PSI - Issue 48

Oleh Yasniy et al. / Procedia Structural Integrity 48 (2023) 149–154 Yasniy et al/ Structural Integrity Procedia 00 (2023) 000–000

153

5

4. Results and discussion Dependencies between experimental and predicted values of jump-like creep were plotted by the method of neural networks (Fig. 4), whereas dependencies between the tensile stress on the value of jump-like increments of creep log10(  )  log10(  p) are presented in Fig. 5. The obtained prediction results are in good correlation with the experimental data. It was determined that the predicted error of the machine learning method, particularly neural networks, was equal to 4.8%.

Fig. 4. Predicted log10 (  p prediction ) and experimental log10(  p true ) values of jump-like creep obtained by the method of neural networks by Didych et al. (2022)

Fig. 5. Predicted and experimental dependencies between the corresponding tensile stress and the value of jump-like creep obtained by the method of neural networks by Didych et al. (2022)

5. Conclusions Method selection and dataset are important to obtain good prediction results. In addition, methods of supervised learning usually show better results than other methods. It was found that the prediction accuracy of the neural networks method is 96.2%. In particular, the error between the results obtained by FE modeling and experimental data does not exceed 12%. Therefore, comparing these results, it follows that machine learning is a powerful tool that can be used to solve mechanics problems. References Yasnii, P. V., Fedak, S. I., Glad’o, V. B., Galushchak, M. P., 2004. Jumplike Deformation in AMg6 Aluminum Alloy in Tension. Strength of Materials 36, 113–118. Yasnii, P. V., Glad’o, V. B., 2002. The influence of the cyclic tensile loading component on the dislocation structure in AMg6 alloy. Fiz.-Khim. Mekh. Mater. 3, 63–68. Yasnii, P.V., Glad’o, V.B. & Fedak, S.I., 2004. Relationship between plastic strains and microstructural parameters of AMg6 alloy under conditions of active tension and creep. Strength of Materials 36, 582–590. Yasniy, P., Hlado, V., Shulhan, I., Fedak, S., Lapusta, Yu., 2010. Modeling of discontinuous deformation in Al-6%Mg alloy. 18 th European Conference on Fracture. Dresden. Germany, 364. Javaheri, E., Kumala, V., Javaheri, A., Rawassizadeh, R., Lubritz, J., Graf, B., Rethmeier, M., 2020. Quantifying Mechanical Properties of Automotive Steels with Deep Learning Based Computer Vision Algorithms. Metals 10 , 163. Ghatak, A., and Robi, P. S., 2018. Prediction of creep curve of HP40Nb steel using artificial neural network. Neural Computing and Applications 30, 2953-2964. Fedak, S., 2003. Jumplike deformation of the AMg6 alloy during creep. Visn. Ternopil. Derzh. Tekhn. Univ. 8, 16–23. [in Ukrainian]