MOSFET modelling as a digital twin and use of machine learning models to find current leakage

Abstract

The object of this thesis is the presentation of a digital twin of the MOSFET, with the aim of providing tools for a better and more in-depth study and monitoring of its functions. Through the digital twin of the MOSFET, the peculiarities and challenges it presents, such as threshold leakage current, a common occurrence in non-ideal MOSFETs, are investigated. While digital twins have found applications in many industrial fields due to their versatility, in the case of this thesis, Machine Learning is utilized to enhance the accuracy and sensitivity of the analysis in the MOSFET. The corresponding digital twin, apart from its description, has been implemented within the thesis through the use of Python programming language, incorporating advanced Machine Learning models for even deeper understanding. The aim is not just to present a technological innovation, but to highlight how the integration of Machine Learning can enhance the performance and applicability of digital twins in real industrial situations.