Experimental force reconstruction using a neural network and simulated training data

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Force reconstruction is the determination of unknown applied forces using operational vibration information. This inverse problem is made more difficult by limiting the number of operational sensors and increasing the number of modes excited in a given frequency range. Artificial neural networks have been generated to do a wide variety of tasks in vibrations, including inverse problems. However, the amount of training data required is usually too much for a single experiment to generate. In this research, a neural network was generated to estimate the location of a hammer impact on a plate using accelerometer data. Instead of experimental data, however, the network was trained on a finite element model of the structure. To account for variations between the model and reality, the training data was generated with a range of physical properties, such as material stiffness and damping. The resulting model was able to identify the location of a hammer impact to within approximately 12% of the size of the plate.

Original languageEnglish (US)
Title of host publicationProceedings of 2020 International Congress on Noise Control Engineering, INTER-NOISE 2020
EditorsJin Yong Jeon
PublisherKorean Society of Noise and Vibration Engineering
ISBN (Electronic)9788994021362
StatePublished - Aug 23 2020
Event49th International Congress and Exposition on Noise Control Engineering, INTER-NOISE 2020 - Seoul, Korea, Republic of
Duration: Aug 23 2020Aug 26 2020

Publication series

NameProceedings of 2020 International Congress on Noise Control Engineering, INTER-NOISE 2020

Conference

Conference49th International Congress and Exposition on Noise Control Engineering, INTER-NOISE 2020
Country/TerritoryKorea, Republic of
CitySeoul
Period8/23/208/26/20

All Science Journal Classification (ASJC) codes

  • Acoustics and Ultrasonics

Fingerprint

Dive into the research topics of 'Experimental force reconstruction using a neural network and simulated training data'. Together they form a unique fingerprint.

Cite this