METHOD FOR MULTI-FEATURE RECOGNITION IN MULTIFUNCTIONAL RADAR STATION OF AIRCRAFT CLASS ON PRINCIPLE OF "AIRPLANE WITH TURBOJET ENGINE - AIRPLANE WITH TURBOPROP ENGINE - HELICOPTER - ROCKET - UNMANNED AERIAL VEHICLE" BASED ON COMBINED APPLICATION OF KALMAN FILTRATION AND NEURAL NETWORK Russian patent published in 2024 - IPC G01S7/41 G01S13/52 G06N3/02 G06T7/20 

FIELD: radar.

SUBSTANCE: invention relates to the field of radar and can be used for identification in a multifunctional radar station of the class of an aircraft on the principle of "airplane with a turbojet engine (TJE) – airplane with a turboprop – helicopter – rocket – unmanned aerial vehicle (UAV)" based on combined application of Kalman filtering and neural network (NN). Method consists in the fact that reflected from aircraft radar signal is received by multifunctional radar station and is subjected to narrow-band Doppler filtering at an intermediate frequency based on a fast Fourier transform (FFT) procedure, converted into an amplitude-frequency spectrum (AFS), the spectral components of which are caused by reflections of the radar signal from the aircraft airframe. Only those Doppler frequency readings Fq are generated by threshold processing of the AFS signal, where q=1,…, Q; Q is the number of aircraft classes at frequency positions of which the amplitudes of the spectral components have exceeded the preset threshold. In the NN training mode, in accordance with q-th dynamic models of Doppler frequencies, Doppler frequency readings are successively formed, based on which values of the corresponding q-th autocorrelation function (ACF) of Doppler frequency fluctuations are calculated in R cycles. Corresponding values of ACF coefficients for aircraft of q-th class are calculated, which together with additional recognition features: aircraft flight altitude and speed of the q-th class, the value of Doppler frequencies caused by the radar signal reflection from the rotating blades of the low-pressure compressor of the aircraft of the classes "airplane with turbojet engine", "airplane with a turboprop", range of Doppler frequencies caused by the radar signal reflections from the engine and blades of the aircraft of the "helicopter" class, from the propellers of the aircraft of the "airplane with a turboprop" class, are supplied to the corresponding input neurons of the NN for its training. In aircraft class recognition mode, Doppler frequency readings are filtered using each q-th Kalman filter. As a result, Doppler frequency fluctuation estimate is formed at the output of each q-th Kalman filter, according to which the ACF estimate and the ACF coefficient estimates are calculated, which with additional features for the q-th class aircraft are transmitted to the NN inputs to make a decision that the aircraft has the q-th class.

EFFECT: creation of a method which enables to recognize an aircraft class in a multifunctional radar station with a probability not lower than a given one based on the principle of "airplane with turbojet engine – airplane with turboprop – helicopter – rocket – UAV".

1 cl, 1 dwg, 1 tbl