000001
GroupedDBID	0R~ 97E AASAJ AAWTH ABAZT ABJNI ABQJQ ABVLG AGQYO AHBIQ AKJIK AKQYR ALMA_UNASSIGNED_HOLDINGS ATWAV BEFXN BFFAM BGNUA BKEBE BPEOZ EBS IEDLZ IFIPE JAVBF M~E OCL RIA RIE AAYXX CITATION
ID	FETCH-LOGICAL-c132t-ecc7f0d00ec7aee1ca28a6e8a0300542441bbb505d8ba0ef84a20ca0abe9511f3
IEDL.DBID	RIE
ISSN	2691-4581
IngestDate	Sat Nov 29 07:35:57 EST 2025 Wed Sep 10 07:40:58 EDT 2025
IsPeerReviewed	true
IsScholarly	true
Issue	9
Language	English
License	https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html https://doi.org/10.15223/policy-029 https://doi.org/10.15223/policy-037
LinkModel	DirectLink
MergedId	FETCHMERGED-LOGICAL-c132t-ecc7f0d00ec7aee1ca28a6e8a0300542441bbb505d8ba0ef84a20ca0abe9511f3
ORCID	0000-0003-4333-859X 0000-0002-8336-8542 0000-0003-4683-7810 0000-0003-3856-5704
PageCount	16
ParticipantIDs	ieee_primary_10909618 crossref_primary_10_1109_TAI_2025_3547878
PublicationCentury	2000
PublicationDate	2025-Sept.
PublicationDateYYYYMMDD	2025-09-01
PublicationDate_xml	– month: 09 year: 2025 text: 2025-Sept.
PublicationDecade	2020
PublicationTitle	IEEE transactions on artificial intelligence
PublicationTitleAbbrev	TAI
PublicationYear	2025
Publisher	IEEE
Publisher_xml	– name: IEEE
References	ref13 ref12 ref15 ref14 ref31 ref30 ref11 ref10 ref2 ref1 Dinh (ref35) 2016 ref17 ref16 Oono (ref34) 2019 ref38 ref19 ref18 Maaten (ref46) 2008; 9 Rusch (ref33) 2023 (ref37) 2024 ref24 ref23 Kipf (ref32) 2016 ref45 ref26 ref25 Hossin (ref39) 2015; 5 ref20 ref42 ref41 ref22 ref44 ref21 ref43 ref28 ref27 ref29 ref8 ref7 ref9 ref4 ref3 ref6 ref5 Zhang (ref36) 2018; 31 ref40
References_xml	– year: 2023 ident: ref33 article-title: A survey on oversmoothing in graph neural networks – ident: ref45 doi: 10.1109/ONCON56984.2022.10126859 – volume: 9 start-page: 2579 issue: 11 year: 2008 ident: ref46 article-title: Visualizing data using T-SNE publication-title: J. Mach. Learn. Res. – ident: ref9 doi: 10.1109/TPWRD.2023.3268201 – ident: ref21 doi: 10.1109/TIA.2021.3083645 – ident: ref42 doi: 10.5555/3454287.3455008 – ident: ref1 doi: 10.3390/pr11082494 – ident: ref4 doi: 10.1109/ACCESS.2014.2323353 – ident: ref40 doi: 10.1016/j.tej.2022.107137 – ident: ref12 doi: 10.1002/ese3.1573 – ident: ref5 doi: 10.1016/j.energy.2018.06.111 – ident: ref44 doi: 10.1109/BDICN58493.2023.00039 – volume: 31 volume-title: Proc. Adv. neural Inf. Process. Syst. year: 2018 ident: ref36 article-title: Generalized cross entropy loss for training deep neural networks with noisy labels – ident: ref13 doi: 10.1002/eng2.12950 – ident: ref15 doi: 10.1109/ISGT50606.2022.9817473 – ident: ref19 doi: 10.1016/j.epsr.2022.108085 – ident: ref26 doi: 10.1109/JSAC.2019.2951964 – ident: ref6 doi: 10.1016/j.epsr.2023.109998 – year: 2019 ident: ref34 article-title: Graph neural networks exponentially lose expressive power for node classification – ident: ref11 doi: 10.1016/j.ijepes.2013.09.011 – ident: ref29 doi: 10.1017/ATSIP.2020.13 – ident: ref16 doi: 10.3390/app122311903 – ident: ref30 doi: 10.1561/2200000056 – year: 2016 ident: ref35 article-title: Density estimation using real NVP – ident: ref14 doi: 10.1109/PTC.2017.7980907 – ident: ref25 doi: 10.1007/s00202-021-01223-7 – ident: ref27 doi: 10.1016/j.ijepes.2021.107102 – volume: 5 start-page: 1 issue: 2 year: 2015 ident: ref39 article-title: A review on evaluation metrics for data classification evaluations publication-title: Int. J. Data Mining Knowl. Manage. Process doi: 10.5121/ijdkp.2015.5201 – ident: ref3 doi: 10.1016/j.rser.2017.03.021 – ident: ref7 doi: 10.1007/s00521-021-06541-2 – year: 2016 ident: ref32 article-title: Semi-supervised classification with graph convolutional networks – ident: ref20 doi: 10.1109/ISGT59692.2024.10454243 – ident: ref31 doi: 10.1109/SEST50973.2021.9543326 – ident: ref2 doi: 10.1016/j.apenergy.2023.120932 – ident: ref17 doi: 10.1109/ICEEE2.2018.8391345 – ident: ref23 doi: 10.1016/j.asej.2023.102427 – ident: ref38 doi: 10.1016/B978-0-12-809633-8.20349-X – ident: ref22 doi: 10.1109/TIM.2023.3238059 – ident: ref28 doi: 10.35833/MPCE.2022.000204 – ident: ref43 doi: 10.1016/j.measurement.2021.109330 – ident: ref24 doi: 10.1109/ACCESS.2020.2971582 – ident: ref41 doi: 10.1109/TNNLS.2023.3280078 – year: 2024 ident: ref37 article-title: Powerfactory – ident: ref8 doi: 10.1109/TPWRD.2018.2875598 – ident: ref18 doi: 10.1016/j.epsr.2020.106914 – ident: ref10 doi: 10.1109/TIA.2018.2885045
SSID	ssj0002512227
Score	2.340944
Snippet	Accurate fault classification and location are critical to ensure the reliability and resilience of large-scale power distribution systems (PDSs). The existing...
SourceID	crossref ieee
SourceType	Index Database Publisher
StartPage	2448
SubjectTerms	Accuracy Artificial intelligence Conditional normalizing flow (CNF) Convolution Convolutional neural networks Current measurement Decoding deep convolution network deep spars architectures fault classification Fault location Feature extraction Logic gates Transmission line measurements
Title	Deep Graph Convolutional Autoencoder With Conditional Normalizing Flow for Power Distribution Systems Fault Classification and Location
URI	https://ieeexplore.ieee.org/document/10909618
Volume	6
hasFullText	1
inHoldings	1
isFullTextHit
isPrint
journalDatabaseRights	– providerCode: PRVIEE databaseName: IEEE Electronic Library (IEL) customDbUrl: eissn: 2691-4581 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0002512227 issn: 2691-4581 databaseCode: RIE dateStart: 20200101 isFulltext: true titleUrlDefault: https://ieeexplore.ieee.org/ providerName: IEEE – providerCode: PRVHPJ databaseName: ROAD: Directory of Open Access Scholarly Resources customDbUrl: eissn: 2691-4581 dateEnd: 99991231 omitProxy: false ssIdentifier: ssj0002512227 issn: 2691-4581 databaseCode: M~E dateStart: 20200101 isFulltext: true titleUrlDefault: https://road.issn.org providerName: ISSN International Centre
link	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV07T8MwELZoxcBCeRRRHpUHFoa0TprUyVi1DSBB1aGIbpFfkSJVSVWSIjGw8rfxOS7qwsCWxLYU-bN1d_bd9yF0BxQhPo30_vYJd_wgGDhcsNQhcki5Ukrb-Fpsgs5m4XIZzW2xuqmF0e0m-Uz14NHc5ctCVHBU1ockQlAoaaAGpbQu1vo9UAFD7Xl0dxVJov5i9KQDQC_oDYC0CoTU9kzPnpaKMSVx658_cYKOrc-IRzXIp-hA5WeotdNjwHZ7nqPviVJr_AAU1Hhc5Fu7qmBoVRbAWCl197esNM0ys40zcFtX2ac2YjheFR9Y-7F4DuppeAK0ulYRC1tycxyzalVio6YJeUYGWsxyiZ-L-qWNXuPpYvzoWKEFR-hgtHQ0jDQlkhAlKFPKFcwL2VCFjACZPZTCuZxz7SvJkDOi0tBnHhGMMK60g-amgwvUzItcXSLs8yDVIZT2ypTre4JxKSOZ8jBgEoyk6KD7HQbJuubTSEwcQqJE45UAXonFq4PaMP17_eqZv_rj-zU6guF1BtgNapabSt2iQ7Ets_dNFzVevqZds2Z-AMjBxCE
linkProvider	IEEE
linkToHtml	http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV07T8MwELagIMFCeRRRnh5YGNI6adIkY1UIrShRhyK6RX5FilQlVUmKxB_gb-NzXNSFgS2JnSjyZ-vu7LvvQ-geKEJcP1Tr2yXMcj2vZzFOU4uIvs-klMrG12ITfhwH83k4NcXquhZGtevkM9mBS32WLwpewVZZF5IIQaFkF-15ruvYdbnW75YKmGrH8TeHkSTszgZjFQI6XqcHtFUgpbZlfLbUVLQxiZr__I1jdGS8RjyoYT5BOzI_Rc2NIgM2C_QMfT9KucTPQEKNh0W-NvMKXq3KAjgrher-npW6WWSmMQbHdZF9KTOGo0XxiZUni6egn4YfgVjXaGJhQ2-OI1otSqz1NCHTSIOLaS7wpKhvWugtepoNR5aRWrC4CkdLSwHpp0QQIrlPpbQ5dQLalwElQGcPxXA2Y0x5SyJglMg0cKlDOCWUSeWi2WnvHDXyIpcXCLvMS1UQpfwyabsOp0yIUKQs8KgAM8nb6GGDQbKsGTUSHYmQMFF4JYBXYvBqoxYM_1a_euQv_3h-hw5Gs9dJMhnHL1foED5V54Ndo0a5quQN2ufrMvtY3eqZ8wN668Y3
openUrl	ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Deep+Graph+Convolutional+Autoencoder+With+Conditional+Normalizing+Flow+for+Power+Distribution+Systems+Fault+Classification+and+Location&rft.jtitle=IEEE+transactions+on+artificial+intelligence&rft.au=Saffari%2C+Mohsen&rft.au=Khodayar%2C+Mahdi&rft.au=Khodayar%2C+Mohammad+E.&rft.au=Fazlhashemi%2C+Seyed+Saeed&rft.date=2025-09-01&rft.pub=IEEE&rft.eissn=2691-4581&rft.volume=6&rft.issue=9&rft.spage=2448&rft.epage=2463&rft_id=info:doi/10.1109%2FTAI.2025.3547878&rft.externalDocID=10909618
thumbnail_l	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=2691-4581&client=summon
thumbnail_m	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=2691-4581&client=summon
thumbnail_s	http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=2691-4581&client=summon