Abstract: Characterize the representative of met mast by the correlation coefficient between the wind speed of met mast and nacelle anemometer,a method based on neural network algorithm is proposed to determine the layout of met masts. Decomposing the distance vector between met mast and every wind turbine into three mutual perpendicular components of X,Y and Z,by using neural network algorithm,seven schemes have been carried out for fitting the relationship between the correlation coefficient and distance vector components. The results show that,the fitting effect of X,Y,Z three-variable input scheme is the best with an error of 0.026. The representativeness of met mast is in inverse proportion to the distance component Y,that is,the closer the distance is in the Y direction,the stronger the representativeness of the met mast is,and vice versa,weaker. The determination method of the layout scheme of met masts proposed in this paper has a guiding significance for other wind energy engineering.