Abstract
We have developed an identification method of thermal quantities for high accurate thermal simulation of a power semiconductor module. Thermal quantities are extracted by the inverse analysis from transient thermal resistance, which includes all information of a thermal path from junction to ambient and is an only quantity we can measure with a high degree of accuracy. In order to execute the inverse analysis efficiently and stably, we have adopted a genetic algorithm which is one of optimization techniques. It has been verified that the developed method works well with the test model.
