A radar antenna scanning type recognition method based on a lightGBM algorithm

Abstract

The application relates to a radar antenna scanning type recognition method based on a lightGBM algorithm and belongs to the technical field of electronic countermeasure reconnaissance. The method comprises the following steps: sampling a pulse signal to obtain a pulse sequence; converting the amplitude value of the pulse sequence into a voltage value to obtain a pulse amplitude data sequence; performing maximum main lobe extraction on the pulse amplitude data sequence after normalization processing to obtain the features of the main lobe sequence; performing autocorrelation processing on the pulse amplitude data sequence after the normalization processing to obtain relevant pulse amplitude data in multiple antenna scanning periods; performing smoothing processing on the relevant pulse amplitude data in the multiple antenna scanning periods to determine the characteristic parameters of the pulse amplitude data after the smoothing processing, and then performing radar antenna scanning type recognition according to the characteristic parameters. Compared with a traditional decision tree (DT) and a support vector machine (SVM), the application can more accurately recognize the radar antenna scanning mode.

Description

Technical Field

[0001] This invention belongs to the field of electronic countermeasures and reconnaissance technology, and relates to a radar antenna scanning type identification method based on the lightGBM algorithm. Background Technology

[0002] Radar countermeasures reconnaissance refers to an electronic countermeasures reconnaissance method that searches for, intercepts, measures, analyzes, and identifies enemy radar signals to obtain intelligence information such as their technical parameters, functions, types, and locations. Obtaining radar radiation source parameter information includes signal parameters, location parameters, and functional parameters. Signal parameters mainly describe the characteristics of the radar pulse, such as pulse amplitude, pulse width, carrier frequency, and repetition period. Location parameters involve the radar's spatial location and platform information; functional parameters include the radar type and operating status. Currently, the analysis of intercepted radar radiation source signals is mainly based on information such as pulse width, carrier frequency, repetition period, and intra-pulse modulation to achieve radar signal sorting and individual identification. However, this information is not sufficient for assessing the threat level of radar, especially in identifying radar antenna scanning types. How to utilize intercepted radar pulse data to analyze and judge the radar's operating status and threat level is an important research direction in electronic countermeasures reconnaissance.

[0003] A radar antenna is a device used to radiate electromagnetic wave energy, and its characteristics directly reflect the radar's operational performance. The primary tasks of radar are target detection, tracking, and identification. To achieve these tasks, the radar antenna needs to search a designated airspace in a specific manner; this is called antenna beam scanning. However, the beam pattern of a radar antenna typically only covers a portion of the area of ​​interest. To search and detect targets over a larger area, the radar antenna beam needs to move in space; this beam movement, i.e., the change in angle over time, is called scanning. Different types of radar use different antenna beam shapes and antenna scan styles (ASTs) for different purposes and operating conditions. For example, early warning radar is mainly used for target search, and its antenna scan type typically employs circular scanning or bidirectional sector scanning. In electronic countermeasures reconnaissance, accurately determining the antenna scan type of enemy radar is crucial for identifying the radar's operational status and assessing its threat level.

[0004] Radar antenna scanning methods can be divided into mechanical scanning and electronic scanning. Mechanical scanning uses the mechanical rotation of the radar antenna to move the beam in space. This scanning method typically repeats at a certain period and mainly includes circular scanning, planar sector scanning (unidirectional sector scanning, bidirectional sector scanning, multi-sector scanning), helical scanning, and row scanning. Electronic scanning, on the other hand, uses electronic technology to move the beam without mechanical rotation and mainly includes one-dimensional electronic scanning and two-dimensional electronic scanning. Electronic scanning is more flexible and faster than mechanical scanning and can achieve more complex scanning modes. Identifying radar antenna scanning types is an important task in electronic countermeasures reconnaissance, which mainly relies on the selection of characteristic parameters, parameter estimation, and identification methods. Characteristic parameters are key indicators used to describe the antenna scanning type, including scanning period, kurtosis, number of main lobes, maximum difference in main lobe peak values, main lobe peak time interval ratio, and main-side lobe gain ratio. Accurate estimation of these parameters is crucial for identifying radar antenna scanning types; however, the estimation performance of characteristic parameters is affected by various factors, such as noise, pulse loss, and "glitch" interference pulses. How to identify the radar antenna scanning type has become a key issue in radar threat assessment and situational analysis, and it is also a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a radar antenna scanning type identification method based on the lightGBM algorithm. This method uses radar feature parameters, including new features, to achieve more accurate identification of radar antenna scanning modes.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A radar antenna scan type identification method based on the lightGBM algorithm includes the following steps:

[0008] S1. The intercepted radar signal power is sampled to obtain a pulse sequence;

[0009] S2. Convert the amplitude values ​​of the pulse sequence into voltage values ​​to obtain the pulse amplitude data sequence, and then normalize the pulse amplitude data sequence.

[0010] S3. Extract the maximum main lobe sequence from the normalized pulse amplitude data sequence to obtain the maximum main lobe feature parameters;

[0011] S4. Distinguish between mechanical and electronic scanning of the radar antenna by the maximum main lobe characteristic parameter; if it is electronic scanning, distinguish the dimension of electronic scanning and end the identification; if it is mechanical scanning, proceed to the next step.

[0012] S5. Perform autocorrelation processing on the normalized pulse amplitude data sequence to obtain correlated pulse amplitude data within multiple antenna scanning cycles;

[0013] S6. Extract features from the relevant pulse amplitude data sequences within multiple antenna scanning cycles. The features include kurtosis, number of main lobes, maximum difference in main lobe peak values, maximum ratio of main lobe time intervals, ratio of period to 3dB width of maximum main lobe, number of silent intervals, and maximum value of the first-order difference between the beginning and end of the silent interval.

[0014] S7. Identify the radar antenna mechanical scanning type in the lightGBM classifier based on the features extracted in step S6.

[0015] Furthermore, in step S3, the maximum main lobe feature parameters include the maximum value M of the first-order difference of the maximum main lobe sequence. d The proportion R of the first-order difference of the largest main lobe sequence that is less than the threshold after normalization. d The mean square error V of the sequence consisting of the first-order difference of the largest main lobe sequence, after normalization, and values ​​less than a threshold. d .

[0016] Therefore, if the characteristic parameter M d and R d If the value is large, the radar antenna scanning type is electronic scanning. If the characteristic parameter V... d If the characteristic parameter V is large, then the radar antenna is a one-dimensional electronically scanned array. d If the size is small, the radar antenna is a two-dimensional electronically scanned array (ESA).

[0017] Further, in step S6, the ratio of period to the maximum main lobe 3dB width represents the width corresponding to when the maximum amplitude of the main lobe decreases to 0.707 times the original amplitude. Its extraction method is as follows: in the pulse amplitude sequence {x} of a single radar scanning cycle... r Find the coordinates of the peak point of the maximum main lobe, then find the points on both sides of the peak point where the pulse amplitude drops to 0.707 times the peak value. Take the distance between these two points as the 3dB width of the maximum main lobe. Then the ratio of the period to the 3dB width of the maximum main lobe is:

[0018]

[0019] In the formula, T 3dB T represents the width of the maximum main lobe (3dB). p This indicates the antenna scanning period.

[0020] In step S6, the method for extracting the number of silent intervals is as follows: in a single radar scan cycle pulse amplitude sequence {x r Find the point k where the radar gain is almost zero. i From k iSearch to the right for the first point k where the radar gain is not zero. i +Δk, calculate the time length corresponding to Δk. If it is greater than 0, then Δk is a silent interval; find the sequence {x}. r The number of silent intervals is obtained from all silent intervals in}.

[0021] Furthermore, step S2 also includes: determining whether the repetition period of the pulse amplitude data sequence is a fixed repetition period; if yes, proceed to step S3; if no, return to step S1 to resample the pulse signal.

[0022] When the pulse repetition period of the pulse amplitude data sequence is jittery, the resampling interval is the average value of the pulse repetition intervals of the pulse amplitude data sequence; when the pulse repetition period of the pulse amplitude data sequence is uneven or slipping, the resampling interval is the minimum pulse repetition interval of the pulse amplitude data sequence.

[0023] Further, in step S3, the method for extracting the maximum main lobe sequence is as follows: first, find the peak value of the main lobe and the peak value of the side lobes in the pulse amplitude data sequence, set the peak value of the side lobes as the threshold value, and then find the points where the pulse amplitude values ​​on the left and right sides of the main lobe peak drop to the threshold value, and take the signal between the two points as the maximum main lobe sequence.

[0024] The beneficial effects of this invention are as follows: This invention identifies the scanning type of a radar antenna by extracting characteristic parameters of the radar pulse signal. Specifically, for mechanical scanning types, this invention proposes three new features to improve the accuracy of type identification, which is of great significance for applications such as target identification, target tracking, and target localization in radar systems. This invention can achieve automatic identification of radar antenna scanning types without manual assistance, representing a fundamental improvement to typical automatic target threat assessment technologies. Compared to traditional decision trees (DT) and support vector machines (SVM), it has significant advantages. The relationship between the probability of successful radar antenna scanning type identification and the signal-to-noise ratio (SNR) of this invention is as follows:

[0025] Under low signal-to-noise ratio (SNR) conditions, with an SNR of around 10dB, the accuracy rate for identifying radar antenna scan types is no less than 74%; under higher SNR conditions, with an SNR of around 30dB, the accuracy rate for identifying radar antenna scan types is 93.54%; and under extremely high SNR conditions, with an SNR of around 50dB, the accuracy rate for identifying radar antenna scan types reaches as high as 99.39%.

[0026] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0027] To make the objectives, technical solutions, and advantages of the present invention clearer, the preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings, wherein:

[0028] Figure 1 A schematic diagram of the radar antenna scan type identification method based on the lightGBM algorithm provided in this embodiment of the invention;

[0029] Figure 2 This is a schematic diagram illustrating the probability of radar antenna scanning type identification according to an embodiment of the present invention;

[0030] Figure 3 A comparison chart of the total probability of radar antenna scanning type identification according to an embodiment of the present invention with the total probability of identification by traditional decision tree (DT) and support vector machine (SVM);

[0031] Figure 4 This is a schematic diagram illustrating how the application of new features in this embodiment of the invention improves the recognition accuracy. Detailed Implementation

[0032] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0033] like Figure 1 The figure shows a radar antenna scan type identification method based on the light GBM algorithm provided in an embodiment of the present invention. The method includes:

[0034] S1. Sample the radar pulse signal to obtain a pulse sequence.

[0035] The sampling process is as follows:

[0036] S11. Obtain radar countermeasures and reconnaissance parameters to determine the intercepted radar signal power;

[0037] In this embodiment, the radar countermeasures and reconnaissance parameters include radar transmit power, radar transmit antenna receiving gain in the direction of electronic warfare receiver, radar azimuth and elevation angles relative to electronic warfare receiver, radar signal wavelength, and distance between the electronic warfare reconnaissance platform and radar.

[0038] The intercepted radar signal power can be expressed as:

[0039]

[0040] In the formula, P t λ represents the radar transmit power, λ represents the radar signal wavelength, R represents the distance between the electronic warfare reconnaissance platform and the radar; L represents the system loss factor, including atmospheric propagation loss and polarization mismatch loss between the radar antenna and the electronic warfare reconnaissance platform antenna; θ t (t) represents the azimuth angle, φ t (t) represents the elevation angle; G t [θ t (t),φ t [(t)] represents the gain of the radar transmitting antenna in the direction of the electronic warfare receiver, which can be expressed as follows:

[0041] G t [θ t (t),φ t [(t)]=G T F t [θ t (t),φ t (t)] (2)

[0042] In the formula, G T F represents the maximum gain of the antenna; where F t [θ t (t),φ t [t] is represented as:

[0043]

[0044] The direction of the main beam can be changed by increasing or decreasing k1. A larger k1 value makes the main beam more focused in a specific direction, while a smaller k1 value makes the main beam's radiation range wider. This can be used to adjust the antenna's pointing. The width of the main beam can be changed by adjusting k2. A larger k2 value makes the main beam narrower, while a smaller k2 value makes the main beam wider. This can be used to adjust the antenna's radiation range and radiation angle. The number and distribution of the antenna's sidelobes can be changed by adjusting k3, thus changing the beam pointing. A larger k3 value indicates that the antenna has more sidelobe radiation energy, while a smaller k3 value indicates that the antenna has relatively less sidelobe radiation energy.

[0045] S12. The power of the intercepted radar signal is sampled by pulse signal to obtain a pulse sequence.

[0046] First, P r (t) is converted to decibel representation;

[0047] P r,dBm (t)=P t,dBm +Gt,dB -L dB (4)

[0048] Among them, P r,dBm (t) represents the power value of the radar signal intercepted by the reconnaissance antenna, P t,dBm G represents the radar transmit power value. t,dB L represents the antenna transmit gain of the radar antenna in the direction of the reconnaissance antenna. dB Let P represent the system loss factor, and t represent time. r,dBm (t), P t,dBm L dB All are in dB format.

[0049] Considering the effects of pulse loss and noise, the actual received signal is:

[0050] P rc,dB (t)=(1-X m (1-X) s )P r,dBm (t)+(1-X m )X s A s +N dB (t) (5)

[0051] Among them, P rc,dB (t) represents the actual received signal, X m and X s Let A represent the Bernoulli random variables representing the occurrence of pulse loss and "glitch" interference pulse events, respectively. s N represents the amplitude of the "glitch" interference pulse. dB (t) represents Gaussian white noise.

[0052] For the actual received signal P rc,dB (t) is sampled to obtain a digital signal.

[0053]

[0054] In the formula, n represents the pulse sequence value of the digital signal, n = 0, 1, ..., N p -1; T s N represents the sampling or resampling period of an analog signal. p This indicates the length of the pulse amplitude data sequence.

[0055] When sampling or resampling received data, if there is no radar signal at a certain sampling time, the nearest neighbor interpolation method can be used for interpolation processing. Specifically, when the sampling value at time t is missing in the pulse sequence, the sampling value at the nearest time of time t is interpolated to obtain the sampling value at time t. The nearest time of time t is time t-1 or time t+1.

[0056] S2. Convert the amplitude values ​​of the pulse sequence into voltage values ​​to obtain the pulse amplitude data sequence.

[0057] The amplitude value of the received pulse sequence is converted from dB to voltage value using the following formula:

[0058]

[0059] in, This represents a pulse amplitude data sequence.

[0060] S3. Normalize the pulse amplitude data sequence.

[0061] To eliminate the effects of different transmission losses and radar transmit power, and to avoid the influence of factors other than angle-related amplitude differences, the pulse amplitude data sequence is normalized as shown in the following formula:

[0062]

[0063] In the formula, This represents the largest pulse amplitude data in the pulse amplitude data sequence.

[0064] The normalized data is denoised to reduce the impact of noise on the radar sequence characteristics. It is then determined whether the repetition period of the pulse amplitude data sequence is a fixed repetition period. If yes, proceed to step S4; otherwise, return to step S1 to resample the pulse signal.

[0065] Wherein, when the pulse repetition period of the pulse amplitude data sequence is jitter, the resampling interval is the average value of the pulse repetition interval of the pulse amplitude data sequence;

[0066] When the pulse repetition period of the pulse amplitude data sequence is uneven or slip-slip, the resampling interval is the minimum pulse repetition interval of the pulse amplitude data sequence.

[0067] S4. Extract the maximum main lobe sequence from the normalized pulse amplitude data sequence to extract the corresponding maximum main lobe feature parameters to distinguish between mechanical scanning and electronic scanning of the radar antenna scanning mode.

[0068] Since the obtained pulse amplitude data includes the main lobe sequence and the side lobe sequence, and the value of the main lobe sequence is larger and the value of the side lobe sequence is smaller, the extraction method of the main lobe sequence is as follows: first find the peak value of the main lobe and the peak value of the side lobe, set the peak value of the side lobe as the threshold H, and then find the points where the pulse amplitude values ​​on both sides of the main lobe peak drop to H, and take the signal between these two points as the maximum main lobe sequence.

[0069] The method for feature extraction of the largest main lobe sequence is as follows:

[0070] Let the maximum main lobe pulse amplitude sequence be a m [n], first calculate the absolute value of the first-order difference of the maximum main lobe pulse amplitude sequence:

[0071] d a [n] = |a m [n+1]-a m [n] (9)

[0072] Where n = 0, 1, ..., N p -2. Take the characteristic parameter M d For sequence {d a The maximum value of} is generally the M value of electronic scanning. d Larger, mechanically scanned M d Smaller.

[0073] Furthermore, in order to distinguish pulse trains within the same wave position of electronic scanning, the sequence {d} is... a Normalize:

[0074] u d [n] = d a [n] / max(d a [n]) (10)

[0075] Let sequence {u d} less than H d The elements form a new sequence {h d}. Among them, H d This represents a preset threshold, typically slightly greater than 0. The feature parameter R is taken. d =N d / (N p -1), then R d This represents the proportion of pulses whose adjacent pulses are all within the same wavelength, where N d Represents {h d The length of}. R in typical mechanical scanning. d Smaller, electronically scanned R d Relatively large.

[0076] Meanwhile, in order to distinguish between one-dimensional and two-dimensional electronic scanning, the characteristic parameter V is taken. dFor sequence {h d The mean squared error of}. Using the characteristic parameter M. d R d V d The distinction between electronic scanning and mechanical scanning will be explained in detail in subsequent steps.

[0077] S5. Perform autocorrelation processing on the normalized pulse amplitude data sequence to obtain correlated pulse amplitude data within multiple antenna scanning periods. Specifically:

[0078] S51. The normalized pulse amplitude data sequence is resampled for subsequent radar scan period estimation.

[0079] The pulse arrival time is t n =t n-1 +Δt, where t n and t n-1 Let Δt represent the arrival times of the nth and (n-1th)th pulses of the radar signal, respectively, and Δt be the change in pulse arrival time. The change in pulse arrival time is related to the type of pulse repetition period (PRI). However, due to the influence of PRI variations and pulse loss, the sampling rate of the sequence is inconsistent, which is inconvenient for subsequent processing. Therefore, the pulse amplitude sequence needs to be resampled to unify the sampling rate. It can be assumed that the signal intercepted by the receiver is a discrete sample of a continuous signal a(t), and therefore resampling can be performed using the following formula:

[0080] x[n]=a(nT s (11)

[0081] In the formula, n = 0, 1, ..., N-1, N = t[N p -1] / T s Using floor function, T s The sampling period.

[0082] S52. Perform normalized autocorrelation processing to obtain the antenna scanning period.

[0083] The normalized autocorrelation function of the sequence x[n] is calculated as follows:

[0084]

[0085] In the formula, l = 0, 1, ..., NW is the delay variable, and W is the window length. Then the sequence {r xx The maximum value of} corresponds to l (denoted as N). l Let be the period of sequence x[n], and let T be the antenna scanning period. p =N l *T s .

[0086] S6. Feature extraction is performed on the relevant pulse amplitude data within multiple antenna scanning cycles to obtain feature parameters for various types of mechanical scanning. Feature extraction for mechanical scanning identification is based on the pulse amplitude sequence {x} of a single radar scanning cycle. r}, denoted as X r Its signal length is N r .

[0087] S61. Extract kurtosis; kurtosis reflects the sequence X. r The smoothness and sharpness of the surface can be extracted using the following formula:

[0088] K X =(E[X r -μ]) 4 / δ 4 (13)

[0089] Where E[·] represents the expected value, and μ and δ represent the sequence X respectively. r The mean and standard deviation.

[0090] S62, Extract the number of main lobes.

[0091] The process of obtaining the number of main lobes (the number of main lobes in a single cycle) of the smoothed pulse amplitude data is as follows:

[0092] Let the obtained maximum main lobe sequence be {y}, and its signal length be N. y Through the sequence {x r The normalized cross-correlation of {y} and {y} is used to detect other main lobes within a single antenna scan period:

[0093]

[0094] Where l = 0, 1, ..., N r -N y Let N be the delay variable. Then, the number of main lobes N within a single antenna scanning period... B For sequence {r xy} greater than H b The number of elements, where H b ∈[0.98,1).

[0095] When the number of main lobes is greater than 1, the maximum difference D between the main lobe peak values ​​is extracted. Y Let the peak values ​​of each main lobe be x. r [m i ], m i For the peak values ​​of each main lobe in the sequence {x r The coordinates in}, where i = 1, 2, ..., N B Feature D Y Extract using the following formula:

[0096] D Y =max({x r})-min({x r}) (15)

[0097] When the number of main lobes is greater than 2, it is also necessary to extract the maximum ratio R of the main lobe time intervals. Y Extraction is performed using the following formula:

[0098] R Y =max(Y m ) / min(Y m (16)

[0099] Among them, Y m ={m2-m1,m3-m2,…,m NB -m NB-1}

[0100] S7. Extract the new features proposed in this invention to improve recognition accuracy. The new features proposed in this invention include: the ratio of period to the maximum main lobe 3dB width, the number of silent intervals, and the maximum value of the first-order difference between the beginning and end of the silent interval.

[0101] S71. The ratio of extraction period to the maximum main lobe 3dB width, where the 3dB width is defined as the width corresponding to the main lobe's maximum amplitude decreasing to 0.707 times its original amplitude.

[0102] First, find the peak point of the largest main lobe and its position in the sequence {x}. r Find the coordinates w1 in the equation, and then find the closest points on both the left and right sides where the pulse amplitude values ​​drop to 0.707 times their original values. Take the distance between these two points as the width of the maximum main lobe (3dB). This feature is shown in the following formula:

[0103]

[0104] In the formula, T 3dB T represents the width of the maximum main lobe (3dB). p This indicates the antenna scanning period.

[0105] S72, Extract the number of silent intervals.

[0106] When a radar performs a mechanical scan, such as a unidirectional sector scan, the radar scanning beam reaches its endpoint after passing through a certain arc and needs to return to its starting position. The time interval during which the radar travels from the endpoint back to the starting point is called the silent interval. During the silent interval, the radar gain is close to zero. Therefore, the number of silent intervals can be extracted based on this characteristic.

[0107] Specifically, first find the point where the radar gain is almost zero in the sequence {x}. r Position k in}i Then expand to the right until the first point with non-zero radar gain appears, which is in the sequence {x}. r The position in} is denoted as k. i +Δk, calculate the time length corresponding to Δk. If it is greater than 0.1s, it is considered to be a silent interval.

[0108] S73. Extract the maximum value of the first-order difference between the first and last points of the silent interval. This feature can be obtained using the first-order difference function built into MATLAB. Specifically, take about 100 points near the silent interval, perform a first-order difference, and find the maximum value.

[0109] S8. Use the lightGBM model to automatically identify the radar antenna scanning type based on the extracted feature parameters.

[0110] S81, Determine Feature M d and R d If the value is large, proceed to step S82; otherwise, proceed to step S83.

[0111] S82, Determine Feature V d If the value is large, it is determined to be a one-dimensional electrophysiological scan; otherwise, it is determined to be a two-dimensional electrophysiological scan.

[0112] S83, based on kurtosis K X Number of main lobes N B The maximum difference D of the motherboard peak value Y The maximum ratio R of the main lobe time interval Y The ratio of period to maximum main lobe 3dB width, the number of silent intervals, and the maximum value of the first-order difference between the beginning and end of the silent intervals are used to determine the specific mechanical scanning method using the LightGBM classifier. For example, if N B =1,D Y R Y The maximum value of the first-order difference between the first and last parts of the silent interval has no value, K X If the value is large, the ratio of the period to the maximum main lobe 3dB width is relatively large, and the number of silent intervals is zero, then the mechanical scanning type is determined to be circular scanning; if N B =1,D Y R Y No value, K X If the value is small, the ratio of the period to the maximum main lobe 3dB width is small, the number of silent intervals is 1, and the maximum value of the first-order difference between the beginning and end of the silent interval is large, then the mechanical scanning type is determined to be unidirectional scanning.

[0113] The probability of successfully identifying the radar antenna scan type using the method described in the above embodiments is shown in [reference]. Figure 2This embodiment converts the amplitude values ​​of the pulse sequence from dB to voltage values, then normalizes them to obtain correlated pulse amplitude data within multiple antenna scanning periods. Next, the correlated pulse amplitude data within multiple antenna scanning periods is smoothed to determine the characteristic parameters of the smoothed pulse amplitude data. Based on the maximum value of the first-order difference of the main lobe sequence (M) among the characteristic parameters... d The proportion of the main lobe sequence that is less than the threshold H value after first-order difference and normalization (R) d The mean square error (V) of the sequence is composed of the values ​​of the main lobe sequence that are less than the threshold H after the first difference. d This invention distinguishes between electronic and mechanical radar scanning. Based on characteristic parameters such as kurtosis, number of main lobes, maximum difference in main lobe peak values, maximum ratio of main lobe time intervals, ratio of period to 3dB main lobe width, number of silent intervals, and the maximum value of the first-order difference between the first and last silent intervals, it automatically identifies six types of radar antenna scanning. The application of a new classification algorithm and the introduction of new features give this invention significant advantages over traditional decision trees (DT) and support vector machines (SVM). Figure 3 As shown, under low signal-to-noise ratio (SNR) conditions, with an SNR of around 10dB, the accuracy rate of radar antenna scan type identification using this invention is no less than 74%; under higher SNR conditions, with an SNR of around 30dB, the accuracy rate of radar antenna scan type identification using this invention is 93.54%; and under high SNR conditions, with an SNR of around 50dB, the accuracy rate of radar antenna scan type identification using this invention is as high as 99.39%.

[0114] Furthermore, this invention improves the accuracy of classification results by introducing new features, such as... Figure 4 As shown, the improvement is particularly significant under low signal-to-noise ratio (SNR) conditions. At an SNR of 10 dB, the accuracy improved by 17%, and even at high SNR, the classification results improved by 5%. This demonstrates that the introduction of new features greatly improves the accuracy of classification results.

[0115] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A radar antenna scan type identification method based on the lightGBM algorithm, characterized in that: The method includes the following steps: S1. The intercepted radar signal power is sampled to obtain a pulse sequence; S2. Convert the amplitude value of the pulse sequence into a voltage value to obtain a pulse amplitude data sequence, and normalize the pulse amplitude data sequence. S3. Extract the maximum main lobe sequence from the normalized pulse amplitude data sequence to obtain the maximum main lobe feature parameters; S4. The radar antenna is distinguished as mechanically scanned or electronically scanned based on the maximum main lobe feature parameters. If it is electronically scanned, the dimensions of the electronic scan are distinguished, and the identification ends. If it is mechanically scanned, proceed to the next step. S5. Perform autocorrelation processing on the normalized pulse amplitude data sequence to obtain correlated pulse amplitude data within multiple antenna scanning cycles; S6. Feature extraction is performed on the relevant pulse amplitude data sequence within multiple antenna scanning cycles. The features include kurtosis, number of main lobes, maximum difference in main lobe peak values, maximum ratio of main lobe time intervals, ratio of period to maximum main lobe 3dB width, number of silent intervals, and maximum value of the first-order difference between the beginning and end of the silent intervals. The maximum value of the first-order difference between the beginning and end of the silent intervals is extracted by performing a first-order difference operation on one hundred data points near the silent interval and extracting the maximum value of the first-order difference. The ratio of the period to the maximum main lobe 3dB width represents the width corresponding to when the maximum amplitude of the main lobe drops to 0.707 times the original amplitude. It is extracted by analyzing the pulse amplitude sequence within a single radar scan cycle. Find the coordinates of the peak point of the maximum main lobe, and then find the points on both sides of the peak point where the pulse amplitude drops to 0.707 times the peak value. Take the distance between the two points as the 3dB width of the maximum main lobe. Then the ratio of the period to the 3dB width of the maximum main lobe is: In the formula, This indicates the width of the maximum main lobe, which is 3dB. T p Indicates the antenna scanning period; The method for extracting the number of silent intervals is as follows: in the pulse amplitude sequence of a single radar scan cycle. Find the point where the radar gain is almost zero. ,from Search to the right for the first point where the radar gain is not zero. ,calculate If the corresponding time length is greater than 0.1 s, then Given a silent interval; search for a sequence The number of silent intervals is obtained from all silent intervals. S7. Identify the radar antenna mechanical scanning type in the lightGBM classifier based on the features extracted in step S6.

2. The radar antenna scanning type identification method according to claim 1, characterized in that: In step S3, the maximum main lobe feature parameters include the maximum value of the first-order difference of the maximum main lobe sequence. M d The proportion of the largest main lobe sequence with first-order difference that is less than the threshold after normalization. R d The mean square error of the sequence consisting of the first-order difference of the largest main lobe sequence, after normalization and being less than a threshold value. V d .

3. The radar antenna scanning type identification method according to claim 1, characterized in that: Step S2 further includes: determining whether the repetition period of the pulse amplitude data sequence is a fixed repetition period; if yes, proceed to step S3; if no, return to step S1 to resample the pulse signal. When the pulse repetition period of the pulse amplitude data sequence is jittery, the resampling interval is the average value of the pulse repetition intervals of the pulse amplitude data sequence; when the pulse repetition period of the pulse amplitude data sequence is uneven or slipping, the resampling interval is the minimum pulse repetition interval of the pulse amplitude data sequence.

4. The radar antenna scanning type identification method according to claim 1, characterized in that: In step S3, the extraction method of the maximum main lobe sequence is as follows: first, find the peak value of the main lobe and the peak value of the side lobes in the pulse amplitude data sequence, set the peak value of the side lobes as the threshold value, and then find the points where the pulse amplitude values ​​on the left and right sides of the main lobe peak drop to the threshold value, and take the signal between the two points as the maximum main lobe sequence.

Images