Predictive modeling of phase transformation temperatures in NiTiCu shape memory alloys: Integrating electronic factors through artificial neural network

Abstract

The primary objective of this research is to investigate the correlation between the electronic factors of NiTiCu Shape memory alloy (SMA) on phase transformation temperature (PTT) and thermal hysteresis. The study yields valuable insights into understanding the phase transition behavior of NiTiCu shape memory alloys, emphasizing the intricate interplay between electronic factors and alloy composition on PTT. Theoretical analyses highlight the critical role of considering electronic structure in designing shape memory alloys. A pivotal aspect of this study is the development of an artificial neural network (ANN) model to predict PTT, encompassing austenite start (As), austenite finish (Af), martensite start (Ms), martensite finish (Mf), and thermal hysteresis is calculated from the predicted PTT. This research draws on datasets from prior studies on diverse ternary NiTiCu SMA compositions, focusing exclusively on homogenized, solution-treated data to mitigate precipitate-induced impacts on PTT. The proposed ANN model, configured with 9–14–4 neurons, demonstrates precise and reliable predictions for PTT in NiTiCu. The model's robust performance, characterized by a strong correlation (high R values), low relative errors within ±1% from the Gaussian distribution, and successful validation, underlines its applicability for predicting PTT across diverse NiTiCu compositions. Furthermore, experimental validations for unseen NiTiCu compositions affirm the model's reliability and robust performance in forecasting PTT in NiTiCu SMAs.