RADAR WITH FREQUENCY RETUNING BY APERTURE INVERSE SYNTHESISING AND TWO-LEVEL NEURON-NETWORK IDENTIFICATION OF OBJECTS BY COMBINATION OF ATTRIBUTES Russian patent published in 2011 - IPC G01S13/90 

FIELD: physics.

SUBSTANCE: proposed method consists in using known devices, an altitude determination unit, a unit for isolation effective are of dissipation and a unit for isolation flying object full speed, with appropriate modification of communication between units. Neuron-network classifiers used in the radar two-level identifier are synthesized by principle of transposition of sub networks into complex artificial neuron network. Proposed approach allows improving ability of each sub network to identify particular attribute. Particular solutions in output layers of neurons are integrated into general solution by special rules corresponding to the structure of complex neuron network.

EFFECT: higher validity of flying objects identification brought about be expanded combination of attributed in neuron network classifier and of two-level system.

1 dwg