Identification of Convection Constants for Electronic Packages Using Modified Genetic Algorithm and Reduced - Basis Method

-A new inverse analysis method is presented to identify parameters of heat convection in microelectronic packages. This approach adopts a modified Micro Genetic Algorithm (μGA) in finding the global optimum of parameters. A reduced-basis approach is introduced in the forward heat transfer analysis so as to significantly improve the efficiency in the calculation. Different identification procedures are employed to identify heat convection coefficients of a typical microelectronic package. Comparisons between different algorithms are performed. Results show that the use of the reduced-basis method together with the modified μGA outperforms the conventional GAs significantly. The presented method of coefficient identification is ideal for practical applications. It is efficient enough even for online analysis of both forward and inverse problem.