Analysis Of Microwave Nonlinear Scattering Function Theory And Study On Large-Signal Modeling With BP Neural Network

With the applications of microwave technology, the design of the microwave circuit becomes the toughest problem to solve.Modern communication developments tend to apply larger power level and more sophisticated modulation program.It makes the components and ciucuits of the system represent nonlinear characteristic that leads to worse performance.So constructing models, which can accurately describe the hard-nonlinear behavior of transistors, becomes popular in research field.In the last decades, classic large-signal analysis is based on S-parameter since only S-parameter can be measured.S-parameter theory is linear theory.Under large-signal conditions, the components usually represents nonlinear characteristic.So S-parameter theory no longer fits for large-signal analysis.This paper described a new large-signal analysis method, namely nonlinear scattering function.The method includes the influence of harmonic waves, so it can used for the analysis of harmonic waves produced by nonlinear characteristic.In this article, a black-box model based on frequency domain measurement is introduced.The modeling method is built on the idea of large-signal network analysis that includes measuring and modeling solutions.The core of the modeling is the fitting technology of the nonlinear scattering function.Whether the scattering function can accurately characterize hard nonlinear behavior of microwave components is key to choose the fitting technology.BP Neural Network(BPNN) as a nonlinear system in nature can approximate any continuous function with any accuracy in theory.In recent years, many people have proved the approximation ability of BPNN.So it is used to build the model.This model is more accurate than ever because it is based on real condition.Finally, we validate the model that have been built above.The results indicate BPNN can used for modeling scattering function, and the model is accurate.