PSI - Issue 68

Oleh Yasniy et al. / Procedia Structural Integrity 68 (2025) 132–138 O. Yasniy et al. / Structural Integrity Procedia 00 (2025) 000–000

135

4

a

b

Fig. 3. The predicted versus true (experimental) values of the strain, obtained using the SVM method for the period of loading (a) and unloading (b) of the SMA material. Fig. 4 shows the relationship between the experimental values of material strain ! ! " $ #%! $ and the predicted values &!$ " % $ ! '. for KNN. a b

Fig. 4. The predicted versus true (experimental) values of the strain, obtained using the KNN method for the period of loading (a) and unloading (b) of the SMA material.

To predict the up direction, the Multi-Layer Perceptron (MLP) ANN by Rosenblatt (1961) 2-13-1 was utilized. The input layer contained two neurons corresponding to two input parameters. The hidden layer, consisting of 13 neurons, had a Tangential activation function (hyperbolic tangent). The output layer, which is designed to predict the material strain e , contained one neuron with a linear activation function (Linear). To predict the unloading period (down direction), we used ANN MLP 2-100-1 with similar activation functions. Fig. 5 shows the relationship between the experimental values of material strain ! ! " $ #%! $ and the predicted values &!$ " % $ ! '. for ANN.