Power module thermal network model parameter identification method based on least square method

Abstract

The invention discloses a power module thermal network model parameter identification method based on a least square method. According to the thermal network parameter identification method provided by the invention, the thermal conductivity of the thermal network model is calculated and the thermal capacity of the thermal network model is identified by utilizing a plurality of groups of finite element model simulation results based on a least square method principle. The heat network model parameter identification method is simple in principle, the extracted heat network model is high in precision, and the heat network model parameter identification method can be applied to online calculation of chip temperature and long-time-scale chip temperature calculation.

Description

technical field [0001] The invention relates to a method for identifying parameters of a thermal network model of a power module, in particular to a method for identifying parameters of a thermal network model of a power module based on a least square method. Background technique [0002] Industry studies have shown that power modules are the weakest components in power electronics systems. About 55% of power module failures are induced by temperature factors. The chip temperature estimation of the power module is the basis for life prediction and reliability evaluation. The thermal network model is widely used in the industry to predict and estimate the chip temperature of power modules because it is easy to embed in the digital signal processor and does not need to design additional hardware circuits. The finite element simulation model is often used to predict the chip temperature. The simulation results of the finite element simulation model have high accuracy, but the...

AI Technical Summary

Problems solved by technology

[0005] 1) The implementation process involves the initial finite element model, the finite element volume model and the new finite element model, and there are two mapping operations, and the operation of extracting the reduced-order model is complicated;

[0006] 2) The initial finite element model includes the heat dissipation system, and the new finite element model uses the convective heat transfer coefficient to model the actual heat dissipation system, which will cause large errors in the extracted reduced-order model

[0008] 1) The extracted thermal network model does not consider the thermal coupling effect between chips, and is not suitable for temperature estimation of multi-chip power modules;

Images