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

133

2

phase transformations that occur in the material during such impacts and are the basis of two key effects: the shape memory effect (SME) and superelasticity (SE). Due to these properties, shape memory alloys have been widely used in various industries, such as aviation by Quan and Hai (2015), medicine by Abitha et al. (2020), robotics by Yang et al. (2024), and automotive and construction by Vahedi et al. (2024). The superelasticity of SMA is manifested in the ability of the material to return to its original shape after the mechanical loading was removed. This is due to the phase transformation between austenite and martensite under load, as well as the reverse transformation during unloading. Phase transformations in such alloys are accompanied by significant hysteresis, which is clearly visible in the stress-strain diagram during loading and unloading cycles of SMA samples. The hysteresis loops show a complex nonlinear relationship between stress and strain, which is difficult to approximate with traditional mathematical models. Machine learning (ML) is a part of artificial intelligence that allows computer systems to learn from data and improve their predictions. The basic idea behind ML is that algorithms can detect patterns and structures in data, which are used to make predictions or automate tasks. Machine learning is used in the financial sector to analyses credit risk, detect fraud, and automate trading algorithms by Silva et al. (2024), in medicine to diagnose and predict treatment by Melchane et al. (2024), in the automotive industry for the development of autonomous vehicles and maintenance forecasting by Lyashuk et al. (2024), in materials science for predicting material properties and optimizing their composition by Yasniy et al. (2024), Yasniy et al. (2020), Didych et al. (2018), and in cybersecurity for threat detection by Klots et al. (2023), by Petliak et al. (2023).

Nomenclature s stress e strain N

number of the loading and unloading cycle n volume of the test data set ! ! " $ #%! $ ( ) true value of material strain in the test data set &!$ " % $ ! '. ( ) predicted value of material strain in the test data set 2. Material and methods

One of the most common shape memory alloys is nickel-titanium alloys (NiTi), also known as Nitinol. In this paper, machine learning methods such as boosted trees, random forests, support vector machines (SVM), k-nearest neighbors (KNN) and artificial neural networks (ANN) were utilized to predict the strain of NiTi alloy as a function of the applied stress. The functional relationships were modelled on the basis of experimental data by Iasnii et al. (2022). The input data for the machine learning included the stress value s (MPa) and the number of loading-unloading cycles N , while the output parameter was the material strain e (%). For the training and testing process, we used data from 100-120 loading and unloading cycles of the SMA material. The data set was divided into two unequal parts: training and test samples. In total, the sample contained 1089 elements for the up-direction loading period, of which 80% were randomly selected for training and 20% for testing and evaluation of the model's prediction quality. For the down direction, the sample included 1032 items, which were divided in the same way. The mean absolute percentage error (MAPE) was chosen as the forecasting error, which was calculated using the formula: = * ) ∑ +, ! ! "$ #%! $ (.)0, &!$ " % $ ! '. (.)+ 1, ! ! "$ #%! $ (.)1 * .2) ∙ 100% (1) The Mean Absolute Percentage Error (MAPE) is a popular metric for evaluating the accuracy of predictions in regression tasks. It measures the average percentage difference between the actual and predicted values, which allows you to understand how well the model predicts. The smaller the MAPE value, the more accurate the forecast. A MAPE value close to zero indicates that the model predicts the actual data very well.