The different factors such as material, height, thickness, installation status, and geometry of the wall's head can influence on the efficiency of the sound walls to decrease the noise pollution of theairports. This paper is presented to improve the effective geometry of the wall's head as well as finding the best wall's head to maximize the noise reduction in an airport. To investigate the performance of the sound wall, the boundary element method is used. Then, in order to modeling the sound walls with different dimensions and sizes, PATRAN software is utilized. In the next step, the models are meshed and finite element method is used to analyze the vibrations of the models. Consequently, the natural frequencies and the mode shapes of sound's walls are predicted and finally the insertion loss via modeling of sources of noise and sound receivers are calculated. The design method of Taguchi experiments is applied to decrease the total numbers of the different models of Y shape geometries. Lastly,the governing equations with approximately fitted over the test cases are determined by neural network. Finally, the genetic algorithm is used to obtain the ideal head's wall.