Automobile parts reliability prediction based on claim data: The comparison of predictive effects with deep learning
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Lee, Jun-Guel | - |
dc.contributor.author | Kim, Taehyeong | - |
dc.contributor.author | Sung, Ki Woo | - |
dc.contributor.author | Han, Sung Won | - |
dc.date.accessioned | 2022-02-15T15:42:15Z | - |
dc.date.available | 2022-02-15T15:42:15Z | - |
dc.date.created | 2022-02-08 | - |
dc.date.issued | 2021-11 | - |
dc.identifier.issn | 1350-6307 | - |
dc.identifier.uri | https://scholar.korea.ac.kr/handle/2021.sw.korea/135878 | - |
dc.description.abstract | An automobile company provides quality assurance service based on warranty period, which includes duration and distance driven. The concerned automobile company provides a warranty that begets quality assurance costs in case an accident occurs during this time. If the probability of failure can be ascertained based on the quality assurance area, it will be convenient to manage each part, and it will reduce the cost of quality assurance. It has been reported that failure data is required to predict reliability, thus, claim data containing information about the cause of failure, parts involved, number of days of use, mileage was used in this study. Because these claim data depend on unpredictable accidents, they can be intermittent and irregular. However, the recent accumulation of claim data has provided further opportunities to improve our ability to predict reliability. We collected up to 305,965 claims for a given automobile model over a period of 6-14 years. Using these data, we devise various deep learning model to predict the number of failures and estimate reliability in the presence of these number of failures using the following deep learning methods: 1D convolutional neural network, recurrent neural network (RNN), and sequence to sequence(Seq2Seq). Further, various approaches were used to compare the effectiveness of the proposed models: parametric methods, including Weibull distribution assumption; time-series methods including ARIMA model; and machine learning methods including support vector machine and random forest. Of all the models we tested, the proposed RNN model in this work produces superior performance in failure and reliability prediction. Using this method, an automobile company can significantly reduce costs by predicting the problems of new vehicle models in advance by securing reliability, as well as an accurate prediction of the number of future failures of each automobile part. | - |
dc.language | English | - |
dc.language.iso | en | - |
dc.publisher | PERGAMON-ELSEVIER SCIENCE LTD | - |
dc.subject | MODEL | - |
dc.subject | TIME | - |
dc.subject | FAILURE | - |
dc.title | Automobile parts reliability prediction based on claim data: The comparison of predictive effects with deep learning | - |
dc.type | Article | - |
dc.contributor.affiliatedAuthor | Han, Sung Won | - |
dc.identifier.doi | 10.1016/j.engfailanal.2021.105657 | - |
dc.identifier.scopusid | 2-s2.0-85113355321 | - |
dc.identifier.wosid | 000700926600002 | - |
dc.identifier.bibliographicCitation | ENGINEERING FAILURE ANALYSIS, v.129 | - |
dc.relation.isPartOf | ENGINEERING FAILURE ANALYSIS | - |
dc.citation.title | ENGINEERING FAILURE ANALYSIS | - |
dc.citation.volume | 129 | - |
dc.type.rims | ART | - |
dc.type.docType | Article | - |
dc.description.journalClass | 1 | - |
dc.description.journalRegisteredClass | scie | - |
dc.description.journalRegisteredClass | scopus | - |
dc.relation.journalResearchArea | Engineering | - |
dc.relation.journalResearchArea | Materials Science | - |
dc.relation.journalWebOfScienceCategory | Engineering, Mechanical | - |
dc.relation.journalWebOfScienceCategory | Materials Science, Characterization & Testing | - |
dc.subject.keywordPlus | FAILURE | - |
dc.subject.keywordPlus | MODEL | - |
dc.subject.keywordPlus | TIME | - |
dc.subject.keywordAuthor | Automobile claim data | - |
dc.subject.keywordAuthor | Convolutional Neural Network | - |
dc.subject.keywordAuthor | Deep learning | - |
dc.subject.keywordAuthor | Recurrent Neural Network | - |
dc.subject.keywordAuthor | Reliability prediction | - |
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.
(02841) 서울특별시 성북구 안암로 14502-3290-1114
COPYRIGHT © 2021 Korea University. All Rights Reserved.
Certain data included herein are derived from the © Web of Science of Clarivate Analytics. All rights reserved.
You may not copy or re-distribute this material in whole or in part without the prior written consent of Clarivate Analytics.