Fault prediction algorithm of slow variable signal based on multidimensional data driving

Yifan Li, Ping Song, Hongbo Liu, Chuangbo Hao

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

Abstract

With the increase of fault history data, the problem of high precision and long-time fault prediction under different failure modes is presented. We propose a multi-channel fusion fault prediction algorithm based on Long Short-Term Memory (LSTM) deep network. The prediction ability of the algorithm increases with the increase of training samples. Based on the analysis of the influence of different network parameters on the prediction accuracy, the optimal network parameters are selected to realize the long-time and high-precision fault prediction. It can recognize fault prediction without historical data. And it can integrate multi-channel information for off-line training to achieve the goal of self-increasing fault prediction ability.

Original languageEnglish
Title of host publication2022 Cross Strait Radio Science and Wireless Technology Conference, CSRSWTC 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665460965
DOIs
Publication statusPublished - 2022
Event2022 Cross Strait Radio Science and Wireless Technology Conference, CSRSWTC 2022 - Haidian, China
Duration: 17 Dec 202218 Dec 2022

Publication series

Name2022 Cross Strait Radio Science and Wireless Technology Conference, CSRSWTC 2022

Conference

Conference2022 Cross Strait Radio Science and Wireless Technology Conference, CSRSWTC 2022
Country/TerritoryChina
CityHaidian
Period17/12/2218/12/22

Keywords

  • Long Short-Term Memory (LSTM)
  • Loss Function
  • RNN
  • Remaining Useful Life RUL

Fingerprint

Dive into the research topics of 'Fault prediction algorithm of slow variable signal based on multidimensional data driving'. Together they form a unique fingerprint.

Cite this