Physics-Informed Machine Learning for Solder Joint Qualification Tests
S. D.M. de Jong (TU Delft - Electronic Components, Technology and Materials)
Amir Ghorbani Ghorbani Ghezeljehmeidan (TU Delft - Electronic Components, Technology and Materials)
W. D. van Driel (TU Delft - Electronic Components, Technology and Materials)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
The ability to accurately predict the reliability and lifetime of electronics is of great importance to the industry. The failure of the solder joint is of particular interest for these predictions, because of their susceptibility to failure under thermo-mechanical stress. However, the experimental or even conventional simulation techniques employed to estimate the lifetime of a solder joint are often too expensive or time consuming to be of practical use. Therefore, this work introduces a physics-informed Long Short-Term Memory (LSTM) to predict the plastic strain in the critical area of the solder joint. The predicted values are in agreement with the values gained from finite elements, thereby demonstrating the advantage of applying the proposed methodology.