Research On Hydrometeors Classification Technology Of Dual-Pol Weather Radar Based On FNN

The reasonable classification of hydrometeors in the cloud has important application value to improve the accurate measurement of quantitative precipitation,and can provide important reference for the operation decision and evaluation of artificial weather.In this paper,using Dual-Pol weather radar to study the classification of hydrometeors.The main research contents are as follows:Firstly,the characteristics of meteorological echo and non-meteorological echo are studied,and the polarization characteristics of each hydrometeors in meteorological echo are analyzed emphatically.In the study of the microphysical characteristics of meteorological echo and non-meteorological echo,the size,shape and orientation of hydrometeors in meteorological echo are mainly studied.The results show that the echoes and non-meteorological echoes wave intensity and radial velocity were studied and analyzed.The polarization characteristics of the hydrometeors in the meteorological echo are explained by the polarization parameter study of the Dual-Pol weather radar.Secondly,in order to establish the membership function of each polarization parameter in the study of hydrometeors classification of Dual-pol weather radar,empirical value is often used to classify the hydrometeors accurately.A FNN(Fuzzy Neural Network)based on the T-S(Takagi-Sugeno)model is proposed to supervise the classification of hydrometeors.In this paper,a fuzzy neural network with adaptive membership function is established by combining fuzzy logic and neural network learning and training.Then,the FNN error feedback learning feature is used to calculate the membership parameters of the different parameters of the different hydrometeors types in the fuzzy process.The polarization parameters of the polarization parameters are calculated,and re-establish a new membership function to ensure the accuracy of hydrometeors classification.The effectiveness of the method is proved by the results of the measured data of S-band,C-band and X-band Dual-pol weather radar.Thirdly,a method of classification of hydrometeors based on FNN-CM(Fuzzy Neural Network-C Mean)is proposed to solve the classification of hydrometeors in the presence of ground clutter.In this method,the membership function parameters of the polarized parameters of the clutter are calculated by FNN training of the clutter data in the clear sky mode,and the ground clutter in the rainfall mode can be used to eliminate the clutter on the classification accuracy of hydrometeors.Then,the classification of hydrometeors after clutter suppression is studied.The clustering center of each hydrometeors type and the membership degree of each hydrometeors belonging to each type of precipitation are calculated,and then the cost function of the membership degree of hydrometeors is constructed.When the cost function of the hydrometeors satisfies the condition,the calculated fuzzy membership matrix is retreated to obtain the type of each hydrometeors.The validity of the method is proved by the results of the Dual-Pol weather radar data of C-band and S-band.