Implementation Of Artificial Neural Network Forprediction Of Rain Attenuation In Microwave And millimeter Wave Frequencies

摘要

The artificial neural network model is developed on the extinction cross section data derived from the modified Pruppacher-and-Pitter (MPP) raindrop model, using finite element method (FEM) in microwave and millimeter wave frequencies ranging from 1 to 100GHz, with mean raindrop radii from 0.025 to 0.35cm for horizontal and vertical polarizations. The mean square error and correlation coefficient R~2 values between derived data using FEM on MPP raindrop and artificial neural network (ANN) are found to be 1.62 × 10~-4 and 0.9994 for vertical polarization, and 4.677 × 1O~-3 and 0.9943 for horizontal polarization. The artificial neural network model gives results with good accuracy for calculating extinction cross section of raindrop. It is then applied on raindrop size distributions of Singapore and the Indian regions, for the prediction of specific rain attenuation. The results of specific rain attenuation obtained using ANN are compared with reported experimental data and they are found to be in close agreement.