Heating elements layout design on printed circuit board by Bayesian optimization

Abstract

This paper describes effectiveness of Bayesian optimization (BO) for thermal design of electronic devices. Recently, as the heating power density of electronic devices has increases due to miniaturization and performance enhancement, more effective thermal design is required. However, it is difficult to optimize the thermal design due to need to consider various trade-offs associated with packaging and non-steady heating power of components. In this study, we applied a method which combined BO with lumped-capacitance thermal circuit network model to chip layout optimization of a printed circuit board.