Abstract: The turbulence warning of civil airports requires refined wind field data. In this paper, a wind speed prediction model of particle swarm optimization(PSO) based on Elman neural network(PSO-Elman) is established to achieve the purpose of wind field refinement using lidar detection related data. Firstly, the spectrum width, signal to noise ratio and echo distance measured by the lidar experimental platform of Lanzhou Zhongchuan International Airport are used as the input data of PSO-Elman network. Then, the PSO algorithm is used to optimize the internal parameters of Elman network, and fitness function was established to improve the prediction accuracy and to reduce the convergence time. Finally, the non-linear function mapping of the convergent network is established to predict the wind speed between the radial distance gates. The simulation results show that the relative error between the measured and extended wind speed and the predicted wind speed is 6%, and the linear regression coefficiency between the measured and predicted wind speed is 0.919, which proves the effectiveness and feasibility of the wind speed prediction model.

Key words: Elman neural networks, particle swarm optimization(PSO) algorithm, prediction of wind speed, lidar