Rapid radar signal sorting method based on simplified multi-parameter clustering

Abstract

The invention discloses a rapid radar signal sorting method based on simplified multi-parameter clustering, and the method comprises the steps: carrying out the matching of frequency and pulse width parameters in a pulse description word with each known radiation source parameter in a target library, and carrying out the shunting of the pulse description word according to a matching calculation result; when successfully matching with the target library, acquiring the current phase difference of the equipment, calculating the azimuth angle and the pitch angle at the current moment, and outputting the target number, the frequency, the pulse width, the repetition frequency and the calculated angle information in the library in real time; a K-Means-based dynamic clustering algorithm is selected, simplified pulse parameters are used, and unknown pulse data streams are divided into a plurality of clusters through two parameters of frequency and pulse width; taking the pulse arrival time sequence corresponding to each clustering result as the input of main sorting in sequence, and carrying out pulse repetition period sorting and detection on unknown radar signals; and carrying out harmonic verification on the counted first-level pulse arrival time difference, judging the type of the radar, and sorting the repetition frequency of the single radar.

Description

Technical Field

[0001] The present invention relates to the technical field of signal processing, and in particular to a fast radar signal sorting method based on simplified multi-parameter clustering. Background Art

[0002] Radar signal sorting is an important step in radar signal processing, especially in complex electromagnetic environments. With the increase in the number of radars on modern battlefields, multiple radar signals are superimposed, interfered and crossed, making the received signals more complex. Therefore, in order to effectively extract the required information from complex signals, signal sorting technology becomes necessary. Its main task is to identify, separate and classify radar signals from different sources from mixed signals in order to analyze and identify the target radar.

[0003] Pulse repetition frequency and pulse arrival time are two important characteristic parameters in radar signal sorting. PRF (Pulse Repetition Frequency) of different radar systems is usually different, while TOA (Time of Arrival) provides information about the time distribution of radar pulses. By analyzing PRF and TOA, the received signals can be preliminarily grouped. However, with the diversification of radar signal forms, these two parameters alone are no longer sufficient to accurately sort all signals.

[0004] In the case of PRF overlap, PW (Pulse Width), CF (Carrier Frequency) and DOA (Direction of Arrival) provide additional classification basis. Preprocessing the signal with these parameters can effectively distinguish different signal sources. However, due to the multipath effect and reflection in the electromagnetic environment, DOA measurement may produce errors, and the angle measurement of multiple pulses is also easy to consume a lot of time.

[0005] With the introduction of multi-feature fusion and complex algorithms, the computational complexity of the algorithm has increased significantly, making it difficult for the sorting algorithm to meet real-time requirements, especially in scenarios with high real-time requirements such as battlefields, which may lead to sorting delays or response lags. Summary of the invention

[0006] The purpose of the present invention is to provide a fast radar signal sorting method based on simplified multi-parameter clustering to solve the problems of angle measurement errors and time-consuming multi-pulse angle measurement in actual electromagnetic environments.

[0007] In order to solve the above technical problems, the present invention provides a fast radar signal sorting method based on simplified multi-parameter clustering, comprising:

[0008] Step S1: Preprocessing of known radar signals: matching the frequency and pulse width parameters in the pulse description word with each known radiation source parameter in the target library, and dividing the pulse description word according to the result of the matching calculation, so as to prepare for the main signal sorting;

[0009] Step S2: Main processing of known signals: When successfully matched with the target library, obtain the current phase difference of the device, calculate the azimuth and elevation angle at the current moment, and output the target number, frequency, pulse width, repetition rate and calculated angle information in the library in real time;

[0010] Step S3: Pre-sorting of unknown signals: Using a dynamic clustering algorithm based on K-Means, using simplified pulse parameters, and dividing the unknown pulse data stream into multiple clusters by two parameters: frequency and pulse width;

[0011] Step S4: Main sorting of unknown signals: The pulse arrival time series corresponding to each clustering result is used as the input of the main sorting in turn, and the pulse repetition period of the unknown radar signal is sorted and detected; the statistical primary pulse arrival time difference is harmonically checked, and then the radar type is judged, and the repetition frequency of a single radar is sorted out.

[0012] In one embodiment, the pulse description word PDW input in step S1 includes N pulses, PDW=[d1,…d i ,…d N ], PDW i =[CF i ,PW i ,TOA i ], CF i is the ith carrier frequency, PW i is the ith pulse width, TOA i is the arrival time of the ith pulse; there is no parameter information in the initial target library. In actual engineering, after the previous packet of data is sorted, it is updated to the target library as a known signal. The target library contains frequency, pulse width and repetition rate. target = [cf target ,pw target ,prf target ], where cf target is the frequency in the target library, pw target is the pulse width in the target library, prf target is the repetition frequency in the target library;

[0013] Perform frequency and pulse width matching calculations to dilute and divert the input pulse descriptor. When the parameters are matched, as long as the difference is within the tolerance range, the match is considered successful. The pulse descriptor PDW sequence corresponding to the matching of both carrier frequency CF and pulse width PW is used as the input for the main processing of known signals. The pulse descriptor PDW that does not match the radar target library is sent to the input of the unknown signal preprocessing.

[0014] In one embodiment, the n pulse description words PDW={d1, d2, ..d n} Perform dynamic clustering based on K-Means, including the following steps:

[0015] Step S31: taking the first data d1 of the pulse width PW and the carrier frequency CF in the pulse description word as the center value C1={PW1, CF1} of the first cluster;

[0016] Step S32: Read the next data d in the data set in sequence j = {PW j ,CF j}, and the center value C of the cluster k = {PW k ,CF k} to compare and determine whether the error satisfies |PW j -PW k |<Δpw and |CF j -CF k |<Δcf;

[0017] If the error is within the range, the match is successful, the data is merged into the cluster where the cluster center value is located, and the average value of the cluster is used instead of the cluster center value c. k ={(PW k +PW j ) / 2,(CF k +CF j ) / 2};

[0018] Step S33: After the center value is calculated, read the next data in the data set and continue to execute step S32; if it exceeds the error range, determine whether it has been compared with all cluster center values. If not, continue to compare with the next cluster center value C k-1 Compare until all cluster center values ​​are compared and still not matched, then add a new cluster category as the new cluster center value C k+1 = {PW k+1 ,CF k+1};

[0019] Step S34: Execute step S33 repeatedly until all data are classified and finally divided into M clusters, where M is the number of radars.

[0020] In one embodiment, in step S4, the TOA (time of arrival) sequence corresponding to each clustering result is sequentially used as the input of the main sorting, and it is considered as the TOA sequence of a single radar. The specific sorting process includes:

[0021] Step S41: Calculate the primary TOA difference, count the primary difference histogram within the tolerance range of the difference concept pri, perform sub-harmonic check on the pri that exceeds the threshold, and calculate the detection threshold as shown in formula (1):

[0022]

[0023] In formula (1), E is the total number of pulses, C is the level of the difference histogram, and N is the total scale value of the pulse interval on the histogram; τ is the independent variable of the histogram, x and k are constants, k∈[0,1], and the values ​​of x and k are adjusted according to the actual situation of the TOA sequence to improve the sorting effect of the algorithm;

[0024] Step S42: In actual engineering, the arrival time error is included, and the TOA sequence of a single radar is processed. The harmonic check is changed to: if one of the items in formula (2) is satisfied, it is judged as a harmonic, and the corresponding pri(j) and pri_total(j) are deleted to ensure that only the pri of a single radar is left;

[0025]

[0026] The pri_total array represents the number of pri values, and Δpri represents the pri tolerance. It is determined whether the remaining pri, i.e., ppri, varies between several different values. The ppri_total array represents the number of ppri values ​​updated after harmonic verification. It is visually observed through the simulation diagram of the pulse sequence and the first-level TOA difference, or adaptively determined through the following formula (3). If more than two ppri_total values ​​exceed the threshold, the jitter and stagger determination is performed, otherwise the conventional and jitter radar determination is performed.

[0027]

[0028] in represents the mean, represents the median, σ ppri_total represents the standard deviation of ppri_total, Δppri_total and Δppri are the set thresholds; in the case of pulse loss and arrival time error, the pri of the staggered radar is similar to the value in ppri_total, and the array value of ppri is more dispersed than that of the jitter radar; if the conditions of formula (3) are met, further determine whether they are met:

[0029] pri_total(p)-thr(p)>delta diff (4)

[0030] Where pri_total(p) indicates the pth pri_total value is the largest, thr(p) indicates the threshold calculated by formula (1), and delta diff is the set difference threshold; the pri distribution of the radar with a jitter rate of 15% is pri×(1±15%), and the corresponding values ​​in the pri_total array are relatively small; formula (4) determines the difference between its value and the overall threshold, and diff If the value is less than 100, it is considered as a jitter radar.

[0031] If one ppri_total exceeds the threshold and the part exceeding the threshold is large, it is judged as a regular radar; if no part exceeds the threshold or some parts exceed the threshold and are not much different from the threshold line, it means that they are scattered in multiple ppris, which reflects the characteristics of a jitter radar.

[0032] The present invention provides a fast radar signal sorting method based on simplified multi-parameter clustering, which reduces the complexity of data processing and significantly reduces the consumption of computing resources by focusing on core parameters (frequency and pulse width) during clustering; the main signal sorting effectively simplifies the calculation steps through fast feature screening and formula matching, and can quickly identify the radar type and its repetition frequency, thereby meeting real-time sorting requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The invention provides a design flow chart of a fast radar signal sorting method based on simplified multi-parameter clustering.

[0034] Figure 2 It is a flowchart of the known radar signal pre-sorting provided by the present invention.

[0035] Figure 3 It is a flowchart of the unknown radar signal pre-sorting provided by the present invention. DETAILED DESCRIPTION

[0036] The following is a further detailed description of a fast radar signal sorting method based on simplified multi-parameter clustering proposed by the present invention in conjunction with the accompanying drawings and specific embodiments. The advantages and features of the present invention will become clearer according to the following description. It should be noted that the accompanying drawings are all in a very simplified form and are not in precise proportions, and are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention.

[0037] The present invention provides a fast radar signal sorting method based on simplified multi-parameter clustering, the process of which is as follows: Figure 1 As shown, the method comprises the following steps:

[0038] Step S1: Preprocess the known radar signal: match the frequency and pulse width parameters in the PDW (Pulse Description Word) with each known radiation source parameter in the target library, and shunt the pulse description word according to the result of the matching calculation to prepare for the main signal sorting;

[0039] Step S2: Perform main processing on the known signal: when successfully matched with the target library, obtain the current phase difference of the device, calculate the azimuth and elevation angle at the current moment, and output the matching result in real time, output the target number, frequency, pulse width, repetition rate and calculated angle information in the library;

[0040] Step S3: Pre-sorting the unknown signal: using a dynamic clustering algorithm based on K-Means, using simplified pulse parameters, and dividing the unknown pulse data stream into multiple clusters by two parameters: frequency and pulse width;

[0041] Step S4: perform primary sorting on unknown signals: use the TOA sequence corresponding to each clustering result as the input of the primary sorting in turn, perform PRI (Pulse Repetition Interval) sorting and detection on the unknown radar signal, perform harmonic verification on the statistical primary TOA difference, and then judge the radar type and sort out the repetition frequency of a single radar.

[0042] In this embodiment, 2554 radar pulse signals received within 100 milliseconds are sorted and processed. Each pulse is recorded by a pulse description word (PDW), which contains the following main parameters and corresponding units: Time of arrival (TOA): in nanoseconds (ns), identifies the precise arrival time of the pulse, and is used for subsequent calculation of the pulse repetition interval and frequency. Carrier frequency (CF): in megahertz (MHz), that is, the center frequency of the signal, used for frequency matching and signal classification. Pulse width (PW): in nanoseconds (ns), describes the duration of the pulse, and is an important feature for sorting.

[0043] like Figure 2 As shown in the figure, the target library in the known radar signal preprocessing currently contains two target information {CF1, PW1, PRF1, CF2, PW2, PRF2}, and the input PDW is matched with the frequency and pulse width, and then the pulse description word is diluted and diverted. The tolerance of CF is ±5MHz, and the tolerance of PW is PW×10%. When the parameters are matched, as long as the difference is within the tolerance range, the match is considered successful. The PDW sequence corresponding to the matching of both CF and PW parameters is used as the input of the main processing of the known signal; the pulse description words that do not match the radar target library are sent to the input of the unknown signal preprocessing (clustering). The known signal pre-sorting results are shown in Table 1:

[0044] Before treatment After treatment Number of pulses 2554 1785

[0045] Table 1 Pre-sorting results of known signals

[0046] like Figure 3 As shown in the figure, the 1785 unknown pulse signals remaining after known radar preprocessing are dynamically clustered based on K-Means, including the following steps:

[0047] Step S31: First, the first data d1 of PW and CF in the pulse description word is used as the center value C1 = {PW1, CF1} of the first cluster.

[0048] Step S32: Then read the next data d in the data set in sequence j = {PW j ,CF j}, and the center value C of the cluster k = {PW k ,CF k} to compare and determine whether the error satisfies:

[0049] |PW j -PW k |<Δpw and |CF j -CF k |<Δcf, where Δpw=PW×10%, Δcf=5MHz. If it is within the error range, the match is successful, and the data is merged into the cluster where the cluster center value is located, and the average value of the cluster is used instead of the cluster center value C k ={(PW k +PW j ) / 2,(CF k +CF j ) / 2}.

[0050] Step S33: After the center value is calculated, read the next data in the data set and continue to execute step S32; if it exceeds the error range, determine whether it has been compared with all cluster center values. If not, continue to compare with the next cluster center value C k-1 Compare until all cluster center values ​​are compared and still not matched, then add a new cluster category as the new cluster center value C k+1 = {PW k+1 ,CF k+1}.

[0051] Step S34: Execute step S33 repeatedly until all data are classified and finally divided into 5 clusters; the unknown signal pre-sorting results are shown in Table 2:

[0052] Number of pulses Cluster 1 277 Cluster 2 625 Cluster 3 175 Cluster 4 500 Cluster 5 208

[0053] Table 2 Unknown signal pre-sorting (clustering) results

[0054] The TOA sequence corresponding to each clustering result is used as the input of the main sorting, and it is considered as the TOA sequence of a single radar. The specific sorting process includes:

[0055] Step S41: Calculate the primary TOA difference, count the primary difference histogram within the tolerance range of pri (difference concept), and perform sub-harmonic verification on pri exceeding the threshold. The calculation formula of the detection threshold is shown in formula (1):

[0056]

[0057] Where E is the total number of pulses, C is the level of the difference histogram, N is the total scale value of the pulse interval on the histogram; τ is the independent variable of the histogram, x and k are constants, k∈[0,1], x is 0.2, and k is 0.9.

[0058] Step S42: In actual projects, arrival time errors are included and the TOA sequence of a single radar is processed. The harmonic check is changed to: if one of the items in formula (2) is satisfied, it is judged as a harmonic, and the corresponding pri(j) and pri_total(j) are deleted to ensure that only the pri of a single radar is left.

[0059]

[0060] In this embodiment, Δpri=5us, and the pri_total array represents the number of pri values. It is determined whether the remaining pri, i.e., ppri, varies between several different values. The ppri_total array represents the number of ppri values ​​updated after harmonic verification. It can be observed intuitively through the simulation diagram of the pulse sequence and the first-level TOA difference, or it can be adaptively determined through the following formula (3). If more than two ppri_totals exceed the threshold, the staggered and jitter judgments are performed, otherwise the conventional and jitter radar judgments are performed.

[0061]

[0062] in represents the mean, represents the median, σppri _total represents the standard deviation of ppri_total, Δppri_total and Δppri are the set thresholds. In this embodiment, Δppri total=5, Δppri=5us. In the case of pulse loss and arrival time error, the pri of the staggered radar is similar to the value in ppri_total, and the array value of ppri is more dispersed than that of the jitter radar. If the conditions of formula (3) are met, further determine whether

[0063] pri_total(p)-thr(p)>delta diff (4)

[0064] Where pri_total(p) indicates the pth pri_total value is the largest, thr(p) indicates the threshold calculated by formula (1), and delta diff is the set difference threshold. In this embodiment, delta diff =20. The pri distribution of a radar with a jitter rate of 15% is pri×(1±15%), and the corresponding values ​​in the pri_total array are relatively small. Formula (4) determines the difference between its value and the overall threshold value, and calculates the delta diff If the value is less than 20, it is judged as a jitter radar.

[0065] If one ppri_total exceeds the threshold and the part exceeding the threshold is large, it is judged as a regular radar; if no part exceeds the threshold or some parts exceed the threshold and are not much different from the threshold line, it means that they are scattered in multiple ppris, which reflects the characteristics of a jitter radar.

[0066]

[0067]

[0068] Table 3 Main sorting results of unknown signals

[0069] The above description is only a description of the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention. Any changes or modifications made by a person skilled in the art in the field of the present invention based on the above disclosure shall fall within the scope of protection of the claims.

Claims

1. A fast radar signal sorting method based on simplified multi-parameter clustering, characterized in that: include: Step S1: Preprocessing of known radar signals: matching the frequency and pulse width parameters in the pulse description word with each known radiation source parameter in the target library, and dividing the pulse description word according to the result of the matching calculation, so as to prepare for the main signal sorting; Step S2: Main processing of known signals: When successfully matched with the target library, obtain the current phase difference of the device, calculate the azimuth and elevation angle at the current moment, and output the target number, frequency, pulse width, repetition rate and calculated angle information in the library in real time; Step S3: Pre-sorting of unknown signals: Using a dynamic clustering algorithm based on K-Means, using simplified pulse parameters, and dividing the unknown pulse data stream into multiple clusters by two parameters: frequency and pulse width; Step S4: Main sorting of unknown signals: The pulse arrival time series corresponding to each clustering result is used as the input of the main sorting in turn, and the pulse repetition period of the unknown radar signal is sorted and detected; the statistical primary pulse arrival time difference is harmonically checked, and then the radar type is judged, and the repetition frequency of a single radar is sorted out.

2. The fast radar signal sorting method based on simplified multi-parameter clustering as claimed in claim 1, characterized in that: The pulse description word PDW input in step S1 includes N pulses, PDW=[d1, ...d i ,...d N ], PDW i =[CF i , PW i , TOA i ], CF i is the ith carrier frequency, PW i is the ith pulse width, TOA i is the arrival time of the ith pulse; there is no parameter information in the initial target library. In actual engineering, after the previous packet of data is sorted, it is updated to the target library as a known signal. The target library contains frequency, pulse width and repetition rate. target = [cf target , pw target ,prf target ], where cf target is the frequency in the target library, pw target is the pulse width in the target library, prf target is the repetition frequency in the target library; Perform frequency and pulse width matching calculations to dilute and shunt the input pulse description words; when the parameters are matched, as long as the difference is within the tolerance range, the match is considered successful; the pulse description word PDW sequence corresponding to the matching of both carrier frequency CF and pulse width PW parameters is used as the input for the main processing of the known signal; The pulse description word PDW that does not match the radar target library is sent to the input of the unknown signal preprocessing.

3. The fast radar signal sorting method based on simplified multi-parameter clustering as claimed in claim 2, characterized in that: In step S3, the n pulse description words PDW={d1, d2, ..d n } Perform dynamic clustering based on K-Means, including the following steps: Step S31: taking the first data d1 of the pulse width PW and the carrier frequency CF in the pulse description word as the center value C1={PW1, CF1} of the first cluster; Step S32: Read the next data d in the data set in sequence j = {PW j , C.F. j }, and the center value C of the cluster k = {PW k , C.F. k } to compare and determine whether the error satisfies |PW j -PW k |<Δpw and |CF j -CF k |<Δcf; If the error is within the range, the match is successful, the data is merged into the cluster where the cluster center value is located, and the average value of the cluster is used instead of the cluster center value C k ={(PW k +PW j ) / 2, (CF k +CF j ) / 2}; Step S33: After the center value is calculated, read the next data in the data set and continue to execute step S32; if it exceeds the error range, determine whether it has been compared with all cluster center values. If not, continue to compare with the next cluster center value C k-1 Compare until all cluster center values ​​are compared and still not matched, then add a new cluster category as the new cluster center value C k+1 = {PW k+1 , C.F. k+1 }; Step S34: Execute step S33 repeatedly until all data are classified and finally divided into M clusters, where M is the number of radars.

4. The fast radar signal sorting method based on simplified multi-parameter clustering as claimed in claim 3, characterized in that: In step S4, the TOA (time of arrival) sequence corresponding to each clustering result is used as the input of the main sorting, and it is considered as the TOA sequence of a single radar. The specific sorting process includes: Step S41: Calculate the primary TOA difference, count the primary difference histogram within the tolerance range of the difference concept pri, perform sub-harmonic check on the pri that exceeds the threshold, and calculate the detection threshold as shown in formula (1): In formula (1), E is the total number of pulses, C is the level of the difference histogram, and N is the total scale value of the pulse interval on the histogram; τ is the independent variable of the histogram, x and k are constants, k∈[0,1], and the values ​​of x and k are adjusted according to the actual situation of the TOA sequence to improve the sorting effect of the algorithm; Step S42: In actual engineering, the arrival time error is included, and the TOA sequence of a single radar is processed. The harmonic check is changed to: if one of the items in formula (2) is satisfied, it is judged as a harmonic, and the corresponding pri(j) and pri_total(j) are deleted to ensure that only the pri of a single radar is left; The pri_total array represents the number of pri values, and Δpri represents the pri tolerance. It is determined whether the remaining pri, i.e., ppri, varies between several different values. The ppri_toral array represents the number of ppri values ​​updated after harmonic verification. It is visually observed through the simulation diagram of the pulse sequence and the first-level TOA difference, or adaptively determined through the following formula (3). If more than two ppri_total values ​​exceed the threshold, the staggered and jitter judgments are performed, otherwise the conventional and jitter radar judgments are performed. in represents the mean, represents the median, σ ppri_total represents the standard deviation of ppri_total, Δppri_total and Δppri are the set thresholds; in the case of pulse loss and arrival time error, the pri of the staggered radar is similar to the value in ppri_total, and the array value of ppri is more dispersed than that of the jitter radar; if the conditions of formula (3) are met, further determine whether they are met: pri_total(p)-thr(p)>delta diff (4) Where pri_total(p) indicates the pth pri_total value is the largest, thr(p) indicates the threshold calculated by formula (1), and delta diff is the set difference threshold; the pri distribution of the radar with a jitter rate of 15% is pri×(1±15%), and the corresponding values ​​in the pri_total array are relatively small; formula (4) determines the difference between its value and the overall threshold, and diff If the value is less than 100, it is considered as a jitter radar. If one ppri_total exceeds the threshold and the part exceeding the threshold is large, it is judged as a regular radar; if no part exceeds the threshold or some parts exceed the threshold and are not much different from the threshold line, it means that they are scattered in multiple ppris, which reflects the characteristics of a jitter radar.

Images