Analysis Of Non-linear Scattering Function And Modeling Research Based On Recurrent Neural Network

Since 1850s, the influence in microwave correspondence of non-linear behavior has received extensive attention. In 1980s, the development of digital mobile phone makes the complicated wireless system very sensitive to non-linear distortion. Especially some microwave apparatus put up strong non-linear behavior in big-signal operating, such as microwave power amplifier. So the conventional linear circuit S parameters theory can't be used to analyze those non-linear apparatus.A new way called big-signal non-linear scattering function to token non-linear behavior has been described in this paper. The linearization, the definitions and the characteristics of non-linear scattering function have been studied in this paper. The conversions among the non-linear scattering function matrix, non-linear impedance function matrix and non-linear conductance function matrix are also given. We design a measurement system to measure the non-linear scattering of a non-linear system. The data from the measurement system has been used to train the recursion neural network to modeling the big-signal non-linear scattering function. The error is in precision request.