Radar pulse sorting method and system based on machine learning

Abstract

The invention provides a radar pulse sorting method and system based on machine learning, and the method comprises the steps: S1, receiving a radar signal, and carrying out the preprocessing of the radar signal, and obtaining a radar pulse parameter; the pretreatment comprises frequency conversion treatment and measurement treatment; s2, performing parameter matching on the radar knowledge base and the radar pulse parameters to obtain a radar pulse pre-sorting result; the pre-sorting result comprises a known radar pulse flow and an unknown radar pulse flow; s3, performing main sorting on the pre-sorting result by using a decision tree algorithm and a clustering algorithm to obtain corresponding radar parameters; and S4, updating the radar knowledge base by using the obtained radar parameters, and completing the sorting of the interlaced pulse streams. According to the method, the known radar pulses are subjected to main sorting by using the supervised learning algorithm in machine learning, the sorting success rate is high, the real-time performance is good, the unknown radar pulses are subjected to main sorting by using the unsupervised learning algorithm, and the number and parameters of the unknown radar pulses can be effectively identified.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a radar pulse sorting method and system based on machine learning. Background Art

[0002] Radar radiation source signal sorting is the first step in radar electronic countermeasures. Its purpose is to sort out the pulse sequences of each radar from the intertwined radar signals and estimate and identify the parameters of each radar. With the emergence of new radar systems, the radar's frequency, pulse interval and other parameters no longer maintain a fixed mode. In addition, the battlefield electromagnetic environment is becoming more and more complex, and the pulse density received by the receiver at the same time is increasing, which greatly increases the difficulty of radar radiation source sorting and identification.

[0003] Traditional pulse repetition interval-based sorting algorithms include cumulative difference histogram algorithm and sequence difference histogram algorithm, which use a single parameter of the repetition interval of signal pulses for sorting. Such methods are greatly affected by pulse repetition interval jitter and are inefficient when the number of pulses increases sharply.

[0004] The clustering-based sorting algorithm uses radar pulse description words PDW (frequency, amplitude, pulse width, pulse repetition period, pulse arrival time, etc.) to cluster pulses, and considers pulses clustered in the same category as signals radiated by the same radiation source. The amount of computation in this method increases sharply as the number of pulses increases. Using only the clustering algorithm for sorting cannot guarantee real-time sorting requirements. Summary of the invention

[0005] In view of the defects in the prior art, the object of the present invention is to provide a radar pulse sorting method and system based on machine learning.

[0006] A radar pulse sorting method based on machine learning provided by the present invention includes:

[0007] Step S1: receiving radar signals and preprocessing them to obtain radar pulse parameters;

[0008] The preprocessing includes frequency conversion processing and measurement processing;

[0009] Step S2: matching the radar knowledge base with the radar pulse parameters to obtain the radar pulse pre-sorting result;

[0010] The pre-sorting result includes a known radar pulse stream and an unknown radar pulse stream;

[0011] Step S3: Use decision tree algorithm and clustering algorithm to perform main sorting on the pre-sorting results to obtain corresponding radar parameters;

[0012] Step S4: Use the obtained radar parameters to update the radar knowledge base and complete the sorting of the interlaced pulse stream.

[0013] Preferably, step S1 includes using a radar receiver to receive a radar radiation signal through an antenna, converting the radar radiation signal to an intermediate frequency and then processing it to obtain and measure radar pulse parameters; the radar pulse parameters include arrival time, pulse width, pulse amplitude, arrival angle and frequency.

[0014] Preferably, the arrival time, pulse width, pulse amplitude, arrival angle and frequency constitute a radar pulse description word; and the change value of the same parameter in the radar pulse description word is within a preset range.

[0015] Preferably, step S3 includes using a decision tree algorithm to perform primary sorting on known radar pulse streams to obtain known radar parameters; using a clustering algorithm to perform primary sorting on unknown radar pulse streams to obtain unknown radar parameters; the clustering algorithm includes a DBSCAN algorithm.

[0016] Preferably, the sorting result of the interleaved pulse stream is a radar radiation source parameter; the radar radiation source parameters include average azimuth, average amplitude, frequency range and average pulse width; the pulse repetition interval is calculated according to the pulse arrival time of the clustering result and the radar PRI type is obtained, the pulse repetition period is calculated according to the arrival time to the same radar radiation source and the pulse inter-modulation type of the radar radiation source is analyzed, the radar frequency modulation and agile frequency type are analyzed according to the frequency distribution, and the radar radiation source knowledge base is improved.

[0017] A radar pulse sorting system based on machine learning is provided according to the present invention, comprising:

[0018] Module M1: receives radar signals and pre-processes them to obtain radar pulse parameters;

[0019] The preprocessing includes frequency conversion processing and measurement processing;

[0020] Module M2: matching the radar knowledge base with the radar pulse parameters to obtain the radar pulse pre-sorting result;

[0021] The pre-sorting result includes a known radar pulse stream and an unknown radar pulse stream;

[0022] Module M3: Use decision tree algorithm and clustering algorithm to perform main sorting on the pre-sorting results to obtain corresponding radar parameters;

[0023] Module M4: Use the obtained radar parameters to update the radar knowledge base and complete the sorting of the interlaced pulse stream.

[0024] Preferably, the module M1 includes using a radar receiver to receive a radar radiation signal through an antenna, converting the radar radiation signal to an intermediate frequency and then processing it to obtain and measure radar pulse parameters; the radar pulse parameters include arrival time, pulse width, pulse amplitude, arrival angle and frequency.

[0025] Preferably, the arrival time, pulse width, pulse amplitude, arrival angle and frequency constitute a radar pulse description word; and the change value of the same parameter in the radar pulse description word is within a preset range.

[0026] Preferably, the module M3 includes using a decision tree algorithm to perform primary sorting on known radar pulse streams to obtain known radar parameters; using a clustering algorithm to perform primary sorting on unknown radar pulse streams to obtain unknown radar parameters; the clustering algorithm includes a DBSCAN algorithm.

[0027] Preferably, the sorting result of the interleaved pulse stream is a radar radiation source parameter; the radar radiation source parameters include average azimuth, average amplitude, frequency range and average pulse width; the pulse repetition interval is calculated according to the pulse arrival time of the clustering result and the radar PRI type is obtained, the pulse repetition period is calculated according to the arrival time to the same radar radiation source and the pulse inter-modulation type of the radar radiation source is analyzed, the radar frequency modulation and agile frequency type are analyzed according to the frequency distribution, and the radar radiation source knowledge base is improved.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] 1. The present invention utilizes radar knowledge base parameter matching for pre-sorting, effectively reduces the pulse density of unknown radiation sources, and divides radar pulse streams into known radar pulse streams and unknown radar pulse streams; utilizes decision trees to perform primary sorting on known radar pulse streams, and can quickly establish decision tree models based on known radar knowledge bases, and realize real-time sorting; utilizes DBSCAN clustering to perform primary sorting on unknown radar pulse streams, and clustering parameters include azimuth, pulse width, frequency, and amplitude, and obtains the number of unknown radar radiation sources and the parameters of each unknown radar radiation source, and adds the unknown radar radiation sources to the radar knowledge base, and sorts out unknown radar radiation sources in real time on the basis of ensuring sorting accuracy.

[0030] 2. The present invention utilizes the supervised learning algorithm in machine learning to perform the main sorting of known radar pulses, with a high sorting success rate and good real-time performance; and utilizes the unsupervised learning algorithm to perform the main sorting of unknown radar pulses, which can effectively identify the number and parameters of unknown radar pulses.

[0031] Other beneficial effects of the present invention will be explained in the specific implementation manner through the introduction of specific technical features and technical solutions. Through the introduction of these technical features and technical solutions, those skilled in the art should be able to understand the beneficial technical effects brought about by the technical features and technical solutions. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0033] Figure 1 Schematic diagram of radar radiation source sorting.

[0034] Figure 2 The figure is a flow chart of the method of the present invention.

[0035] Figure 3 It is the sorting model of the decision tree in the present invention.

[0036] Figure 4 It is the density clustering sorting model in the present invention. DETAILED DESCRIPTION

[0037] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0038] Reference Figure 1 As shown in the figure, the purpose of radar radiation source signal sorting is to sort out the pulse sequences of each radar from the interlaced radar signals, and to estimate and identify the parameters of each radar. The radar parameters that need to be obtained include: pulse arrival time, pulse width, pulse amplitude, pulse arrival angle, pulse signal frequency, pulse repetition period (the time difference between adjacent pulses of the same radiation source), pulse repetition period type (fixed pulse repetition period, sliding, jittering), and inter-pulse frequency type (frequency agility, frequency diversity, frequency coding, etc.).

[0039] In order to solve the above problems, the present invention adopts a pre-sorting algorithm based on a radar knowledge base combined with a main sorting algorithm based on machine learning to sort out unknown radar radiation sources in real time while ensuring the sorting accuracy.

[0040] Specifically, refer to Figure 2 As shown, a radar pulse sorting method based on machine learning includes:

[0041] Step S1: The radar receiver receives the radar radiation signal through the antenna, and the frequency conversion module converts the radar signal to the intermediate frequency, and measures the radar pulse parameters of the spatial radiation after signal processing. The radar pulse parameters include: arrival time TOA, pulse width PW (the pulse width of the continuous wave signal takes the maximum value measured), pulse amplitude PA, arrival angle DOA, and frequency RF. The above five parameters are combined into the radar pulse description word PDW.

[0042] Step S2: In the pulse description word, the parameters such as pulse amplitude PA, arrival angle DOA, pulse width PW of the same radar radiation source will not change dramatically; while the frequency of frequency modulation and frequency agile radar will change within a certain range. Through parameter matching, known radar radiation sources can be screened out. The pre-sorting method based on parameter matching of the radar knowledge base can divide the radar pulse stream into unknown radar pulse stream and known radar pulse stream, effectively reducing the pulse density of unknown radiation sources.

[0043] Step S3: Refer to Figure 3 As shown, the main sorting of the known radar pulse stream is performed based on the decision tree: according to the known radar knowledge base, a decision tree model can be quickly established and real-time sorting can be achieved.

[0044] The decision tree algorithm establishes a tree structure based on the overall rules and internal connections of the input data set, dividing the parameters into several intervals to determine which type of target a sample belongs to. The decision tree algorithm will continuously optimize the interval division according to the sorting accuracy to improve the sorting accuracy, so it is suitable for the main sorting of known radar emitters.

[0045] Step S4: Refer to Figure 4 As shown in the figure, the unknown radar pulse stream is sorted based on clustering. The clustering algorithm selects the density-based clustering algorithm, which does not need to specify the number of clustering categories and can find data categories of any shape. The unknown radar main sorting does not know how many unknown radar radiation sources there are in the data stream in advance, and the characteristic parameters of the unknown radiation sources are uncertain, so it is suitable to use the DBSCAN algorithm based on density clustering.

[0046] Step S5: After the DBSCAN algorithm is completed, the cluster center becomes a new known radar radiation source and is written into the radar knowledge base.

[0047] Finally, the interlaced pulse stream is sorted, and the sorting results are radar radiation source parameters: average azimuth (DOA), average amplitude (PA), frequency range (RFmin~RFmax), average pulse width (PW), etc. Then, the pulse repetition interval (PRI) can be calculated based on the pulse arrival time (TOA) of the clustering result and the radar PRI type can be obtained. The pulse repetition period can be calculated based on the arrival time to the same radar radiation source to analyze the pulse inter-pulse modulation type of the radar radiation source. The radar frequency modulation and agile frequency type can be analyzed based on the frequency distribution, and the radar radiation source knowledge base can be improved.

[0048] The present invention utilizes the supervised learning algorithm in machine learning to perform main sorting on known radar pulses, with high sorting success rate and good real-time performance. It utilizes the unsupervised learning algorithm to perform main sorting on unknown radar pulses, and can effectively identify the number and parameters of unknown radar pulses.

[0049] The present invention also provides a radar pulse sorting system based on machine learning. The radar pulse sorting system based on machine learning can be implemented by executing the process steps of the radar pulse sorting method based on machine learning, that is, those skilled in the art can understand the radar pulse sorting method based on machine learning as a preferred implementation of the radar pulse sorting system based on machine learning.

[0050] Specifically, a radar pulse sorting system based on machine learning includes:

[0051] Module M1: receives radar signals and pre-processes them to obtain radar pulse parameters;

[0052] The preprocessing includes frequency conversion processing and measurement processing;

[0053] Module M2: matching the radar knowledge base with the radar pulse parameters to obtain the radar pulse pre-sorting result;

[0054] The pre-sorting result includes a known radar pulse stream and an unknown radar pulse stream;

[0055] Module M3: Use decision tree algorithm and clustering algorithm to perform main sorting on the pre-sorting results to obtain corresponding radar parameters;

[0056] Module M4: Use the obtained radar parameters to update the radar knowledge base and complete the sorting of the interlaced pulse stream.

[0057] The module M1 includes using a radar receiver to receive radar radiation signals through an antenna, converting the radar radiation signals to an intermediate frequency and then processing them to obtain and measure radar pulse parameters; the radar pulse parameters include arrival time, pulse width, pulse amplitude, arrival angle and frequency.

[0058] The arrival time, pulse width, pulse amplitude, arrival angle and frequency constitute a radar pulse description word; and the change value of the same parameter in the radar pulse description word is within a preset range.

[0059] The module M3 includes using a decision tree algorithm to perform primary sorting on known radar pulse streams to obtain known radar parameters; using a clustering algorithm to perform primary sorting on unknown radar pulse streams to obtain unknown radar parameters; the clustering algorithm includes a DBSCAN algorithm.

[0060] The sorting result of the interlaced pulse stream is the radar radiation source parameter; the radar radiation source parameters include average azimuth, average amplitude, frequency range and average pulse width; the pulse repetition interval is calculated according to the pulse arrival time of the clustering result and the radar PRI type is obtained, the pulse repetition period is calculated according to the arrival time to the same radar radiation source and the pulse inter-modulation type of the radar radiation source is analyzed, the radar frequency modulation and agile frequency type are analyzed according to the frequency distribution, and the radar radiation source knowledge base is improved.

[0061] Those skilled in the art know that, in addition to realizing the system and its various devices, modules, and units provided by the present invention in a purely computer-readable program code, it is entirely possible to realize the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for realizing various functions can also be regarded as structures within the hardware component; the devices, modules, and units for realizing various functions can also be regarded as both software modules for realizing the method and structures within the hardware component.

[0062] In the description of the present application, it should be understood that the terms "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0063] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A radar pulse sorting method based on machine learning, characterized in that: include: Step S1: receiving radar signals and preprocessing them to obtain radar pulse parameters; The preprocessing includes frequency conversion processing and measurement processing; Step S2: matching the radar knowledge base with the radar pulse parameters to obtain the radar pulse pre-sorting result; The pre-sorting result includes a known radar pulse stream and an unknown radar pulse stream; Step S3: Use decision tree algorithm and clustering algorithm to perform main sorting on the pre-sorting results to obtain corresponding radar parameters; Step S4: Use the obtained radar parameters to update the radar knowledge base and complete the sorting of the interlaced pulse stream.

2. The radar pulse sorting method based on machine learning according to claim 1 is characterized in that: The step S1 includes using a radar receiver to receive a radar radiation signal through an antenna, converting the radar radiation signal to an intermediate frequency and then processing it to obtain and measure radar pulse parameters; the radar pulse parameters include arrival time, pulse width, pulse amplitude, arrival angle and frequency.

3. The radar pulse sorting method based on machine learning according to claim 2 is characterized in that: The arrival time, pulse width, pulse amplitude, arrival angle and frequency constitute a radar pulse description word; and the change value of the same parameter in the radar pulse description word is within a preset range.

4. The radar pulse sorting method based on machine learning according to claim 3 is characterized in that: The step S3 includes using a decision tree algorithm to perform primary sorting on known radar pulse streams to obtain known radar parameters; using a clustering algorithm to perform primary sorting on unknown radar pulse streams to obtain unknown radar parameters; the clustering algorithm includes a DBSCAN algorithm.

5. The radar pulse sorting method based on machine learning according to claim 4 is characterized in that: The sorting result of the interlaced pulse stream is the radar radiation source parameter; the radar radiation source parameters include average azimuth, average amplitude, frequency range and average pulse width; the pulse repetition interval is calculated according to the pulse arrival time of the clustering result and the radar PRI type is obtained, the pulse repetition period is calculated according to the arrival time to the same radar radiation source and the pulse inter-modulation type of the radar radiation source is analyzed, the radar frequency modulation and agile frequency type are analyzed according to the frequency distribution, and the radar radiation source knowledge base is improved.

6. A radar pulse sorting system based on machine learning, characterized in that: include: Module M1: receives radar signals and pre-processes them to obtain radar pulse parameters; The preprocessing includes frequency conversion processing and measurement processing; Module M2: matching the radar knowledge base with the radar pulse parameters to obtain the radar pulse pre-sorting result; The pre-sorting result includes a known radar pulse stream and an unknown radar pulse stream; Module M3: Use decision tree algorithm and clustering algorithm to perform main sorting on the pre-sorting results to obtain corresponding radar parameters; Module M4: Use the obtained radar parameters to update the radar knowledge base and complete the sorting of the interlaced pulse stream.

7. The radar pulse sorting system based on machine learning according to claim 6, characterized in that: The module M1 includes using a radar receiver to receive radar radiation signals through an antenna, converting the radar radiation signals to an intermediate frequency and then processing them to obtain and measure radar pulse parameters; the radar pulse parameters include arrival time, pulse width, pulse amplitude, arrival angle and frequency.

8. The radar pulse sorting system based on machine learning according to claim 7, characterized in that: The arrival time, pulse width, pulse amplitude, arrival angle and frequency constitute a radar pulse description word; and the change value of the same parameter in the radar pulse description word is within a preset range.

9. The radar pulse sorting system based on machine learning according to claim 8, characterized in that: The module M3 includes using a decision tree algorithm to perform primary sorting on known radar pulse streams to obtain known radar parameters; using a clustering algorithm to perform primary sorting on unknown radar pulse streams to obtain unknown radar parameters; the clustering algorithm includes a DBSCAN algorithm.

10. The radar pulse sorting system based on machine learning according to claim 9, characterized in that: The sorting result of the interlaced pulse stream is the radar radiation source parameter; the radar radiation source parameters include average azimuth, average amplitude, frequency range and average pulse width; the pulse repetition interval is calculated according to the pulse arrival time of the clustering result and the radar PRI type is obtained, the pulse repetition period is calculated according to the arrival time to the same radar radiation source and the pulse inter-modulation type of the radar radiation source is analyzed, the radar frequency modulation and agile frequency type are analyzed according to the frequency distribution, and the radar radiation source knowledge base is improved.

Images