METHOD OF RECOGNIZING NATURE OF FLIGHT OF PAIR OF AIRCRAFT DURING ITS TRACKING BASED ON KALMAN FILTERING AND NEURAL NETWORK Russian patent published in 2024 - IPC G01S13/52 G01S7/41 G06N3/02 

FIELD: data processing.

SUBSTANCE: invention relates to the field of secondary digital processing of radar signals and can be used to recognize the nature of the flight of a pair of aircraft flying in a close combat formation, according to the principle of: stationary flight of a pair (at different distances between aircraft of the pair) – manoeuvre in a pair (for example, acceleration, deceleration with different parameters) – manoeuvre by a pair (for example,"backward turn", "S-turn" with different parameters). In the disclosed method, according to the Doppler frequency input readings, generated based on the dynamic models of the various nature of the flight of the aircraft from a pair thereof, procedure of optimal multidimensional linear discrete Kalman filtration is carried out in each optimal filter OF_j of their matrix-rows (where J is the number of variants of aircraft flight dynamics in close combat order) with different a priori data received in the form of corresponding dynamic models. For each optimal filter of their matrix row, corresponding estimates of values of probabilities of recognition of the j-th version of flight of each aircraft from their pair are calculated. Calculated values of estimates during K operation cycles of all optimal filters of their matrix-row are simultaneously supplied to the j-th inputs of two identical neural networks and their training is performed. By threshold processing of the amplitude-frequency spectrum of the radar signal reflected from the pair of aircraft, two Doppler frequency readings are formed, which are supplied to combined inputs of optimal filters OF_1j of their first matrix-row and combined inputs of optimal filters OF_2j of their second matrix-row. Further, for each optimal filter OF_1j and OF_2j of the corresponding matrix-row for K cycles of their operation, estimates of the values of the recognition probabilities of the j-th version of the flight of each aircraft from their pair are calculated, which are transmitted in parallel to the corresponding j-th inputs of the corresponding neural networks and through which a decision is made on the nature of the flight of each aircraft of the pair.

EFFECT: creation of a method for recognition of the flight pattern of a pair of aircraft flying in a close combat formation, according to the principle of "stationary flight of a pair – manoeuvre of aircraft as part of a pair – manoeuvre by a pair".

1 cl, 1 dwg