Master'sOpen Access

Prediction of self-loosening of bolts under the vibration via artificial neural networks

Is this your thesis?

This record came from a bulk archive import. If it’s yours, link it to your profile.

2019
0 views
0 downloads

Abstract (EN)

prediction of loosening rates of M8x100x1,25 – 6g 10.9 bolts under the transverse loading vibration has been tried via artificial neural networks. The two different types of coated bolts with different clamping lengths, different clamping loads, different types of coated nuts fastened and transverse loading vibration tests were conducted in all combinations. In result of vibration tests, effects of all researched parameters to self-loosening were found out. 70% of test results were used for training data for artificial neural networks, 30% of test results were used for test and validation. Approximately, 430.000 neural networks' performances were evaluated and highest performance network was detected. The highest performance neural network predicted the loosening rates to real test results closely and satisfying. MSE (Mean Square Error) criteria of predictions is 1,6652x10^(-5).

Author

Deniz Civan Özuğurlu

How to Cite

Deniz Civan Özuğurlu (Master Thesis). Prediction of self-loosening of bolts under the vibration via artificial neural networks, 2019, Ege University.

Keywords

License

Tüm Hakları Saklıdır

This work is shared under the specified license terms.

More theses from Ege University