RU2844360C1 - Method for recognition of type of aerial object with several turbojet engines at asynchronous operation by secondary radiation - Google Patents

Abstract

FIELD: radar ranging.

SUBSTANCE: invention relates to radar ranging of point aerial objects and can be used in radioelectronic systems for radar recognition of types of observed aerial objects. Claimed method is based on the pulse-Doppler radar receiving reflected radar coherent signals from the observed aerial objects, primary processing of received signals in form of narrow-band spectral analysis for detection of frequency-resolved harmonic components of rotating structural elements of observed aerial object, forming a spectral-Doppler portrait of the observed object, calculating an estimate of the rotor speed of the low-pressure compressor, forming a base of reference portraits. At that, reference portraits feature space consists of number of blades at first two stages of low-pressure compressor or turbine and permissible low-pressure compressor rotor speed. Formed spectral-Doppler portrait is compared with portraits of aerial objects from a database of references. Minimum value of calculated weighted square of difference between values of frequencies is compared with threshold determined by criterion χ². If the value is less than the threshold, a decision is made to recognize the type of aerial object from the reference database, otherwise a decision is made to observe an unknown type of aerial object. After calculation of the reference spectral-Doppler portrait, an additional procedure of estimating the speed of the engines in the Z-dimensional search window is carried out, the dimension of which is determined by the number of engines on the aerial object, and calculating the likelihood functional, determined by the difference in frequencies of the observed spectral components and the spectral components of the reference portrait with the Z-number of engines with a set range of variation of the rotor speed of their low-pressure compressors, corresponding to the given type of engines, to take into account their asynchronous operation.

EFFECT: high reliability of recognizing the type of an aerial object.

1 cl, 1 dwg

Description

The invention relates to the technology of radar detection of point aerial objects and can be used in radio-electronic systems for radar recognition of types of observed aerial objects.

The "Method for Recognizing the Type of an Aircraft with a Turbojet Engine in a Pulse-Doppler Radar Station" (RU No. 2731878 C1, published 09/08/2020 IPC G01S 13/52) is known, which consists in the fact that the radar signal reflected from an aircraft with a turbojet engine is subjected to narrow-band Doppler filtering based on the fast Fourier transform (FFT) procedure and converted into an amplitude-frequency spectrum (AFS) of signal reflections from the airframe of an aircraft with a turbojet engine and the rotating blades of the low-pressure compressor (LPC) impeller of its power plant. By threshold processing of the AFS signal, only those Doppler frequency readings F _i with the corresponding amplitudes of the spectral components that exceed the established threshold are formed. Simultaneously, during the time T of each space survey, two values of the range D ₁ and D ₂ to the aircraft with a turbojet engine are measured and the frequency position of the Doppler frequency F _n is calculated, depending on the speed of approach of the carrier of the pulse-Doppler radar to the airframe of the aircraft with a turbojet engine. In the signal AFS, the position of the Doppler frequency with the maximum amplitude of the spectral component that has exceeded the established threshold is determined, which corresponds to the value of the Doppler frequency F _k , caused by the speed of approach of the carrier of the pulse-Doppler radar to the rotating blades of the first stage of the LPC of the power plant of the aircraft with a turbojet engine, and the difference in Doppler frequencies ΔF _nk = (F _n -F _k ) is calculated. Additionally, during the time T of each space survey, the values of the onboard bearings f _r in azimuth and f v _in elevation, the average range are measured, and the flight altitude of the aircraft with a turbojet engine is calculated. For each altitude H of a turbojet aircraft, the range of differences ΔF _nk is divided into Q non-overlapping sub-ranges. When the Doppler frequency difference ΔF _nk falls into the q-th sub-range, a decision is made about the q-th type of turbojet aircraft flying at altitude H. The disadvantages of this method are:

1. when using only the first stage of the low-pressure compressor, a truncated spectral-Doppler portrait of the observed aerial objects is received and a truncated reference portrait of these objects is formed, which sharply reduces the feature space for making a decision on the type of the observed object and reduces the efficiency of recognizing the type of object;

2. there is no procedure for calculating the estimated rotation speed of the low-pressure compressor rotor;

3. from step to step of space survey in the method under consideration for making a decision on the type of air object, a procedure for calculating the difference in Doppler frequencies and comparing it with a certain subrange at typical altitudes is proposed. At the same time, at the stage of observation of an air object, the change in the frequencies of the spectral components requires a procedure for identifying the observed components, which is absent from the description of the method.

The closest in technical essence and achieved result to the claimed method for recognizing the type of an airborne object with several turbojet engines in asynchronous operation by secondary radiation (the prototype for the proposed invention) is the method for recognizing the type of an airborne object by the turbine effect (RU No. 2790143 C1, “Method for recognizing the type of an airborne object by the turbine effect”, published 14.02.2023, IPC G01S 13/52).

The method for recognizing the type of an aerial object based on the turbine effect described in the prototype includes the following main stages: receiving by a pulse-Doppler radar station reflected radar coherent signals from observed aerial objects, primary processing of the received signals in the form of narrow-band spectral analysis to detect frequency-resolvable harmonic components of rotating structural elements of the observed aerial object, forming a spectral-Doppler portrait of the observed object, calculating an estimate of the rotation frequency of the low-pressure compressor rotor, forming a database of reference portraits, the feature space of which consists of the number of blades in the first two stages of the low-pressure compressor or turbine and the permissible rotation frequencies of the low-pressure compressor rotor, comparing the formed spectral-Doppler portrait with the portraits of aerial objects from the database of references, calculating the diagonal matrix of errors in the estimates of the frequencies of the spectral components between the reference and the observed spectral-Doppler portrait, calculating according to the criterion of the maximum likelihood functional an estimate of the parameter characterizing the type of aerial object. After the formation of the spectral-Doppler portrait of the observed air object based on the known glider and reference - one of the spectral components of the formed spectral-Doppler portrait, the estimate of the low-pressure compressor rotor speed is calculated. After comparing the observed spectral-Doppler portrait with the reference portrait database, a complete matrix of errors in the frequency estimates of the compared spectral components is formed; when calculating the likelihood functional, a weighted square of the difference in the frequency values of the spectral components of the formed and reference spectral-Doppler portraits is calculated. Additionally, a threshold is determined according to the agreement criterion, or χ ² , with which the minimum value of the calculated weighted square of the difference in frequency values is compared; if the value is less than the threshold, a decision is made to recognize the type of air object whose spectral-Doppler portrait is contained in the reference database; otherwise, a decision is made to observe an unknown type of air object that is absent from the reference database.

The main disadvantages of the prototype method include low reliability of recognizing the type of the observed aerial object in the presence of several engines with their asynchronous operation (different speeds of rotation of the compressor or low-pressure turbine), which entails the appearance in the observed spectral-Doppler portrait of duplicated and separately observed spectral components due to the asynchronous operation of the engines, as well as the possible occurrence of errors in determining the engine speed and incorrect formation of the reference spectral-Doppler portrait and, as a consequence, errors in determining the type of aerial object.

The technical result of the invention of the method for recognizing the type of an airborne object with several turbojet engines during asynchronous operation based on secondary radiation is an increase in the reliability of recognizing the type of an airborne object.

The method for recognizing the type of an aerial object with several turbojet engines during asynchronous operation based on secondary radiation is based on the reception by a pulse-Doppler radar station of reflected radar coherent signals from observed aerial objects, primary processing of the received signals in the form of narrow-band spectral analysis to detect frequency-resolvable harmonic components of rotating structural elements of the observed aerial object, formation of a spectral-Doppler portrait of the observed object, calculation of an estimate of the rotation frequency of the low-pressure compressor rotor based on a known reference spectral component of the formed spectral-Doppler portrait, formation of a database of reference portraits, the feature space of which consists of the number of blades in the first two stages of the low-pressure compressor or turbine and permissible rotation frequencies of the low-pressure compressor rotor, comparison of the formed spectral-Doppler portrait with portraits of aerial objects from the database of standards, calculation of the diagonal matrix of errors in the estimates of the difference in frequencies of the spectral components of the reference and observed portraits, calculation according to the criterion of the maximum likelihood functional estimating the parameter characterizing the type of aerial object based on a truncated spectral-Doppler portrait, excluding the reference component, based on the weighted square of the difference in the frequency values of the spectral components of the received and formed reference spectral-Doppler portraits, calculating the full matrix of errors in the difference in the frequency values of the spectral components being compared, determining the threshold based on the ^χ2 criterion, with which the minimum value of the calculated weighted square of the difference in frequency values is compared; if the value is less than the threshold, a decision is made to recognize the type of aerial object whose spectral-Doppler portrait is contained in the reference database; otherwise, a decision is made to observe an unknown type of aerial object that is not contained in the reference database.

New features of the claimed method are that the procedure of forming a Z-dimensional search window is additionally implemented, carried out after calculating the reference spectral-Doppler portrait, and the window dimension is determined by the number of engines on the air object and the number of spectral components formed for each observed spectral component of the portrait with the established range of change in the rotation frequency of the low-pressure compressor rotor. The likelihood functional is calculated, determined by the number of observed spectral components, in which the diagonal matrix of errors of the difference in frequency values, the observed and reference spectral components being compared is used, and the asynchronous operation of the engines is also taken into account.

The figure shows a graph of the probability of correct recognition of the type of aerial object.

The essence of the claimed method consists in processing a reflected coherent radar signal from an aerial object in a pulse-Doppler radar station, which makes it possible to form a spectral-Doppler portrait of the aerial object, obtained from the amplitude spectrum of the observed signal by detecting frequency-resolvable harmonic components in a mixture of the reflected signal and noise.

Recognition of the type of aerial object requires the presence of a feature space (a priori information):

- the alphabet of all recognizable types of air objects in the form of the number of blades in the first two stages of a low-pressure compressor or turbine N1 _k , N2 _k ; k=1 … K _c , where K _c is the total number of known types of these objects;

- maximum speed of an air object V _ц ^max ;

- maximum and minimum permissible rotation speeds of the low-pressure compressor rotor, respectively F _врk ^max , F _врk ^min ;

Recognition of the type of aerial object is possible upon detection of at least three spectral components (I≥3) of the observed spectral-Doppler portrait, consisting of a glider and at least two non-multiple and non-symmetrical components relative to the glider.

The position of the spectral components of the portrait on the frequency axis is determined by the radial velocity of approach to the glider (V _r ), the number of blades on the corresponding stage of the low-pressure compressor (turbine) of the engine and the rotation frequency of its rotor:

where λ=c/f ₀ is the wavelength of the irradiating radar, n _i are the errors in forming the spectral-Doppler portrait with zero mean and dispersion σ ² ; p1, p2, p3 are the multiplicity of harmonics of the corresponding stages of the low-pressure compressor. The harmonic with the maximum amplitude, or the harmonic assigned by target designation in the tracking mode, is taken as the glider component in the spectral-Doppler portrait.

The method for recognizing the type of airborne object is based on calculating the likelihood functional for each of the hypotheses about the type of object (k) and the rotation speed of the low-pressure compressor rotor (F _vr ):

where D is the matrix of error variances caused by observation noise and errors in the formation of the reference spectral-Doppler portrait. The minimum of the exponent can be used as a criterion for making a decision on the type of air object. The value of the elements of the vector of averages M _k , determining the position on the frequency axis of the spectral components of the reference spectral-Doppler portrait, is calculated using the formula:

since the rotor speed F_vris unknown, then in the process of observation it is necessary to solve the problem of eliminating the parametric uncertainty of its value, i.e. to make an assessmentFor this purpose, in the spectral-Doppler portrait it is proposed to select a reference harmonic with a frequency ƒ_Topand for each type of air object, form several hypotheses about the rotation frequency of the low-pressure compressor rotor. Formation of several hypotheses is necessary to eliminate uncertainty about the belonging ƒ_Topto a harmonic of unknown multiplicity and an unknown stage number of the low-pressure compressor (turbine). In practice, when calculatingwe can limit ourselves to the harmonic multiplicity (p1g, p2g=[-2…2]; p1g and p2g are not equal to 0 simultaneously) and two stages of the compressor or turbine (K=1.2), since the amplitude of other harmonics is significantly lower:

As a result, a set of hypotheses about the rotation speed of the low-pressure compressor is formed, and the rotation speed hypotheses that are outside the permissible rotation speed range are discarded. For each of the remaining rotation speed hypotheses, standard spectral-Doppler portraits are formed Due to the rapid decrease in the amplitudes of harmonics with an increase in their multiplicity, to form a reference portrait it is sufficient to take into account the first two stages of the low-pressure compressor with a multiplicity of p1=[-3…3] and p2=[-2…2]. Consequently, the reference spectral-Doppler portrait for each hypothesis has a set of J=(2⋅p1 _max +1)(2⋅p2 _max +1)=35 Doppler frequencies (where j=1…J; p1 _max and p2 _max are the highest harmonic multiplicity corresponding to reflections from stages 1 and 2 of the low-pressure compressor) under the condition of synchronous operation of the engines:

If the observed aerial object has two Z=2 engines with non-synchronous operation (different rotation speeds of the compressor or low-pressure turbine), duplicated and separately observed spectral components appear in the observed spectral-Doppler portrait.

For the reference spectral-Doppler portrait, arrays of frequency components corresponding to each engine are calculated:

which are combined into the resulting standard:

In the reference spectral-Doppler portrait, for each j-th spectral component, a procedure for forming a search window is implemented, the size of which is determined by the number of hypotheses for the rotation frequencies of the low-pressure compressor rotor of each engine formed with a step

Where - a rough estimate of the engine speed, calculated according to (4), - hypothesis about the rotation frequency of the 1st and 2nd engines of the reference portrait, SP _1,2 - the number of spectral components in the search window, sp _1,2 - the index (number) of the spectral component in the search window.

Then the likelihood functional (2) takes the form:

The error covariance matrix D included in the likelihood ratio (9) is diagonal, since the errors in measuring the low-pressure compressor (turbine) rotation frequency are not common to all of its elements. The diagonal elements of the matrix are determined by the variance of the error in measuring the airframe frequency and the off-diagonal ones are the variances of the errors in measuring the frequency of the glider and the frequency of the analyzed harmonic

Since the “residual” is a normal random variable with zero mean, the quadratic sum of the normalized “residuals” in the numerator of (9) is described by a ^χ2 distribution with I degrees of freedom.

Then, by setting the acceptable probability of recognizing the type of aerial object, it is possible to determine the threshold, upon exceeding which the square sum of the normalized “residuals” makes the decision “I don’t know” (observation of an aerial object of an unknown type).

If the quadratic sums of the normalized “residuals” for several types of airborne objects are below the threshold, then using the corresponding likelihood functionals (9), the probabilities are calculated that the observed spectral-Doppler portrait belongs to these types of objects.

The efficiency of the proposed method for recognizing the type of airborne object was estimated based on the simulation results. To simulate the spectral-Doppler portrait, a hypothetical type of airborne object was used, which had N1=29 blades on the first stage of the low-pressure compressor and N2=35 blades on the second stage. The rotation frequency of the low-pressure compressor rotor was 125 Hz. The asynchrony of the engine rotation frequency was 5%.

The analysis of the influence of the proposed procedure for forming the search window by the frequencies of the spectral components when comparing the spectral-Doppler portraits (drawing) shows that for the signal-to-noise ratio equal to q _s/n = 25 dB, the probability of correct recognition of the airborne object type without taking into account the proposed procedure for forming the search window is equal to P _pr = 0.8 (curve 2 - known prototype method), and the probability of correct recognition of the airborne object type taking into account the proposed procedure for forming the search window (curve 1 - proposed method) increases to 0.95. With the probability of correct recognition of the airborne object type P _pr = 0.95, the required signal-to-noise ratio is equal to q _s/n = 29 dB without taking into account the proposed procedure for forming the search window (curve 2 - known prototype method), and taking into account the proposed procedure for forming the search window, the signal-to-noise ratio decreases by 4 dB (curve 1 - proposed method).

Thus, the proposed method for recognizing the type of an aerial object with two engines, based on taking into account the modulation of the rotating elements of the power plant design, ensures an increase in the reliability of recognizing the types of aerial objects by increasing the probability of their correct recognition.

The proposed technical solution has an inventive level, since it does not clearly follow from published scientific data and known technical solutions that the claimed method for recognizing the type of an aerial object with two engines by the turbine effect ensures an increase in the reliability of recognizing the types of aerial objects.

The proposed technical solution is industrially applicable, since its implementation can be implemented using elements that are widely used in the field of electronic and radio engineering.

Claims (1)

A method for recognizing the type of an air object with several turbojet engines during asynchronous operation based on secondary radiation, based on the reception by a pulse-Doppler radar station of reflected radar coherent signals from observed air objects, primary processing of the received signals in the form of narrow-band spectral analysis to detect frequency-resolvable harmonic components of rotating structural elements of the observed air object, forming a spectral-Doppler portrait of the observed object, calculating an estimate of the rotation frequency of the low-pressure compressor rotor based on a known reference spectral component of the formed spectral-Doppler portrait, forming a database of reference portraits, the feature space of which consists of the number of blades in the first two stages of the low-pressure compressor or turbine and the permissible rotation frequencies of the low-pressure compressor rotor, comparing the formed spectral-Doppler portrait with portraits of air objects from the database of standards, calculating a diagonal matrix of errors in the estimates of the frequencies of the spectral components between the reference and the observed spectral-Doppler portrait, calculating according to the criterion maximum likelihood functional of the parameter estimation characterizing the type of aerial object, based on the truncated spectral-Doppler portrait, with the exclusion of the reference component, based on the weighted square of the difference in the frequency values of the spectral components of the received and generated reference spectral-Doppler portraits, calculation of the full matrix of errors of the difference in the frequency values of the compared spectral components, determination of the threshold according to the criterion with which the minimum value of the calculated weighted square of the difference in frequency values is compared; if the value is less than the threshold, a decision is made to recognize the type of aerial object whose spectral-Doppler portrait is contained in the reference database; otherwise, a decision is made to observe an unknown type of aerial object that is not contained in the reference database, characterized in that after calculating the reference spectral-Doppler portrait, an additional procedure is introduced for estimating the rotation frequencies of engines in a Z-dimensional search window, the dimension of which is determined by the number of engines on the aerial object, and a likelihood functional is also calculated, determined by the difference in frequencies of the observed spectral components and the spectral components of the reference portrait with the Z-number of engines with a set range of changes in the rotation frequencies of the rotors of their low-pressure compressors corresponding to the given type of engines, to take into account their asynchronous operation.