Machine Learning-Enhanced Equivalent Circuit Modeling for Advanced Packaging Components

Signal integrity analysis for advanced packaging is essential for guiding design and manufacturing processes and is crucial to optimizing chip performance. In this paper, we propose a general modeling method for advanced packaging components that employs MLPs to predict the parameters of equivalent circuits, thereby characterizing signal transmission over a wide frequency range. Compared to traditional empirical formulas, our approach offers higher accuracy and computational efficiency, providing valuable guidance for the design and manufacturing of advanced 3D integrated circuits.