A multiobjective genetic algorithm is used to optimize the structure of circular concave cavities and microchannels. The objective functions of thermal resistance and pumping power are constructed by the response plane approximation method according to the simulation results, and then a mathematical model of multiobjective genetic optimization with the structural parameters of microchannels as variables is established. The Pareto optimized solution sets of thermal resistance and pumping power are calculated by the nondominated ranking genetic algorithm NSGA-II, and the comprehensive heat transfer performance is evaluated by the enhanced heat transfer factor. The results show that the multivariate statistical coefficients
R
2
of the thermal resistance and pumping power objective functions are 0.932 9 and 0.996 6, respectively, indicating the high accuracy of the fitted functions. The optimized channel structure (
e
1
=
0.036.8
mm
,
e
2
=
0.019.3
mm
) was used to achieve a more uniform temperature field distribution and better integrated heat transfer performance (enhanced heat transfer factor
η
=
1.23
). When the thermal resistance is larger or the pump work is larger, the comprehensive heat transfer effect is not as good as the working condition when the thermal resistance and pump work are more uniform.