Answers are grounded in this paper and cite their sources.
The safe operation and lifetime of the power module are heavily dependent on the temperature distribution, making it imperative to optimize the thermal performance of the layout at the design stage. During the module design phase, a significant number of layouts are assessed. It is important to constrain computational complexity while achieving high accuracy. mechanistic modelling and data-driven modelling have limitations in terms of complexity and generalizability respectively. Therefore, this paper proposes a hybrid data-driven and mechanistic modelling approach. This approach incorporates the theoretical analysis of mechanistic modelling and the high efficiency of data-driven modelling to accurately and efficiently compute the temperature distribution of a given layout. This approach initially acquires the dataset from simple structures and conducts a dimensionality reduction to obtain the isothermal point set (IPS). Later, IPS of complex structures is developed based on mechanistic modelling. Finally, the IPS is converted to a field to obtain a temperature distribution of the given layout. The accuracy of the approach is verified by a steady-state thermal experiment. With the proposed approach, an accurate assessment of surface thermal distribution of different layouts can be quickly finished. The thermal characteristics evaluation efficiency can be improved roughly 300 times, which provides the necessary technical basis for layout optimization.
No takes yet. Share an insight, caveat, or question.
Zhang et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: