Radar signal sorting method and device, electronic equipment and storage medium

Abstract

The application provides a radar signal sorting method and device, electronic equipment and storage medium, and relates to the technical field of radar signal processing. The method comprises the following steps: sorting a to-be-sorted pulse sequence to obtain a plurality of clusters, and obtaining a sorting number of each pulse descriptor word in the to-be-sorted pulse sequence; determining that the to-be-sorted pulse sequence is a pulse sequence other than a pulse sequence of a reference beat in a plurality of beat pulse sequences, inputting the to-be-sorted pulse sequence into a tracking model, and obtaining a tracking number of each pulse descriptor word in the to-be-sorted pulse sequence output by the tracking model; and correcting the sorting number of each pulse descriptor word in the to-be-sorted pulse sequence based on the tracking number of each pulse descriptor word in the to-be-sorted pulse sequence. The tracking model is trained based on a pulse sequence of a reference beat and a corresponding sorting number label. The application can improve the sorting accuracy of radar signals and improve the sorting efficiency of radar signals.

Description

Technical Field

[0001] This invention relates to the field of radar signal processing technology, and in particular to a radar signal sorting method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the rapid development of technology, applications of radar signal-based identification are becoming increasingly widespread, such as target identification, target tracking, and target localization. However, radar receivers mostly receive multi-target radar pulse data, which is often mixed with other radar signals. To better identify targets based on radar signals, signal sorting is necessary.

[0003] Currently, radar signal sorting algorithms based on PRI (Pulse Repeat Interval) histograms are used for sorting. Specifically, for received pulse data, a pulse arrival time sequence within a certain range is extracted, and the TOA (Time of Arrival) difference between any two pulses is calculated. This difference is used to estimate the possible PRI patterns in the pulse data through statistical analysis, thereby enabling sequence retrieval of the PRI. However, radar signal sorting algorithms based on PRI histograms require calculating a large number of differences, leading to reduced radar signal sorting efficiency. Furthermore, the sorting accuracy of these algorithms is low when dealing with complex radar signals. Summary of the Invention

[0004] This invention provides a radar signal sorting method, apparatus, electronic device, and storage medium to address the shortcomings of low radar signal sorting efficiency and low sorting accuracy in the prior art.

[0005] This invention provides a radar signal sorting method, comprising:

[0006] The pulse sequence to be sorted is sorted to obtain multiple clusters, and the sorting number of each pulse descriptor in the pulse sequence to be sorted is obtained, wherein the sorting number of each pulse descriptor in any cluster is the same;

[0007] The pulse sequence to be sorted is determined to be a pulse sequence other than the pulse sequence of the reference pulse in a pulse sequence of multiple beats. The pulse sequence to be sorted is input into the tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model. The pulse sequence of multiple beats is obtained by dividing the pulse data based on the aliasing of multi-target radar.

[0008] Based on the tracking number of each pulse descriptor in the pulse sequence to be sorted, the sorting number of each pulse descriptor in the pulse sequence to be sorted is corrected;

[0009] The tracking model is trained based on a sample pulse sequence and the sorting number label corresponding to the sample pulse sequence. The sample pulse sequence is the pulse sequence of the reference beat, and the sorting number label is the sorting number of each pulse descriptor in the pulse sequence of the reference beat.

[0010] According to a radar signal sorting method provided by the present invention, the step of correcting the sorting number of each pulse descriptor in the pulse sequence to be sorted based on the tracking number of each pulse descriptor in the pulse sequence to be sorted includes:

[0011] Based on the tracking number of each pulse descriptor in each cluster, the number of target tracking numbers in each cluster is determined, and any target tracking number is the tracking number with the largest number in the cluster.

[0012] Based on the number of each number and the number of pulse descriptors of each cluster, a tracking threshold for each cluster is determined. The tracking threshold for any cluster is determined based on the ratio of the number of numbers of the cluster to the number of pulse descriptors of the cluster.

[0013] A first target cluster whose tracking threshold is greater than a preset tracking threshold is identified, and the sorting number of each pulse descriptor in the first target cluster is corrected to the target tracking number of the first target cluster.

[0014] According to a radar signal sorting method provided by the present invention, the step of determining the tracking threshold of each of the said clusters based on the number of each number and the number of pulse descriptors of each said cluster, further includes:

[0015] A second target cluster is determined where the tracking threshold is less than or equal to a preset tracking threshold, and the sorting number of each pulse descriptor in the second target cluster is incremented by a preset value.

[0016] According to a radar signal sorting method provided by the present invention, the sorting of pulse sequences to be sorted to obtain multiple clusters, and obtaining sorting numbers for each pulse descriptor in the pulse sequences to be sorted, includes:

[0017] Blind clustering is performed on the pulse sequences to be sorted to obtain multiple clusters.

[0018] The multiple clusters are numbered respectively to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted.

[0019] A radar signal sorting method according to the present invention further includes:

[0020] The pulse sequence to be sorted is input into the model identification model to obtain the model identification result of each pulse description word in the pulse sequence to be sorted output by the model identification model;

[0021] The model identification model is trained based on sample pulse data and the corresponding model identification result labels.

[0022] According to a radar signal sorting method provided by the present invention, the step of inputting the pulse sequence to be sorted into a model to obtain the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model further includes:

[0023] Determine the confidence level corresponding to each model identification result;

[0024] If the target model identification result is determined to have a confidence level less than or equal to a preset confidence threshold, the target model identification result is updated to the out-of-collection data identification result.

[0025] The sample pulse data includes pulse sequences of N types of radars, the type identification model is an N-classification model, and the out-of-set data identification results indicate that the radar type does not belong to the N types.

[0026] According to a radar signal sorting method provided by the present invention, the step of inputting the pulse sequence to be sorted into a model to obtain the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model further includes:

[0027] Based on the corrected sorting number of each pulse descriptor in the pulse sequence to be sorted, the pulse sequence to be sorted is re-sorted to obtain multiple re-sorting clusters;

[0028] Based on the model identification results of each pulse descriptor in each of the re-sorting clusters, the most numerous model identification results in each of the re-sorting clusters are determined respectively;

[0029] The model identification results of each pulse descriptor in each of the re-sorting clusters are corrected to the maximum model identification results.

[0030] The present invention also provides a radar signal sorting device, comprising:

[0031] The sorting module is used to sort the pulse sequence to be sorted to obtain multiple clusters, and to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted, wherein the sorting numbers of each pulse descriptor in any cluster are the same.

[0032] The tracking module is used to determine that the pulse sequence to be sorted is a pulse sequence other than the pulse sequence of the reference pulse in a pulse sequence of multiple beats, and inputs the pulse sequence to be sorted into the tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model. The pulse sequence of multiple beats is obtained based on the pulse data division of multi-target radar aliasing.

[0033] The correction module is used to correct the sorting number of each pulse descriptor in the pulse sequence to be sorted based on the tracking number of each pulse descriptor in the pulse sequence to be sorted.

[0034] The tracking model is trained based on a sample pulse sequence and the sorting number label corresponding to the sample pulse sequence. The sample pulse sequence is the pulse sequence of the reference beat, and the sorting number label is the sorting number of each pulse descriptor in the pulse sequence of the reference beat.

[0035] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the radar signal sorting method as described above.

[0036] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the radar signal sorting method as described above.

[0037] The radar signal sorting method, apparatus, electronic device, and storage medium provided by this invention divide the pulse data of multi-target radar aliasing into multiple pulse sequences. Based on the pulse sequence of the reference pulse sequence and the sorting number of each pulse descriptor in the reference pulse sequence, a tracking model is trained. The pulse sequences to be sorted in the multiple pulse sequences, excluding the pulse sequence of the reference pulse, are input into the tracking model. The tracking model outputs the tracking number of each pulse descriptor in the pulse sequences to be sorted. Based on the tracking number of each pulse descriptor in the pulse sequences to be sorted, the sorting number of each pulse descriptor obtained from sorting the pulse sequences is corrected, thereby improving the sorting accuracy of radar signals. Furthermore, the post-processing correction of the sorted numbers has a small computational load, thereby improving the sorting efficiency of radar signals. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0039] Figure 1 This is one of the flowcharts illustrating the radar signal sorting method provided by the present invention;

[0040] Figure 2 This is the second flowchart illustrating the radar signal sorting method provided by the present invention.

[0041] Figure 3 The third schematic diagram of the radar signal sorting method provided by the present invention;

[0042] Figure 4 This is a schematic diagram of the radar signal sorting device provided by the present invention;

[0043] Figure 5 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0045] With the rapid development of technology, applications of radar signal-based identification are becoming increasingly widespread, such as target identification, target tracking, and target localization. However, due to the rapid advancement of radar technology, the operating frequency coverage of radars is wider, signal parameters are more agile, and the reconnaissance coverage is broader. This leads to aliasing of received multi-source signals, meaning that the radar receiver receives mostly pulse data from multiple target radars. To better identify radar signals, it is necessary to sort the radar signals.

[0046] Traditional sorting methods based on PDW (Pulse Describe Words) data such as signal carrier frequency, pulse width, angle of arrival, and pulse repetition period are difficult to effectively sort radar radiation source signals with multiple sources overlapping.

[0047] Currently, radar signal sorting is performed using a radar signal sorting algorithm based on the PRI (Pulse Repeat Interval) histogram. Specifically, for received pulse data, a pulse arrival time sequence within a certain range is extracted, and the TOA (Time of Arrival) difference between any two pulses is calculated. This difference is used to estimate the possible PRI patterns of the pulse data through statistical analysis, thereby performing sequence retrieval for the PRI. More specifically, firstly, pulses received within a certain period are extracted, the TOA difference between any two pulses is calculated, the difference is statistically analyzed, and a histogram is plotted. Secondly, the difference with the most frequent occurrences is identified as the possible PRI value, which is used as a parameter of the radar signal for sequence retrieval. If several PRI values ​​are close, the smallest PRI value is selected for sequence retrieval. If the sequence retrieval is successful, this PRI value is recorded, and the PRI variation pattern is re-estimated based on the retrieved sequence. If the retrieval fails, other larger PRI values ​​are used to continue the retrieval. Finally, the successfully retrieved sequences are removed, and the above steps are repeated for the remaining pulses until sorting is complete. However, radar signal sorting algorithms based on PRI histograms require calculating a large number of differences, leading to reduced sorting efficiency. Furthermore, the sorting accuracy is low when dealing with complex radar signals. Even when improving the algorithm using pulse sequence correlation—that is, calculating only the TOA difference between adjacent pulses for each retrieval to accumulate the difference and reduce computation—the computational load remains high. Moreover, the sorting performance deteriorates under complex radar PRI modulation patterns (e.g., jitter PRI, slip PRI, etc.) and high noise interference conditions, thus reducing the sorting accuracy.

[0048] To address the above problems, the present invention proposes the following embodiments. Figure 1 This is one of the flowcharts illustrating the radar signal sorting method provided by the present invention, such as... Figure 1 As shown, the radar signal sorting method includes:

[0049] Step 110: Sorting the pulse sequence to be sorted to obtain multiple clusters, and obtaining the sorting number of each pulse descriptor in the pulse sequence to be sorted, wherein the sorting numbers of each pulse descriptor in any cluster are the same.

[0050] Here, the pulse sequence to be sorted is a pulse sequence of multi-target radar aliasing, and the pulse sequence to be sorted is a radar pulse sequence to be sorted. The pulse sequence to be sorted includes multiple pulse descriptors (PDWs).

[0051] In some embodiments, prior to step 110 above, the pulse data of the multi-target radar aliasing is divided into multiple pulse sequences, and a pulse sequence to be sorted is determined from these multiple pulse sequences. The multi-target radar aliasing pulse data can be data received by a radar receiver. It is understood that the pulse sequence to be sorted is determined sequentially from the multiple pulse sequences to repeat the same steps.

[0052] In one embodiment, each beat lasts for 200 milliseconds, meaning the pulse sequence to be sorted is a single-beat pulse sequence with a duration of 200 milliseconds. It is understood that the number of pulse waves (PDWs) contained in this single-beat pulse sequence varies depending on the pulse density; for example, this single-beat pulse sequence may contain 20,000 PDWs.

[0053] Here, any cluster includes at least one pulse descriptor, and pulse descriptors with the same sorting number are clustered together.

[0054] Specifically, the pulse descriptors in the pulse sequence to be sorted are clustered to obtain multiple clusters, and each cluster is numbered to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted.

[0055] Step 120: Determine that the pulse sequence to be sorted is a pulse sequence other than the pulse sequence of the reference pulse in a pulse sequence of multiple beats. Input the pulse sequence to be sorted into the tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model. The pulse sequence of multiple beats is obtained by dividing the pulse data of multi-target radar aliasing.

[0056] The tracking model is trained based on a sample pulse sequence and the sorting number label corresponding to the sample pulse sequence. The sample pulse sequence is the pulse sequence of the reference beat, and the sorting number label is the sorting number of each pulse descriptor in the pulse sequence of the reference beat.

[0057] Here, the pulse sequence of the reference beat is the pulse sequence that other beats need to track; that is, the pulse sequence of the reference beat is used to train the tracking model.

[0058] In some embodiments, prior to step 120 above, the method further includes:

[0059] The pulse sequence of the reference beat is sorted to obtain the sorting number of each pulse descriptor in the pulse sequence of the reference beat; based on the pulse sequence of the reference beat and the sorting number of each pulse descriptor in the pulse sequence of the reference beat, the tracking model is trained. More specifically, the pulse sequence to be sorted is determined as the pulse sequence of the reference beat, and the pulse sequence of the reference beat is sorted.

[0060] In one embodiment, the pulse sequence of the reference beat is the pulse sequence of the first beat, so that the other beats in the pulse sequences of multiple beats, excluding the first beat, track the first beat. Based on this, the pulse sequence of the first beat does not need to be used for signal tracking; that is, the pulse sequence of the first beat does not need to execute steps 120 and 130. The pulse sequence of the first beat is only needed for training the tracking model.

[0061] In one embodiment, the pulse sequence of the reference beat is sorted to obtain M clusters, and the corresponding trained tracking model is an M-classifier. Considering that the pulse sequence includes PDW (Pulse-Driven Wave), after comparing various neural networks and traditional machine learning classification algorithms, the tracking model can use an SVM (Support Vector Machine) classifier to improve tracking accuracy, i.e., improve the accuracy of tracking numbering.

[0062] Specifically, the pulse sequence to be sorted is input into the trained tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted, output by the trained tracking model. More specifically, the sorted pulse sequence to be sorted is input into the trained tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted, output by the trained tracking model.

[0063] Step 130: Based on the tracking number of each pulse descriptor in the pulse sequence to be sorted, correct the sorting number of each pulse descriptor in the pulse sequence to be sorted.

[0064] Considering that in the radar signal sorting process, after the pulse sequences of each beat are sorted independently, the sorting results between beats are not correlated. Therefore, a tracking model is trained using the pulse sequence of a reference beat, and the pulse sequences of other beats are input into the tracking model to obtain the tracking number of each pulse descriptor. Then, based on the tracking number of each pulse descriptor, the sorting number of each pulse descriptor is corrected. This ensures that for the pulse descriptor of the same radar target, its sorting number should remain consistent in the sorting results of multiple beats. That is, the radar signal is stably tracked between beats, ensuring that the sorting results between beats are correlated. Ultimately, this improves the stability and accuracy of radar signal sorting.

[0065] In one embodiment, based on the tracking number of each pulse descriptor in each cluster, the number of target tracking numbers in each cluster is determined, with any target tracking number being the most numerous tracking number in the cluster. Based on the number of each tracking number and the number of pulse descriptors in each cluster, a tracking threshold is determined for each cluster, where the tracking threshold is determined based on the ratio of the number of tracking numbers to the number of pulse descriptors in the cluster. A first target cluster with a tracking threshold greater than a preset tracking threshold is determined, and the sorting numbers of each pulse descriptor in the first target cluster are corrected to the target tracking number. The specific execution process of this embodiment is described in the following embodiment.

[0066] In another embodiment, the sorting number of each pulse descriptor in the pulse sequence to be sorted is corrected to the corresponding tracking number. That is, the tracking number is used as the final sorting number.

[0067] It should be noted that if the sorting number of the pulse descriptor is corrected, the pulse descriptor shall be based on the corrected sorting number; if the sorting number of the pulse descriptor is not corrected, the pulse descriptor shall be based on the original sorting number. It is understandable that after the sorting number is corrected, the division of multiple clusters shall be based on the corrected sorting number.

[0068] It is understandable that post-processing correction of the sorted numbers obtained from sorting can effectively solve the sorting problem of complex radar signals, such as radar signal sorting under high noise and high interference conditions, as well as radar signal sorting under complex radar PRI modulation patterns, thereby improving the stability and effectiveness of radar signal sorting.

[0069] The radar signal sorting method provided in this invention divides the pulse data of multi-target radar aliasing into multiple pulse sequences. Based on the pulse sequence of the reference pulse sequence and the sorting number of each pulse descriptor in the reference pulse sequence, a tracking model is trained. The pulse sequences to be sorted in the multiple pulse sequences, excluding the pulse sequence of the reference pulse, are input into the tracking model. The tracking model outputs the tracking number of each pulse descriptor in the pulse sequences to be sorted. Based on the tracking number of each pulse descriptor in the pulse sequences to be sorted, the sorting number of each pulse descriptor obtained from sorting the pulse sequences is corrected, thereby improving the sorting accuracy of radar signals. Furthermore, the post-processing correction of the sorted numbers has a small computational load, thus improving the sorting efficiency of radar signals.

[0070] Based on the above embodiments, Figure 2 This is a second flowchart illustrating the radar signal sorting method provided by the present invention, as shown below. Figure 2 As shown, step 130 above includes:

[0071] Step 131: Based on the tracking number of each pulse descriptor in each cluster, determine the number of target tracking numbers in each cluster, wherein any target tracking number is the tracking number with the largest number in the cluster.

[0072] For any given cluster, count the tracking numbers of each pulse descriptor in that cluster, and determine the number of tracking numbers that are most numerous in that cluster.

[0073] Step 132: Based on the number of each number and the number of pulse descriptors of each cluster, determine the tracking threshold of each cluster. The tracking threshold of any cluster is determined based on the ratio of the number of numbers of the cluster to the number of pulse descriptors of the cluster.

[0074] For any cluster, the tracking threshold for that cluster is determined based on the ratio of the number of target tracking IDs to the number of pulse descriptors in that cluster. This ratio can be used directly as the tracking threshold, or it can be obtained through further data processing. The number of pulse descriptors is the total number of pulse descriptors in the cluster.

[0075] Step 133: Determine the first target cluster whose tracking threshold is greater than the preset tracking threshold, and correct the sorting number of each pulse descriptor in the first target cluster to the target tracking number of the first target cluster.

[0076] Here, the preset tracking threshold is a pre-set threshold, preferably 0.9.

[0077] Here, the number of first target clusters can be one or more, and the first target cluster is a cluster among multiple clusters. That is, multiple clusters may include one or more first target clusters. Of course, multiple clusters may not include the first target clusters.

[0078] For any first target cluster, the sorting number of each pulse descriptor in the first target cluster is corrected to the target tracking number with the largest number in the first target cluster.

[0079] For example, a certain cluster has a sorting number of 2 and a pulse descriptor number of 500. The tracking number of the pulse descriptor can include 1-7. The tracking number of 480 pulse descriptors in this cluster is 5, that is, the target tracking number of this cluster is 5. Based on this, the tracking threshold of this cluster is 0.96. If the preset tracking threshold is 0.9, then the tracking threshold of 0.96 is greater than the preset tracking threshold of 0.9. Therefore, the sorting number 2 of each pulse descriptor in this cluster is corrected to the target tracking number 5, which is the most numerous in this cluster.

[0080] In one embodiment, a second target cluster with a tracking threshold less than or equal to a preset tracking threshold is determined, and the sorting number of each pulse descriptor in the second target cluster is incremented by a preset value. The specific execution process of this embodiment is described in the following embodiment.

[0081] In another embodiment, no processing is performed on the second target cluster whose tracking threshold is less than or equal to the preset tracking threshold, that is, no correction processing is performed on the sorting number of each pulse descriptor in the second target cluster.

[0082] The radar signal sorting method provided in this invention determines the tracking threshold for each target cluster based on the ratio of the number of target tracking numbers (the most numerous among all clusters) to the number of pulse descriptors in each cluster. This corrects the sorting numbers of pulse descriptors in a first target cluster whose tracking threshold is greater than a preset tracking threshold to the target tracking numbers of that first target cluster, thereby improving the sorting accuracy of radar signals. Furthermore, only the sorting numbers of pulse descriptors in a first target cluster whose tracking threshold is greater than the preset tracking threshold are corrected to the target tracking numbers of that first target cluster, ensuring that the corrected sorting numbers are the most numerous tracking numbers in the cluster and that the corrected sorting numbers are the target tracking numbers of clusters whose tracking threshold is greater than the preset tracking threshold, thereby further improving the sorting accuracy of radar signals.

[0083] Based on any of the above embodiments, after step 132, the method further includes:

[0084] A second target cluster is determined where the tracking threshold is less than or equal to a preset tracking threshold, and the sorting number of each pulse descriptor in the second target cluster is incremented by a preset value.

[0085] Here, the number of second target clusters can be one or more, and the second target cluster is a cluster among multiple clusters. That is, the multiple clusters may include one or more first target clusters. Of course, the multiple clusters may not include the second target cluster.

[0086] For any second target cluster, the sorting number of each pulse descriptor in the second target cluster is corrected to the sum of each sorting number and a preset value. The preset value should be greater than or equal to the largest sorting number, so that the sorting number corrected by adding the preset value can distinguish the sorting number obtained from sorting. Preferably, the preset value can be 100.

[0087] For example, a certain cluster has a sorting number of 2 and a pulse descriptor number of 500. The tracking number of the pulse descriptor can include 1-7. The tracking number of 440 pulse descriptors in this cluster is 5, that is, the target tracking number of this cluster is 5. Based on this, the tracking threshold of this cluster is 0.88. If the preset tracking threshold is 0.9, then the tracking threshold of 0.88 is less than the preset tracking threshold of 0.9. Let the preset value be 100, then the sorting number 2 of each pulse descriptor in this cluster is corrected to 102.

[0088] The radar signal sorting method provided in this embodiment of the invention performs a preset value increment on the sorting number of each pulse descriptor in the second target cluster whose tracking threshold is less than or equal to a preset tracking threshold. This distinguishes the sorting number from the previously obtained sorting pulse sequence, thereby separating the pulse descriptors with tracking abnormalities and improving the sorting accuracy of radar signals.

[0089] Based on any of the above embodiments, in this method, step 110 includes:

[0090] Blind clustering is performed on the pulse sequences to be sorted to obtain multiple clusters.

[0091] The multiple clusters are numbered respectively to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted.

[0092] Specifically, blind clustering (blind sorting) is performed on each pulse descriptor in the pulse sequence to be sorted, resulting in multiple clusters. Each cluster corresponds to a sorting number.

[0093] In one embodiment, after comparing various clustering algorithms, DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is used to perform blind clustering of the pulse sequence to be sorted, thus eliminating the need to specify the number of categories. This algorithm has a fast clustering speed, thereby improving the sorting efficiency of radar signals. Furthermore, this algorithm can effectively handle noise points and discover spatial clusters of arbitrary shapes, thereby improving the sorting accuracy of radar signals and ultimately meeting the requirements of radar signal sorting scenarios.

[0094] The radar signal sorting method provided in this embodiment of the invention performs blind clustering processing on the pulse sequence to be sorted, thereby eliminating the need to specify the number of categories, thus improving the clustering speed and improving the sorting efficiency of radar signals; and the blind clustering processing does not require training data, thereby improving the convenience of radar signal sorting.

[0095] Based on any of the above embodiments, the method further includes:

[0096] The pulse sequence to be sorted is input into the model identification model to obtain the model identification result of each pulse description word in the pulse sequence to be sorted output by the model identification model;

[0097] The model identification model is trained based on sample pulse data and the corresponding model identification result labels.

[0098] Here, the model identification model is used to identify which radar model the input signal is from.

[0099] The radar signal sorting method provided in this embodiment of the invention further inputs the pulse sequence to be sorted into a model identification model to obtain the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model identification model. Thus, it not only outputs the sorting number of each pulse descriptor in the pulse sequence to be sorted, but also outputs the model identification result of each pulse descriptor in the pulse sequence to be sorted, so as to comprehensively determine the radar signal sorting result based on the sorting number and the model identification result, thereby further improving the sorting accuracy of radar signals.

[0100] Based on any of the above embodiments, after inputting the pulse sequence to be sorted into the model identification model and obtaining the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model identification model, the method further includes:

[0101] Determine the confidence level corresponding to each model identification result;

[0102] If the target model identification result is determined to have a confidence level less than or equal to a preset confidence threshold, the target model identification result is updated to the out-of-collection data identification result.

[0103] The sample pulse data includes pulse sequences of N types of radars, the type identification model is an N-classification model, and the out-of-set data identification results indicate that the radar type does not belong to the N types.

[0104] Here, the preset confidence threshold is a pre-set confidence threshold, which can be set according to actual needs. Preferably, the preset confidence threshold is 0.75.

[0105] Here, the number of target model identification results can be one or more, and each target model identification result is a model identification result among all model identification results. That is, each model identification result can include one or more target model identification results. Of course, each model identification result may not include target model identification results. For any target model identification result, update that target model identification result to the off-set data identification result.

[0106] It should be noted that no processing is required for model identification results with a confidence level greater than the preset confidence threshold; that is, the model identification results are still based on the output of the model identification model.

[0107] Considering that the model identification model is trained based on pulse sequences of N types of radars, and the trained model is an N-classification model, while the pulse sequences to be sorted may include signals other than those N types of radars, it is designed for an open-set scenario. The test data can include data corresponding to radar models other than the training data. Therefore, it is necessary to determine the confidence level of each model identification result so that the target model identification results with confidence levels less than or equal to the preset confidence threshold are updated to the out-of-set data identification results.

[0108] In one embodiment, considering that the pulse sequence includes PDW, after comparing various neural networks and traditional machine learning classification algorithms, the model identification model can select an SVM classifier to improve the accuracy of model identification.

[0109] The radar signal sorting method provided in this embodiment of the invention determines the confidence level corresponding to each model identification result, so as to update the target model identification result with a confidence level less than or equal to a preset confidence threshold to the out-of-set data identification result, thereby preventing the signal of other radar models that do not belong to the N models in the set of data from being identified as any of the N models, thereby improving the accuracy of radar model identification.

[0110] Based on any of the above embodiments, the step of inputting the pulse sequence to be sorted into the model identification model to obtain the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model identification model further includes:

[0111] Based on the corrected sorting number of each pulse descriptor in the pulse sequence to be sorted, the pulse sequence to be sorted is re-sorted to obtain multiple re-sorting clusters;

[0112] Based on the model identification results of each pulse descriptor in each of the re-sorting clusters, the most numerous model identification results in each of the re-sorting clusters are determined respectively;

[0113] The model identification results of each pulse descriptor in each of the re-sorting clusters are corrected to the maximum model identification results.

[0114] Here, the corrected sorting number is the sorting number corrected in step 130 above, and the sorting number corrected in step 130 above may include sorting numbers that have not actually been corrected.

[0115] Here, the resorting cluster includes at least one pulse descriptor, and pulse descriptors with the same sorting number (the latest one) are grouped together.

[0116] For any sorting cluster, the model identification results of each pulse descriptor in the sorting cluster are counted, and the model identification result with the largest number in the sorting cluster is determined.

[0117] For any re-sorting cluster, the model identification result of each pulse descriptor in that re-sorting cluster is corrected to the model identification result with the highest number in that re-sorting cluster. For example, if the model identification result with the highest number in a certain cluster is 7, then the model identification result of all pulse descriptors in that cluster is corrected to 7.

[0118] The radar signal sorting method provided in this embodiment of the invention takes into account that pulse descriptors belonging to the same sorting number also belong to the same radar model (signals emitted by the same radar individual). Therefore, based on the corrected sorting number of each pulse descriptor in the pulse sequence to be sorted, the pulse sequence to be sorted is re-sorted to obtain multiple re-sorting clusters. Based on the model identification results of each pulse descriptor in each re-sorting cluster, the most numerous model identification results in each re-sorting cluster are determined. The model identification results of each pulse descriptor in each re-sorting cluster are then corrected to the most numerous model identification results, thereby further improving the accuracy of radar model identification.

[0119] Based on any of the above embodiments, after step 110, the method further includes:

[0120] The pulse sequence of the reference beat is input into the model identification model to obtain the model identification result of each pulse descriptor in the pulse sequence of the reference beat output by the model identification model. Based on the model identification results of each pulse descriptor in each type of cluster, the most frequent model identification result in each type of cluster is determined. The model identification results of each pulse descriptor in each type of cluster are then corrected to the most frequent model identification results. The pulse sequence to be sorted at this time is the pulse sequence of the reference beat.

[0121] The radar signal sorting method provided in this embodiment of the invention, for the pulse sequence of the reference beat, also considers that pulse descriptors belonging to the same sorting number also belong to the same radar model (signals emitted by the same radar individual). Therefore, based on the model identification results of each pulse descriptor in each type of cluster, the most numerous model identification results in each type of cluster are determined respectively, so as to correct the model identification results of each pulse descriptor in each type of cluster to the most numerous model identification results, thereby further improving the accuracy of radar model identification.

[0122] To facilitate understanding of the above embodiments, a specific embodiment will be described here, such as... Figure 3 As shown, firstly, a model recognition model is trained based on pulse sequences of N types of radar; secondly, the sorting, tracking, and recognition steps of the radar signals are as follows:

[0123] Blind clustering is performed on the single-beat pulse sequence to be sorted, and multiple clusters are obtained by blind sorting, as well as the sorting number of each pulse descriptor in the pulse sequence to be sorted.

[0124] Determine whether the pulse sequence to be sorted is a pulse sequence of the reference beat;

[0125] Based on the pulse sequence of the benchmark beat and the sorting number of each pulse descriptor in the pulse sequence of the benchmark beat, the tracking model is trained;

[0126] The pulse sequence to be sorted is determined to be a pulse sequence other than the pulse sequence of the reference beat in a pulse sequence of multiple beats. The pulse sequence to be sorted is input into the tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model.

[0127] Based on the tracking number of each pulse descriptor in each cluster, the number of target tracking numbers in each cluster is determined. Based on the number of each number and the number of pulse descriptors in each cluster, the tracking threshold of each cluster is determined.

[0128] Determine if the tracking threshold is greater than the preset tracking threshold;

[0129] Identify the first target cluster whose tracking threshold is greater than the preset tracking threshold, and correct the sorting number of each pulse descriptor in the first target cluster to the target tracking number of the first target cluster;

[0130] A second target cluster is determined whose tracking threshold is less than or equal to a preset tracking threshold, and the sorting number of each pulse descriptor in the second target cluster is incremented by a preset value.

[0131] The pulse sequence to be sorted is input into the model identification model to obtain the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model identification model, and the confidence level corresponding to each model identification result is obtained.

[0132] Determine whether each confidence level is greater than the preset confidence threshold;

[0133] Once the target model identification result is determined to have a confidence level less than or equal to the preset confidence threshold, the target model identification result is updated to the out-of-collection data identification result.

[0134] Based on the latest sorting number of each pulse descriptor in the pulse sequence to be sorted, the pulse sequence to be sorted is re-sorted to obtain multiple re-sorting clusters; based on the model identification results of each pulse descriptor in each re-sorting cluster, the most numerous model identification results in each re-sorting cluster are determined; the model identification results of each pulse descriptor in each re-sorting cluster are corrected to the most numerous model identification results.

[0135] Output the latest sorting number of each pulse descriptor in the pulse sequence to be sorted, and the latest model identification result of each pulse descriptor in the pulse sequence to be sorted.

[0136] To facilitate understanding of the technical effects of the above embodiments, the above embodiments are evaluated using indicators such as sorting accuracy, sorting batch increase rate, sorting incorrect batch rate, model recognition accuracy, and signal tracking stability. The above embodiments show improved overall performance in terms of these five indicators and are more timely. Among them, the sorting accuracy rate is the ratio of the correctly sorted result to all sorted results; the sorting batching rate is the ratio of the batched result to all sorted results, where a pulse emitted by a radar individual is batched into two numbers; the sorting mis-batching rate is the ratio of the mis-batching result to all sorted results, where pulses emitted from different radar individuals are batched into the same number, i.e., sorting fails; the model identification accuracy rate is the ratio of the successfully identified result to all model identification results, where successful identification means that in a sorting cycle, if the sorted signal i is correctly sorted and corresponds to target j, and the model identification result of signal i is the model label corresponding to target j, then target j is considered successfully identified; signal tracking stability refers to the fact that in the signal sorting process, after each cycle performs sorting independently, the sorting results between cycles need to be correlated. The number of the signal corresponding to a certain target should remain consistent throughout the entire sorting task, i.e., the signal is stably tracked between cycles.

[0137] The radar signal sorting device provided by the present invention is described below. The radar signal sorting device described below and the radar signal sorting method described above can be referred to in correspondence.

[0138] Figure 4 This is a schematic diagram of the radar signal sorting device provided by the present invention, as shown below. Figure 4 As shown, the radar signal sorting device includes:

[0139] The sorting module 410 is used to sort the pulse sequence to be sorted to obtain multiple clusters, and to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted, wherein the sorting numbers of each pulse descriptor in any cluster are the same.

[0140] The tracking module 420 is used to determine that the pulse sequence to be sorted is a pulse sequence other than the pulse sequence of the reference pulse in a pulse sequence of multiple beats, input the pulse sequence to be sorted into the tracking model, and obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model. The pulse sequence of multiple beats is obtained based on the pulse data division of multi-target radar aliasing.

[0141] The correction module 430 is used to correct the sorting number of each pulse descriptor in the pulse sequence to be sorted based on the tracking number of each pulse descriptor in the pulse sequence to be sorted.

[0142] The tracking model is trained based on a sample pulse sequence and the sorting number label corresponding to the sample pulse sequence. The sample pulse sequence is the pulse sequence of the reference beat, and the sorting number label is the sorting number of each pulse descriptor in the pulse sequence of the reference beat.

[0143] The radar signal sorting device provided in this embodiment of the invention divides the pulse data of multi-target radar aliasing into multiple pulse sequences. Based on the pulse sequence of the reference pulse sequence and the sorting number of each pulse descriptor in the reference pulse sequence, a tracking model is trained. The pulse sequences to be sorted in the multiple pulse sequences, excluding the pulse sequence of the reference pulse, are input into the tracking model. The tracking model outputs the tracking number of each pulse descriptor in the pulse sequence to be sorted. Based on the tracking number of each pulse descriptor in the pulse sequence to be sorted, the sorting number of each pulse descriptor obtained from the sorting of the pulse sequence to be sorted is corrected, thereby improving the sorting accuracy of radar signals. Furthermore, the post-processing correction of the sorted numbers has a small computational load, thereby improving the sorting efficiency of radar signals.

[0144] Based on any of the above embodiments, the correction module 430 includes:

[0145] The quantity determination unit is used to determine the number of target tracking numbers in each of the clusters based on the tracking number of each pulse descriptor in each cluster, wherein any target tracking number is the tracking number with the largest number in the cluster;

[0146] A threshold determination unit is used to determine the tracking threshold of each of the above-mentioned clusters based on the number of each number and the number of pulse descriptors of each cluster. The tracking threshold of any cluster is determined based on the ratio of the number of the number of the cluster to the number of pulse descriptors of the cluster.

[0147] The numbering correction unit is used to determine a first target cluster whose tracking threshold is greater than a preset tracking threshold, and to correct the sorting number of each pulse descriptor in the first target cluster to the target tracking number of the first target cluster.

[0148] Based on any of the above embodiments, the numbering correction unit is further configured to:

[0149] A second target cluster is determined where the tracking threshold is less than or equal to a preset tracking threshold, and the sorting number of each pulse descriptor in the second target cluster is incremented by a preset value.

[0150] Based on any of the above embodiments, the sorting module 410 includes:

[0151] The sequence clustering unit is used to perform blind clustering of the pulse sequences to be sorted, resulting in multiple clusters.

[0152] The cluster numbering unit is used to number the multiple clusters respectively to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted.

[0153] Based on any of the above embodiments, the device further includes:

[0154] The model recognition module is used to input the pulse sequence to be sorted into the model recognition model and obtain the model recognition result of each pulse description word in the pulse sequence to be sorted output by the model recognition model;

[0155] The model identification model is trained based on sample pulse data and the corresponding model identification result labels.

[0156] Based on any of the above embodiments, the device further includes:

[0157] A confidence level determination module is used to determine the confidence level corresponding to each of the model identification results;

[0158] The result update module is used to determine the target model identification result whose confidence level is less than or equal to a preset confidence threshold, and update the target model identification result to the out-of-collection data identification result.

[0159] The sample pulse data includes pulse sequences of N types of radars, the type identification model is an N-classification model, and the out-of-set data identification results indicate that the radar type does not belong to the N types.

[0160] Based on any of the above embodiments, the device further includes:

[0161] The re-sorting module is used to re-sort the pulse sequence to be sorted based on the corrected sorting number of each pulse descriptor in the pulse sequence to be sorted, so as to obtain multiple re-sorting clusters;

[0162] The identification and determination module is used to determine the most numerous model identification results in each of the re-sorting clusters based on the model identification results of each pulse descriptor in each of the re-sorting clusters.

[0163] The identification correction module is used to correct the model identification result of each pulse descriptor in each of the re-sorting clusters to the maximum model identification result.

[0164] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540. The processor 510 can call logic instructions in the memory 530 to execute a radar signal sorting method, which includes: sorting a pulse sequence to be sorted to obtain multiple clusters, and obtaining a sorting number for each pulse descriptor in the pulse sequence to be sorted, wherein the sorting numbers for each pulse descriptor in any cluster are the same; determining that the pulse sequence to be sorted is a pulse sequence other than the pulse sequence of the reference pulse in a pulse sequence of multiple beats; inputting the pulse sequence to be sorted into a tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model, wherein the pulse sequence of multiple beats is obtained based on the pulse data division of multi-target radar aliasing; and correcting the sorting number of each pulse descriptor in the pulse sequence to be sorted based on the tracking number of each pulse descriptor in the pulse sequence to be sorted; wherein the tracking model is trained based on a sample pulse sequence and the sorting number label corresponding to the sample pulse sequence, the sample pulse sequence is the pulse sequence of the reference pulse, and the sorting number label is the sorting number of each pulse descriptor in the pulse sequence of the reference pulse.

[0165] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0166] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the radar signal sorting method provided by the above methods. The method includes: sorting a pulse sequence to be sorted to obtain multiple clusters, and obtaining a sorting number of each pulse descriptor in the pulse sequence to be sorted, wherein the sorting numbers of each pulse descriptor in any cluster are the same; determining that the pulse sequence to be sorted is a pulse sequence other than the pulse sequence of the reference beat in a pulse sequence of multiple beats, and sorting the pulse sequence to be sorted... The pulse sequence is input to the tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted, which is output by the tracking model. The pulse sequence of multiple beats is obtained by dividing the pulse data of multi-target radar aliasing. Based on the tracking number of each pulse descriptor in the pulse sequence to be sorted, the sorting number of each pulse descriptor in the pulse sequence to be sorted is corrected. The tracking model is trained based on sample pulse sequences and the sorting number labels corresponding to the sample pulse sequences. The sample pulse sequences are the pulse sequences of the reference beat, and the sorting number labels are the sorting numbers of each pulse descriptor in the pulse sequence of the reference beat.

[0167] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a radar signal sorting method provided by the methods described above. This method includes: sorting a pulse sequence to be sorted to obtain multiple clusters, and obtaining a sorting number for each pulse descriptor in the pulse sequence to be sorted, wherein the sorting numbers for each pulse descriptor in any cluster are the same; determining that the pulse sequence to be sorted is a pulse sequence other than the pulse sequence of a reference pulse among multiple pulse sequences; and inputting the pulse sequence to be sorted into a tracking model to obtain the... The tracking model outputs the tracking number of each pulse descriptor in the pulse sequence to be sorted. The pulse sequence of multiple beats is obtained by dividing the pulse data of multi-target radar aliasing. Based on the tracking number of each pulse descriptor in the pulse sequence to be sorted, the sorting number of each pulse descriptor in the pulse sequence to be sorted is corrected. The tracking model is trained based on sample pulse sequences and the sorting number labels corresponding to the sample pulse sequences. The sample pulse sequences are the pulse sequences of the reference beat, and the sorting number labels are the sorting numbers of each pulse descriptor in the pulse sequence of the reference beat.

[0168] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0169] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0170] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A radar signal sorting method, characterized in that, include: The pulse sequence to be sorted is sorted to obtain multiple clusters, and the sorting number of each pulse descriptor in the pulse sequence to be sorted is obtained, wherein the sorting number of each pulse descriptor in any cluster is the same; The pulse sequence to be sorted is determined to be a pulse sequence other than the pulse sequence of the reference pulse in a pulse sequence of multiple beats. The pulse sequence to be sorted is input into the tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model. The pulse sequence of multiple beats is obtained by dividing the pulse data based on the aliasing of multi-target radar. The pulse sequence of the reference beat is the pulse sequence of the first beat among the pulse sequences of the plurality of beats; Based on the tracking number of each pulse descriptor in the pulse sequence to be sorted, the sorting number of each pulse descriptor in the pulse sequence to be sorted is corrected, including: correcting the sorting number of each pulse descriptor in at least one cluster to the tracking number with the largest number in the cluster. The tracking model is trained based on a sample pulse sequence and the sorting number label corresponding to the sample pulse sequence. The sample pulse sequence is the pulse sequence of the reference beat, and the sorting number label is the sorting number of each pulse descriptor in the pulse sequence of the reference beat.

2. The radar signal sorting method according to claim 1, characterized in that, The step of correcting the sorting number of each pulse descriptor in the pulse sequence to be sorted based on the tracking number of each pulse descriptor in the pulse sequence to be sorted includes: Based on the tracking number of each pulse descriptor in each cluster, the number of target tracking numbers in each cluster is determined, and any target tracking number is the tracking number with the largest number in the cluster. Based on the number of each number and the number of pulse descriptors of each cluster, a tracking threshold for each cluster is determined. The tracking threshold for any cluster is determined based on the ratio of the number of numbers of the cluster to the number of pulse descriptors of the cluster. A first target cluster whose tracking threshold is greater than a preset tracking threshold is identified, and the sorting number of each pulse descriptor in the first target cluster is corrected to the target tracking number of the first target cluster.

3. The radar signal sorting method according to claim 2, characterized in that, The step of determining the tracking threshold for each of the aforementioned clusters based on the number of each number and the number of pulse descriptors for each cluster, further includes: A second target cluster is determined where the tracking threshold is less than or equal to a preset tracking threshold, and the sorting number of each pulse descriptor in the second target cluster is incremented by a preset value.

4. The radar signal sorting method according to claim 1, characterized in that, The process of sorting the pulse sequences to be sorted yields multiple clusters, and the sorting number of each pulse descriptor in the pulse sequences to be sorted is obtained, including: Blind clustering is performed on the pulse sequences to be sorted to obtain multiple clusters. The multiple clusters are numbered respectively to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted.

5. The radar signal sorting method according to any one of claims 1 to 4, characterized in that, Also includes: The pulse sequence to be sorted is input into the model identification model to obtain the model identification result of each pulse description word in the pulse sequence to be sorted output by the model identification model; The model identification model is trained based on sample pulse data and the corresponding model identification result labels.

6. The radar signal sorting method according to claim 5, characterized in that, The step of inputting the pulse sequence to be sorted into the model identification model to obtain the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model identification model, and then further includes: Determine the confidence level corresponding to each model identification result; If the target model identification result is determined to have a confidence level less than or equal to a preset confidence threshold, the target model identification result is updated to the out-of-collection data identification result. The sample pulse data includes pulse sequences of N types of radars, the type identification model is an N-classification model, and the out-of-set data identification results indicate that the radar type does not belong to the N types.

7. The radar signal sorting method according to claim 5, characterized in that, The step of inputting the pulse sequence to be sorted into the model identification model to obtain the model identification result of each pulse descriptor in the pulse sequence to be sorted output by the model identification model, and then further includes: Based on the corrected sorting number of each pulse descriptor in the pulse sequence to be sorted, the pulse sequence to be sorted is re-sorted to obtain multiple re-sorting clusters; Based on the model identification results of each pulse descriptor in each of the re-sorting clusters, the most numerous model identification results in each of the re-sorting clusters are determined respectively; The model identification results of each pulse descriptor in each of the re-sorting clusters are corrected to the maximum model identification results.

8. A radar signal sorting device, characterized in that, include: The sorting module is used to sort the pulse sequence to be sorted to obtain multiple clusters, and to obtain the sorting number of each pulse descriptor in the pulse sequence to be sorted, wherein the sorting numbers of each pulse descriptor in any cluster are the same. The tracking module is used to determine that the pulse sequence to be sorted is a pulse sequence other than the pulse sequence of the reference pulse in a pulse sequence of multiple beats, and inputs the pulse sequence to be sorted into the tracking model to obtain the tracking number of each pulse descriptor in the pulse sequence to be sorted output by the tracking model. The pulse sequence of multiple beats is obtained based on the pulse data division of multi-target radar aliasing. The pulse sequence of the reference beat is the pulse sequence of the first beat among the pulse sequences of the plurality of beats; The correction module is used to correct the sorting number of each pulse descriptor in the pulse sequence to be sorted based on the tracking number of each pulse descriptor in the pulse sequence to be sorted, including: correcting the sorting number of each pulse descriptor in at least one cluster to the tracking number with the largest number in the cluster. The tracking model is trained based on a sample pulse sequence and the sorting number label corresponding to the sample pulse sequence. The sample pulse sequence is the pulse sequence of the reference beat, and the sorting number label is the sorting number of each pulse descriptor in the pulse sequence of the reference beat.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the radar signal sorting method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the radar signal sorting method as described in any one of claims 1 to 7.

Images