Neural Network Recognition of Spectral Portraits of Airborne Objects Observed Using the Method of Shadow Inverse Radar Aperture Synthesis

摘要

Recognition of spectral portraits of airborne objects in the course of the shadow inverse radar aperture synthesis is analyzed for targets observed by a bistatic radar system using the forward-scattering radar technique. An artificial multilayer neural network is used for recognition of object classes and types. The network is trained using a set of spectral portraits theoretically modeled from shadow silhouettes of airborne objects in certain intervals of their speeds and flight altitudes. The efficiency of neural network recognition is estimated using sets of shadow spectral portraits obtained from real forward-scattering signals.