A radar signal processing method based on TOA sequence correlation degree

Abstract

The application discloses a radar signal processing method based on TOA sequence correlation degree, which comprises the following steps: parameter initialization; obtaining a radar signal pulse description word; dividing the radar signal pulse description word by using a grid method; extracting grids with a pulse number higher than a threshold value, and traversing all the extracted grids to extract the features of all the pulses contained in the grids in the TOA dimension, and the features are described by using TOA sequence feature vectors to obtain the TOA sequence feature vectors corresponding to each grid; quantifying the correlation degree between the TOA sequence feature vectors extracted by different grids to obtain corresponding correlation coefficients; performing grid merging and clustering, and outputting a radar signal clustering result. The method has high accuracy when clustering a multi-parameter agile periodic scanning radar, and is not affected by the pulse interval modulation mode of the radar. In addition, when facing a multi-parameter agile radar, the method can effectively prevent the problem that a traditional clustering algorithm is prone to produce an increased batch result.

Description

Technical Field

[0001] This invention belongs to the field of radar signal sorting technology, specifically relating to a radar signal processing method based on TOA sequence correlation under complex electromagnetic environments. Background Technology

[0002] Radar signal sorting refers to the process of using the differences in signals from different radar radiation sources to classify a large number of mixed radar pulse signals in the environment into several different categories according to the different radiation sources they belong to. The patterns and characteristics of each type of signal are extracted, and information such as the type and location of the corresponding radiation source is extracted. Finally, the operating mode, combat status, and threat level are determined, providing important reference information for analyzing the battlefield situation, formulating combat plans, and implementing battlefield dispatch.

[0003] However, due to the increasingly fierce electronic warfare situation, the current electromagnetic environment contains various radar signals emitted by both friendly and enemy forces. Furthermore, the continuous advancement of radar technology and the application of many new radar modulation methods have made these signals more complex and diverse, resulting in significant overlap in space. All of these factors pose significant challenges to radar signal sorting and identification systems, and the following problems remain unresolved:

[0004] (1) On the one hand, the intra-pulse and inter-pulse modulation methods of radar signals are complex and can be superimposed on each other; on the other hand, there are many radar operating modes. Different operating modes of the same radar can transmit completely different pulse signals, and the switching of operating modes cannot be predicted. The above two situations are very likely to lead to "additional batches", that is, signals belonging to the same radar are misjudged as multiple radars.

[0005] (2) Most radars transmit pulses by scanning in all directions at a certain period. Therefore, the radar signals received by radar reconnaissance receivers are mostly in the form of discontinuous and periodic pulse clusters in the time dimension. Most existing radar signal sorting algorithms, especially sorting algorithms based on TOA level difference, are designed for radars that transmit pulses continuously. These algorithms cannot achieve good results when facing periodically scanning radar signals.

[0006] Multi-parameter agile periodic scanning radar combines the two aforementioned scenarios, making it an object that current radar signal sorting algorithms cannot effectively handle. Such radars typically exhibit one-dimensional or two-dimensional parameter agility; for example, the carrier frequency may continuously vary over a relatively large range, or they may simultaneously possess several frequency points and pulse widths. Furthermore, these radars periodically and directionally transmit several pulses at regular intervals to perform target search. This results in discontinuous radar signals received by the radar reconnaissance receiver. Both this discontinuity and the agility of multiple parameters hinder radar signal sorting algorithms.

[0007] Accurately identifying and extracting these complex, multi-parameter, agile, periodic scanning radar signals, optimizing current radar radiation source signal sorting algorithms, and improving the accuracy of sorting results are pressing problems in the field of radar signal sorting. Summary of the Invention

[0008] Objective: To overcome the shortcomings of existing technologies, this invention provides a radar signal sorting method based on TOA sequence correlation. It leverages the strong temporal regularity of signals emitted by the same radar, determining the likelihood of different pulses belonging to the same radar based on the correlation between TOA sequences, and then clustering pulses that may belong to the same radar. This method is applied to the pre-sorting stage of radar signal sorting, accurately identifying and extracting periodic scanning radar signals that traditional TOA-based sorting algorithms cannot handle, as well as multi-parameter agile radar signals that might result in batching errors under traditional clustering algorithms. It effectively prevents batching errors and provides a novel approach using pulse descriptor parameters.

[0009] Technical Solution: To achieve the above objectives, this invention provides a radar signal processing method based on TOA sequence correlation, comprising the following steps:

[0010] S1: Initialize the parameters;

[0011] S2: Based on the set parameters, obtain the radar signal pulse description word for a continuous time period of a specified length;

[0012] S3: The radar signal pulse descriptor is divided into grids using a grid method, dividing all pulses into different grids, with each pulse participating in the division once and only once;

[0013] S4: Based on the set parameters, extract the grids with a pulse count higher than the threshold, and traverse all extracted grids to extract the features of all pulses contained in the grid in the TOA dimension, and describe them with the TOA sequence feature vector to obtain the TOA sequence feature vector corresponding to each grid.

[0014] S5: Quantify the correlation between the feature vectors of the TOA sequence extracted from different grids to obtain the corresponding correlation coefficients;

[0015] S6: Based on the set parameters and the obtained correlation coefficients, perform grid merging and clustering, and output the radar signal clustering results.

[0016] Furthermore, the parameters initialized in step S1 include the time slot length (time-gap), the step size (time-step), the pulse number threshold (α), and the correlation coefficient threshold (TH). cov .

[0017] Furthermore, in step S2, the specified length is the set time-gap length.

[0018] Furthermore, in step S3, the radar signal pulse descriptor is divided into grids using the CF-PW two-dimensional plane. The specific process is as follows:

[0019] Given a radar signal pulse descriptor word (PDW), the values ​​of its five-dimensional features are all bounded. Divide the CF-dimensional features into n... CF Each segment is divided into n PW-dimensional features. PW If there are n small segments, then the CF-PW two-dimensional plane is divided into n CF ×n PW 1 grid, of which In the formula, 'a' represents the grid length, 'b' represents the grid width, 'ceil' indicates rounding up, and 'max' and 'min' represent the maximum and minimum values ​​of the corresponding parameters in the PDW, respectively. Each pulse is projected onto the CF-PW two-dimensional plane, and these pulses fall within the corresponding grid according to their CF and PW distributions. For pulses that fall exactly on the grid boundary, their grid affiliation is determined according to the "left closed, right open" principle, meaning that each grid covers a range of [e] in its respective dimension. 下边界 ,e 上边界 ).

[0020] Further, in step S4, grids with a pulse count higher than a threshold are extracted, where the threshold is the pulse count threshold α.

[0021] Furthermore, the method for extracting the TOA sequence feature vector corresponding to each grid in step S4 is as follows:

[0022] A1: Select a grid containing pulses with a number greater than the pulse number threshold α. Arrange the TOA of all pulses contained therein on the time axis in ascending order, with the right side representing the time continuation direction. The time axis length is time-gap. The start time is denoted as t0, and the end time is denoted as t. end Then time-gap = t end -t0;

[0023] A2: Create a sliding window with a width (distance between left and right boundaries) of time-step. Place the left boundary of the window at t0, count and record the number of pulses within the window, then slide the window one step (time-step) in the direction of time progression, and count and record the number of pulses within the window again. Repeat the above process until the right boundary of the window reaches t0. end At this point, time-gap / time-step record values ​​should be obtained. These values ​​should be arranged in the order of the records to form a vector, which is the TOA sequence feature vector of the current grid.

[0024] A3: Determine whether there are still grids containing a number of pulses greater than the pulse number threshold α. If they exist, execute steps A1 and A2 for them; otherwise, execute step A4.

[0025] A4: Output the TOA sequence feature vectors extracted from all grids containing more than α pulses, and their number N.

[0026] Furthermore, the correlation coefficient in step S5 is obtained by forming a correlation coefficient matrix (cov-Matrix) from the quantification results of the correlation between all grids.

[0027] Furthermore, the process of obtaining the correlation coefficient matrix (cov-Matrix) in step S5 is as follows:

[0028] Create an N×N identity matrix, and then calculate the eigenvector X of the TOA sequence. i and X j The correlation coefficient between them is calculated and the result is placed in the i-th row and j-th column of the identity matrix; the calculation formula is as follows: Among them, Cov(X) i ,X j )=E[(X i -EX i (X) j -EX j )], i,j∈[1,N] and i≠j, E represents the expectation, and Var represents the variance;

[0029] After calculating the correlation coefficients between the eigenvectors of all TOA sequences and filling them into the identity matrix, this matrix is ​​the correlation coefficient matrix (cov-Matrix).

[0030] Further, step S6 specifically involves: traversing the correlation coefficient matrix cov-Matrix and selecting those with correlation coefficients higher than the correlation coefficient threshold TH. cov Merge the two grids associated with the correlation coefficient below the threshold TH. cov The two grids associated with the correlation coefficient are not merged. When all correlation coefficients have been examined, some grids will be merged into several grid groups. The clustering result is denoted as C, and the radar signal clustering result C is output.

[0031] This invention directly uses the Time of Arrival (TOA) parameter from the pulse descriptor, employing TOA sequence correlation as the basis for clustering. First, a grid method is used to divide radar pulses into different grids. Then, TOA sequence feature vectors are used to describe the time-dimensional characteristics of the pulses contained in each grid. Finally, by comparing the correlation between the TOA sequence feature vectors of different grids, pulses contained in grids with high correlation are clustered as belonging to the same radar. This method exhibits high accuracy in clustering multi-parameter agile periodic scanning radars and is unaffected by the radar's inter-pulse modulation method. Furthermore, when dealing with multi-parameter agile radars, it effectively prevents the problem of batching errors that traditional clustering algorithms tend to produce.

[0032] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0033] 1. For periodic scanning radar, the radar pulses received by the radar reconnaissance receiver are discontinuous and periodic pulse clusters in the time dimension. Traditional radar signal sorting algorithms that rely on TOA level differences cannot effectively process this type of radar signal. This invention has excellent clustering effect for this type of radar and has high accuracy.

[0034] 2. This invention directly uses the time dimension parameters of radar pulse data in the overall process, and uses the correlation between TOA sequences as the basis for clustering. Therefore, it does not require that the pulse data be spatially distributed in the CF-PW two-dimensional plane. For multi-parameter agile radar, its pulses do not present a "cluster" shape in the CF-PW two-dimensional plane, but still have relatively accurate clustering results, which can effectively prevent the occurrence of additional batches. Attached Figure Description

[0035] Figure 1 This is a flowchart illustrating the radar signal sorting method based on TOA sequence correlation of the present invention.

[0036] Figure 2 This is a schematic diagram illustrating the extraction of the TOA sequence feature vector for each grid in the radar signal sorting method based on TOA sequence correlation according to the present invention.

[0037] Figure 3 A schematic diagram showing the distribution of the generated radar pulse simulation data in the CF-PW-TOA three-dimensional space;

[0038] Figure 4 This is a schematic diagram showing the distribution of radar clustering results obtained after processing by the method proposed in this invention in the CF-PW-TOA three-dimensional space. Detailed Implementation

[0039] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0040] like Figure 1 As shown, this invention provides a radar signal processing method based on TOA sequence correlation, comprising the following steps:

[0041] S1: Initialization parameters, including time slot length (time-gap), step size (time-step), pulse number threshold (α), and correlation coefficient threshold (TH). cov .

[0042] S2: Obtain the radar signal pulse description word within a continuous time period of length time-gap.

[0043] S3: The radar signal pulse descriptor is divided into grids using a grid method, dividing all pulses into different grids, with each pulse participating in the division once and only once;

[0044] The radar signal pulse descriptor is divided into grids using the CF-PW two-dimensional plane. The specific process is as follows:

[0045] Given a radar signal pulse descriptor word (PDW), the values ​​of its five-dimensional features are all bounded. Divide the CF-dimensional features into n... CF Each segment is divided into n PW-dimensional features. PW If there are n small segments, then the CF-PW two-dimensional plane is divided into n CF ×n PW 1 grid, of which In the formula, 'a' represents the grid length, 'b' represents the grid width, 'ceil' indicates rounding up, and 'max' and 'min' represent the maximum and minimum values ​​of the corresponding parameters in the PDW, respectively. Each pulse is projected onto the CF-PW two-dimensional plane, and these pulses fall within the corresponding grid according to their CF and PW distributions. For pulses that fall exactly on the grid boundary, their grid affiliation is determined according to the "left closed, right open" principle, meaning that each grid covers a range of [e] in its respective dimension. 下边界 ,e 上边界 ).

[0046] S4: Extract grids with a pulse count higher than the pulse count threshold α, and traverse all extracted grids to extract the features of all pulses contained in the grid in the TOA dimension, and describe them with TOA sequence feature vectors to obtain the TOA sequence feature vector corresponding to each grid.

[0047] The method for extracting the TOA sequence feature vector corresponding to each grid is as follows:

[0048] A1: Select a grid containing pulses with a number greater than the pulse number threshold α. Arrange the TOA of all pulses contained therein on the time axis in ascending order, with the right side representing the time continuation direction. The time axis length is time-gap. The start time is denoted as t0, and the end time is denoted as t. end Then time-gap = t end -t0;

[0049] A2: Create a sliding window with a width (distance between left and right boundaries) of time-step. Place the left boundary of the window at t0, count and record the number of pulses within the window, then slide the window one step (time-step) in the direction of time progression, and count and record the number of pulses within the window again. Repeat the above process until the right boundary of the window reaches t0. end At this point, time-gap / time-step record values ​​should be obtained. These values ​​should be arranged in the order of the records to form a vector, which is the TOA sequence feature vector of the current grid.

[0050] A3: Determine whether there are still grids containing a number of pulses greater than the pulse number threshold α. If they exist, execute steps A1 and A2 for them; otherwise, execute step A4.

[0051] A4: Output the TOA sequence feature vectors extracted from all grids containing more than α pulses, and their number N.

[0052] S5: Quantify the correlation between the feature vectors of the TOA sequence extracted from different grids, and form a correlation coefficient matrix (cov-Matrix) by quantifying the correlation between all grids.

[0053] The process of obtaining the correlation coefficient matrix (cov-Matrix) is as follows:

[0054] Create an N×N identity matrix, and then calculate the eigenvector X of the TOA sequence. i and X j The correlation coefficient between them is calculated and the result is placed in the i-th row and j-th column of the identity matrix; the calculation formula is as follows: Among them, Cov(X) i ,X j )=E[(X i -EX i (X) j -EX j )], i,j∈[1,N] and i≠j, E represents the expectation, and Var represents the variance;

[0055] After calculating the correlation coefficients between the eigenvectors of all TOA sequences and filling them into the identity matrix, this matrix is ​​the correlation coefficient matrix (cov-Matrix).

[0056] S6: Traverse the correlation coefficient matrix (cov-Matrix) and select those with correlation coefficients higher than the threshold TH. cov Merge the two grids associated with the correlation coefficient below the threshold TH. cov The two grids associated with the correlation coefficient are not merged. When all correlation coefficients have been examined, some grids will be merged into several grid groups. The clustering result is denoted as C, and the radar signal clustering result C is output.

[0057] Based on the above, this embodiment applies the above solution as follows:

[0058] Step 1: Parameter Initialization. Since this invention performs mesh generation in the CF-PW two-dimensional plane, the mesh length a = 6MHz, mesh width b = 2µs; time slot length time-gap = 40s, step size time-step = 4ms, pulse number threshold α = 6, and correlation coefficient threshold TH... cov =0.45.

[0059] Step 2: Obtain the radar signal pulse descriptor (PDW). This example uses MATLAB to simulate and generate simulation data for 9 radars. The pulses from 8 radars are received with a certain scanning period, while the pulses from 1 radar are received continuously. The parameter settings are shown in the table below:

[0060]

[0061] The distribution diagram of these nine radars in the CF-PW-TOA three-dimensional space is as follows: Figure 3 As shown.

[0062] The pulses of the pulse stream data with a duration of time-gap = 40s were extracted and divided into different grids using a grid method. Each pulse participated in the division once and only once.

[0063] More specifically, the grid method is:

[0064] Given a pulse descriptor word (PDW), the parameters of each dimension of the pulse descriptor word are bounded. Divide the CF-dimensional features into n... CF Each segment is divided into n PW-dimensional features. PW The number of segments obtained by dividing the data into CF dimension and PW dimension, according to the formula, are as follows:

[0065]

[0066]

[0067] Therefore, the CF-PW two-dimensional plane is divided into 282×98 grids. Each pulse is projected onto the CF-PW two-dimensional plane, and these pulses fall within the corresponding grids based on their CF and PW distributions. For pulses that fall exactly on the grid boundaries, their grid affiliation is determined according to the "left-closed, right-open" principle, meaning each grid covers a range of [e] in its respective dimension. 下边界 ,e 上边界 );

[0068] For example, the range covered by the grid in the first row and first column is [2150, 2156) MHz, [5, 7) μs, the range covered by the grid in the first row and second column is [2150, 2156) MHz, [7, 9) μs, ..., and the range covered by the grid in the nth row and mth column is [2150 + 6 × (n-1), 2150 + 6 × n) MHz, [5 + 2 × (m-1), 5 + 2 × m) μs.

[0069] Step 3: Extract the features of all pulses contained in the grid in the TOA dimension and describe them with a vector. In this embodiment, it is named TOA sequence feature vector (TOASequence eigenvector, TOAS-eV). Traverse each grid containing more than α pulses so that each grid obtains its own TOA sequence feature vector.

[0070] like Figure 2 As shown, the method for extracting the TOA sequence feature vector for each grid in step 3 is as follows:

[0071] Step 3-1: Select a grid containing pulses with a number greater than the pulse number threshold α = 6. Arrange the TOA of all its pulses in ascending order on the time axis, with the right side of the time axis representing the time continuation direction and a length of time - gap = 40s. Record the start time as t0 and the end time as t end Then time-gap = t end -t0;

[0072] Step 3-2: Create a sliding window with a width (distance between left and right boundaries) of time-step = 4ms. Place the left boundary of the sliding window at t0, count and record the number of pulses within the window, such as... Figure 2As shown, the first sliding window contains 8 pulses, so the first recorded value is 8. Then, the sliding window is slid forward by one step (time-step = 4ms), and the number of pulses in the sliding window is counted again. The second sliding window contains 6 pulses, so the second recorded value is 6, and so on. The third sliding window contains no pulses, so the third recorded value is 0, and so on. This process is repeated until the right boundary of the sliding window reaches t. end At this point, the time-gap / time-step ratio should be 40 × 10. 3 / 4 = 10000 record values. These values ​​are arranged in chronological order to form a vector, which is the TOA sequence feature vector (TOAS-eV) of the current grid. Figure 2 As shown in the example, the TOAS-eV = (8,6,0,12,…,0,2,0,6,9) has a length of 10000;

[0073] Step 3-3: For grids that have not yet participated in the TOAS-eV calculation, check if there are still grids containing more than α pulses. If they exist, perform steps 3-1 and 3-2 for them. If they do not exist, perform step 3-4.

[0074] Steps 3-4: Output all acquired TOAS-eVs and their number N. In the first 40 seconds of the simulation in this example, 62 valid grids can be acquired, that is, a total of 62 TOAS-eVs are obtained, N=62;

[0075] Step 4: Calculate the correlation coefficients between the feature vectors of the TOA sequences extracted from different grids, and form a correlation coefficient matrix (cov-Matrix) based on the correlation coefficients between all grids;

[0076] More specifically, the detailed steps of step 4 in this example are as follows:

[0077] Create a 62×62 identity matrix, and then calculate the eigenvector X of the TOA sequence. i and X j The correlation coefficient between them is calculated, and the result is placed in the i-th row and j-th column of the identity matrix. The calculation formula is as follows: Where Cov(X) i ,X j )=E[(X i -EX i (X) j -EX j )], i,j∈[1,N] and i≠j. After calculating the correlation coefficients between the eigenvectors of all TOA sequences and filling them into the identity matrix, this matrix is ​​the correlation coefficient matrix (cov-Matrix).

[0078] Step 5: Traverse the correlation coefficient matrix (cov-Matrix) and select those with correlation coefficients higher than the threshold TH. cov Two grids with a correlation coefficient of 0.45 are merged; values ​​below the threshold TH are also considered valid. cov The two grids associated with the correlation coefficient are not merged. Since the value of each element in the correlation coefficient matrix represents the correlation coefficient between the TOAS-eV corresponding to its row and column numbers, that is, the value of the element in the i-th row and j-th column is the correlation coefficient between the i-th TOAS-eV and the j-th TOAS-eV. The closer this value is to 1, the more correlated the i-th TOAS-eV and the j-th TOAS-eV are. In other words, the more similar the pulses contained in the grids to which these two vectors belong are in the time dimension, the greater the probability that they belong to the same radar.

[0079] In this example, after all 62 correlation coefficients have been examined, the grids in the 62 grids associated with them that contain pulses that are highly correlated in the time dimension will be clustered together. Since this data contains more than one radar, several classes will be obtained in the end. The clustering result is denoted as C.

[0080] Step 7: Output the final radar signal clustering result C. In this example, the radar signal sorting method based on TOA sequence correlation is used to process the results, and the distribution diagram of the CF-PW-TOA three-dimensional space is shown below. Figure 4 As shown in the image, a point of one color represents the clustering result of a radar. Comparing the clustering results with the parameters set in the table shows that the clustering results are completely correct.

[0081] For periodically scanning radars, pulses belonging to the same radar exhibit high temporal consistency. This invention directly utilizes the Time of Arrival (TOA) parameter in the pulse descriptor to extract and cluster pulses with high correlation between TOA sequences, grouping pulses that may belong to the same radar into one category, thereby achieving radar pre-sorting.

[0082] In summary, the method of this invention directly uses the Time of Arrival (TOA) parameter in the pulse descriptor to examine the correlation of different pulses in the time dimension, and uses this as the basis for clustering, treating pulses with high TOA sequence correlation as a single radar and clustering them into one class. For multi-parameter agile periodic scanning radars, traditional clustering algorithms often lead to batching errors or even fail to obtain correct clustering results. Furthermore, radar signal sorting algorithms based on TOA level differences may even fail when dealing with such periodic scanning radars due to discontinuous level differences. This invention provides accurate clustering results and strong resistance to batching errors for this type of multi-parameter agile periodic scanning radar.

Claims

1. A radar signal processing method based on TOA sequence correlation, characterized in that, Includes the following steps: S1: Initialize the parameters; S2: Based on the set parameters, obtain the radar signal pulse description word for a continuous time period of a specified length; S3: The radar signal pulse descriptor is divided into grids using a grid method, dividing all pulses into different grids, with each pulse participating in the division once and only once; S4: Extract grids with more than a threshold number of pulses, and iterate through all extracted grids. Extract the features of all pulses contained in the grid in the TOA dimension, and describe them with TOA sequence feature vectors to obtain the TOA sequence feature vector corresponding to each grid. S5: Quantify the correlation between the feature vectors of the TOA sequence extracted from different grids to obtain the corresponding correlation coefficients; S6: Based on the obtained correlation coefficients, perform grid merging and clustering, and output the radar signal clustering results; The parameters set during initialization in step S1 include the time slot length. Step length Pulse quantity threshold Correlation coefficient threshold ; In step S3, the radar signal pulse descriptor is divided into grids using the CF-PW two-dimensional plane. The specific process is as follows: Given a radar signal pulse descriptor word (PDW), the values ​​of its five-dimensional features are all bounded. The CF-dimensional features are then divided into... A small segment, PW-dimensional features are divided into The CF-PW two-dimensional plane is divided into small segments. 1 grid, of which , , in the formula For grid length, Let be the grid width, ceil represent rounding up, and max and min represent the maximum and minimum values ​​of the corresponding parameters in the PDW, respectively. Each pulse is projected onto the CF-PW two-dimensional plane, and these pulses fall within the corresponding grid according to their CF and PW distributions. For pulses that fall exactly on the grid boundary, their grid affiliation is determined; that is, the range covered by each grid in its respective dimension is... ; Step S4 extracts the grids with a pulse count higher than a threshold, where the threshold is the pulse count limit. ; The method for extracting the TOA sequence feature vector for each grid in step S4 is as follows: A1: Select an array containing pulse counts greater than the pulse count threshold. The grid contains pulses whose TOA (Time of Arrival) is arranged in ascending order on the time axis, with the right side representing the time continuation direction, and the time slot length is [missing information]. The start time is recorded as The termination time is recorded as ,but ; A2: Create a sliding window with a width of [value missing]. Place the left edge of the sliding window At this point, count and record the number of pulses within the sliding window, and then slide the sliding window one step in the direction of time progression. Count and record the number of pulses within the sliding window again, and repeat the above process until the right boundary of the sliding window is reached. At this time, it should be obtained Each record value is arranged in chronological order to form a vector, which is the TOA sequence feature vector of the current grid. A3: Determine if there are still instances where the number of pulses exceeds the pulse count threshold. If the grid exists, proceed with steps A1 and A2; otherwise, proceed with step A4. A4: Output all contained in TOA sequence feature vectors extracted from grids with more than one pulse and the number of feature vectors N; The correlation coefficients in step S5 are obtained by forming a correlation coefficient matrix from the quantified correlation results between all grids. ; Step S6 specifically involves: traversing the correlation coefficient matrix. It will be higher than the correlation coefficient threshold. Merge the two grids associated with the correlation coefficient, if the correlation coefficient is below the threshold. Two grids associated with a correlation coefficient are not merged. Once all correlation coefficients have been examined, some grids will be merged into several grid clusters. The clustering result is denoted as... Output radar signal clustering results .

2. The radar signal processing method based on TOA sequence correlation according to claim 1, characterized in that, In step S2, the specified length is the set time slot length. .

3. The radar signal processing method based on TOA sequence correlation according to claim 1, characterized in that, The correlation coefficient matrix in step S5 The acquisition process is as follows: Create The identity matrix is ​​then used to calculate the eigenvectors of the TOA sequence. and The correlation coefficient between them is calculated and the result is placed in the i-th row and j-th column of the identity matrix; the calculation formula is as follows: ,in, , , and E represents the expected value, and Var represents the variance. After calculating the correlation coefficients between the eigenvectors of all TOA sequences and filling them into the identity matrix, this matrix becomes the correlation coefficient matrix. .

Images