The Research Of Wavelet Network Based On Sampling Theory In Antenna Radiation

Antenna radiation is necessary and foundation for testing and fault diagnosis of antenna performance,which plays an important role in antenna design,maintenance and performance guarantee.However,in the complex environment,because of the interference of noise,the traditional modeling methods such as networks are easily falling into local optimum and become over fitting,which greatly affects the efficiency of antenna maintenance and fault diagnosis in the complex outfield environment.In order to solve the problems above,the fast online modeling of antenna radiation in complex interference environment has become the focus in this paper.And the wavelet network based on sampling theory is applied to the antenna radiation modeling,which makes the approximation ability and anti-interference of the wavelet network improved obviously.In summary,the following works have been done:(1)Starting from the direction of frequency band,the band characteristics of wavelet network and antenna radiation function are analyzed respectively.which indicate that the two functions have band pass characteristics.Then the feasibility of using the wavelet network to model the antenna radiation is demonstrated.(2)Introducing the sampling theory into the wavelet network.In complex environment,the training data not only contain a lot of noise,but also distribute sparsely and in homogeneously,which makes the approximation ability of wavelet network greatly reduced.To solve this problem,the sampling theory is introduced into wavelet network.Combined with the limited frequency band characteristics of wavelet network,a wavelet network algorithm based on sampling theory is proposed,which makes the approximation ability of wavelet neural network improved significantly.(3)Reset the parameters of the wavelet network.The input weights,output weights and hidden layer nodes of wavelet network are reset separately,based on the new wavelet network algorithm.The approximation ability and regularity of the wavelet network has been proved good,while its filtering capability is just like a low pass filter,has also been proved,at the same time.(4)In order to verify the results in this paper,wavelet network and modeling and Simulation of RBF network based on regularization technology are done in weak noise and strong noise environment respectively.The experimental simulation results show that the modeling method proposed in this paper can effectively realize the fast online modeling of the radiation field of array antenna.Compared with the RBF network based on regularization technology,the method proposed in this paper is more effective in filtering noise interference,avoiding overfitting of wavelet network and robust to noise changes.