Skip to main navigation Skip to search Skip to main content

Failure rate prediction based on AR model and residual correction

  • Beihang University

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Based on the study of advantages and disadvantages of the traditional AR model (autoregressive model) and the characteristics of failure rate prediction, an AR model based on neural network residual correction is proposed in this paper. The basic idea is to establish the AR model first to obtain the residual sequence, and next construct the neural network residual prediction model using the residual sequence, and then correct the predicted value of the original AR model using the residual value predicted by the model. The combined model is used to predict the failure rate of a kind of Boeing aircraft. It is proved that this model is suitable for short-Term failure rate prediction, and the accuracy of the prediction results is better than that of the single AR model.

Original languageEnglish
Title of host publication2017 2nd International Conference on Reliability Systems Engineering, ICRSE 2017
EditorsDongming Fan, Jun Yang, Ziyao Wang, Tingdi Zhao
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781538609187
DOIs
StatePublished - 8 Sep 2017
Event2nd International Conference on Reliability Systems Engineering, ICRSE 2017 - Huairou, Beijing, China
Duration: 10 Jul 201712 Jul 2017

Publication series

Name2017 2nd International Conference on Reliability Systems Engineering, ICRSE 2017

Conference

Conference2nd International Conference on Reliability Systems Engineering, ICRSE 2017
Country/TerritoryChina
CityHuairou, Beijing
Period10/07/1712/07/17

Keywords

  • AR model
  • failure rate
  • neural network
  • time series

Fingerprint

Dive into the research topics of 'Failure rate prediction based on AR model and residual correction'. Together they form a unique fingerprint.

Cite this