This paper is pointed at the optimization of flush atmospheric data system sensors placement, and developed the optimization methodology based on the non-dominated sorting genetic algorithm II (NSGA-II). First, the physical model of sensor system is analysed, then, designs the objective function and finds the best representation of the Pareto frontier, which makes you get the best sensor locations. Furthermore, simulations for different numbers of pressure points using these design methodology is performed based on CFD, the result showed that error of attack angle and dynamic pressure is less than 0.2%, which identifies the availability of the methodology.