Prediction Of Electromagnetic Scattering Characteristics In Typical Areas Of Real Sea And Landforms Based On Deep Learning

The research on the characteristics of the sea and landforms environment is of great value for many fields such as remote sensing and radar detection.Only through more explicit study and analysis of environmental background characteristics can target information be accurately captured.Given the high cost of data acquisition solely through experiments,it is urgent to combine real environmental data to accurately model for different types of background,so as to achieve the coordination of theoretical model and measurement data.With the rapid development of deep learning technology,it provides a new and effective research idea for traditional electromagnetism.In this paper,the adaptive modeling of typical sea and landforms environment is carried out based on real environment data,and the detailed research on its scattering characteristics was given.What's more,the deep learning technology is used to mine the implicit relationship between electromagnetic scattering characteristics and environmental data.The main research results and features of this paper are as follows:1.Based on the ECWMF data,the spatial and temporal distribution database of marine environmental elements in the South China Sea is established,which provides a reliable multi-physical data source for the establishment of real sea surface model,sea clutter modeling and the application of deep learning.Combined with the ECWMF data and DEM data provided by the Chinese geospatial cloud,the spatial and temporal distribution database of the typical surface environmental elements in the Tibet border region and East China region is established,which provides the real multi-physical field data source of the surface environment for the establishment of the real surface electromagnetic scattering model.2.In consideration of the actual sea conditions,the mixed wave spectrum that covers the detailed wave information was given to solve the defect that the traditional wind-wave spectrum can not make difference analysis of different sea conditions under the same wind speed.Then the real sea model in the South China Sea was establishes based on the mixwave spectrum.The time series of sea clutter in the South China Sea is simulated by using the multi-scale high performance parallel computation method.The differences of sea clutter characteristics in the South China Sea are analyzed mainly from two aspects of marine environmental elements and radar parameters.Besides,the main environmental elements that affect the sea clutter characteristics in the South China Sea are analyzed in two dimensions of time and space.3.Combined with the deep learning technology,the deep neural network is used to establish the mapping relationship between the characteristics of sea clutter in the South China Sea and the real marine environmental elements.The prediction and inversion of the characteristics of sea clutter and environmental elements are realized.What's more,the training effect is considerable,which provides a feasible new idea for the study of marine electromagnetic characteristics.4.Based on the real earth surface environment data,suitable electromagnetic scattering calculation model was established for different landforms in China in view of the geographical features of these typical landforms.Tibet border,the Qinghai-Tibet plateau and the East China were regarded as the objects.The distribution characteristics of backscatter coefficient in these representative regions were studied from several aspects as environment parameters,radar wave band,polarization and incident Angle.The research content of this paper provides a certain theoretical support for the terrestrial and marine environment sensing and assessment system and provides a reference direction for the study of environmental characteristics.