Impact load identification base on LSTM neural network

Shuya Liang, Te Yang, Zaifei Kang, Zhichun Yang

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

Abstract

It is difficult to obtain the accurate dynamic model of engineering structures and it is also impossible to directly measure the impact load exerted on the structures. To solve this problem, this paper uses Long Short-Term Memory (LSTM) neural network to identify the impact load on a vertical tail structure model. Firstly, combining with the “memory” characteristics of LSTM neural network and the relationship between impact load and responses of the vertical tail model structure, a time domain identification method of impact load based on LSTM neural network is proposed. A finite element model of vertical tail structure model is taken as research objects to identify the impact load. Furthermore, the impact load identification experiments are performed on the vertical tail structure model, the results show that the average peak error of the impact load identified by the proposed method is 3.49%, and the load time history identified by the proposed method has a high degree of coincidence with the real one, which verifies the effectiveness of the proposed impact load identification method.

Original languageEnglish
Title of host publication"Advances in Acoustics, Noise and Vibration - 2021" Proceedings of the 27th International Congress on Sound and Vibration, ICSV 2021
EditorsEleonora Carletti, Malcolm Crocker, Marek Pawelczyk, Jiri Tuma
PublisherSilesian University Press
ISBN (Electronic)9788378807995
StatePublished - 2021
Event27th International Congress on Sound and Vibration, ICSV 2021 - Virtual, Online
Duration: 11 Jul 202116 Jul 2021

Publication series

Name"Advances in Acoustics, Noise and Vibration - 2021" Proceedings of the 27th International Congress on Sound and Vibration, ICSV 2021
ISSN (Print)2329-3675

Conference

Conference27th International Congress on Sound and Vibration, ICSV 2021
CityVirtual, Online
Period11/07/2116/07/21

Keywords

  • Deep learning
  • Impact load
  • Load identification
  • LSTM neural network

Fingerprint

Dive into the research topics of 'Impact load identification base on LSTM neural network'. Together they form a unique fingerprint.

Cite this