A METHOD FOR RECOGNIZING THE TYPICAL COMPOSITION OF A GROUP AIR TARGET FROM THE CLASS "TURBOJET ENGINE AIRCRAFTS" BASED ON KALMAN FILTERING AND A NEURAL NETWORK Russian patent published in 2022 - IPC G01S13/52 G01S7/41 G06N3/02 G06T7/277 

FIELD: aviation radars.

SUBSTANCE: invention relates to the field of secondary digital processing of radar signals and can be used to recognize the typical composition of a group air target from the class "turbojet aircrafts". The claimed method consists in the fact that the radar signal reflected from the group air target from the "turbojet aircraft" class is subjected to narrow-band Doppler filtering at an intermediate frequency based on the fast Fourier transform procedure and is converted into an amplitude-frequency spectrum, the components of which are due to the reflections of the signal from the airframes of the group aircraft and the rotating blades of the first stages of the impellers of low-pressure compressors the pressure of their power plants. Next, the Doppler frequency count of the centroid F_c is calculated as the average value of the Doppler frequency counts of the local maxima corresponding to the reflections of radar signals from the airframes of the group aircraft, the Doppler frequency counts F_kj of the first compressor components of the signal spectrum are determined, where i=1, …,I; I is the number of aircraft in the group; j =1, …,J; J is the number of types of turbojet aircraft in the group. The offset F_ij of the Doppler frequency samples are calculated between the centroid F of the group air target and each count F_kij in accordance with the dynamic model of the offsets ΔF_ij of the Doppler frequencies, according to which, during K cycles, the neural network is pre-trained for various types of targets with different flight patterns having the corresponding Doppler frequency spacing ΔF_ij (k+1). In the process of recognizing the typical composition of the group air target by the radar signal reflected from its elements, with the help of each ij-th Kalman filter functioning with the corresponding dynamic model with different flight patterns of the group's aircraft, Doppler frequency spacing counts are filtered, as a result of which estimates of Doppler frequency spacing are formed at the output of each ij-th Kalman filter, which are received at the inputs of the neural network to make a preliminary decision for K cycles of the Kalman filters on the presence in the group of the i-th aircraft of the j-type with the corresponding probability P_ij, which is compared with the threshold value P_pore, when the condition is met for each value of the probability P_ij ≥ P_pore, the final decision is made that the i-th aircraft of the j-th type is in the group, otherwise a decision is made on the absence of this type of aircraft in the group.

EFFECT: insurance of the constancy of the probability of recognition of the typical composition of the group air target not lower than the specified one by optimizing the decisive rule that allows adapting the recognition process to the different nature of the flight of the group air target. The present invention enables to recognize the typical composition of a group air target from the class "turbojet aircrafts".

1 cl, 1 dwg, 1 tbl