Pulse description word sorting method based on pulse repetition interval transformation method

Abstract

The invention relates to the technical field of electronic reconnaissance, and discloses a pulse description word sorting method based on a pulse repetition interval transformation method, and the method comprises the steps: firstly carrying out the pre-sorting based on the phase, pulse width, frequency and other characteristics of pulse description words, and achieving the preliminary classification of pulse data; then, on the basis of pre-sorting, an improved PRI transformation method is adopted for main sorting, and through accurate calculation and optimized PRI sequence analysis, the operation speed is increased, and a corresponding PRI value is rapidly determined; and finally, extracting a matched pulse sequence from the original sequence by using the target PRI in combination with a sequence retrieval algorithm, and ensuring that the pulse description words are efficiently and accurately sorted and extracted in a complex signal environment.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic reconnaissance, and particularly relates to a pulse description word sorting method based on the pulse repetition interval transformation method. Background Art

[0002] Traditional sorting methods based on Pulse Description Word (PDW) mainly rely on the Direction of Arrival (DOA) carried in the PDW for sparse processing, and then obtain the secondary feature PRI. Early commonly used algorithms include the cumulative difference histogram method and the sequence difference histogram, etc. These methods are simple, but are vulnerable to the interference of signal harmonics, resulting in incorrect estimation of the Pulse Repetition Interval (PRI), and at the same time, the computational complexity is large. In 2000, Nishiguchi proposed the PRI transformation method to estimate the PRI through the autocorrelation function, but this method is vulnerable to jitter. Subsequently, Nishiguchi improved this method by using PRI bins and considering the jitter rate, thus improving the PRI estimation effect for jittery pulse sequences. However, the calculation speed of this method is largely limited by the preset parameters. Therefore, it is of great practical significance to improve the speed of the PRI transformation method. Summary of the Invention

[0003] In view of the above deficiencies in the prior art, the present invention provides a pulse description word sorting method based on the pulse repetition interval transformation method.

[0004] In order to achieve the above-mentioned invention purpose, the technical solution adopted by the present invention is as follows:

[0005] A pulse description word sorting method based on the pulse repetition interval transformation method, comprising the following steps:

[0006] Cluster the input pulse description word data by using the phase, pulse width, and frequency carried in the pulse description word to obtain pre-sorted pulse description word classes;

[0007] Adaptively estimate the preset parameters of the pulse repetition interval transformation method for the pre-sorted pulse description word classes;

[0008] Perform pulse repetition interval transformation by using the preset parameters of the pulse repetition interval transformation method to obtain the target pulse repetition intervals corresponding to each pulse description word class;

[0009] Retrieve the matching pulse sequences from the pulse description word data by using the target pulse repetition intervals corresponding to each pulse description word class.

[0010] Further, for the input pulse descriptor word data, clustering is performed using the phase, pulse width, and frequency carried by the pulse descriptor word to obtain pre-sorted pulse descriptor word classes, including:

[0011] Input pulse descriptor word data, where each pulse descriptor word data contains the corresponding phase, pulse width, frequency, and arrival time;

[0012] Perform the first clustering on the phase features carried by the pulse descriptor word to obtain the phase clustering center and the clustered pulse descriptor word data;

[0013] Jointly perform the second clustering on the pulse width features and frequency features carried by the clustered pulse descriptor word data to obtain the pulse width clustering center, the frequency clustering center, and the clustered pulse descriptor word data;

[0014] Filter each clustered pulse descriptor word data using an effective number threshold to obtain the effectively clustered pulse descriptor word data.

[0015] Further, the clustering method includes:

[0016] A1. Initialize the clustering center, the number of clustering centers, and the corresponding number within the class;

[0017] A2. Calculate the error between the features carried by the pulse descriptor word data and the clustering center;

[0018] A3. Determine whether the calculated error is less than the set feature clustering threshold; if so, classify the pulse descriptor word as the clustering center, update the clustering center simultaneously, and jump to step A5; otherwise, jump to step A4;

[0019] A4. Determine whether the clustering center serial number is equal to the number of clustering centers; if so, generate a new clustering center and jump to A5; otherwise, update the clustering center serial number and jump to step A3;

[0020] A5. Update the pulse descriptor word serial number and initialize the clustering center serial number, then jump to step A3.

[0021] Further, the preset parameters for adaptively estimating the pulse repetition interval transformation method for the pre-sorted pulse descriptor word classes include:

[0022] Extract the arrival time sequence of the pre-sorted pulse descriptor word, perform differencing on the arrival time sequence to obtain the pulse repetition interval sequence, and statistically calculate the minimum value and the maximum value of the pulse repetition interval sequence;

[0023] Generate an equidistant sequence from the minimum value to the maximum value of the pulse repetition interval sequence according to the set step size;

[0024] Statistically count the frequency of each pulse repetition interval falling into the corresponding equal-interval range according to the equal-interval sequence, and calculate the scoring probability of the equal-interval range based on the ratio of the frequency of the equal-interval range to the length of the pulse repetition interval sequence;

[0025] Select the equal-interval ranges with a scoring probability greater than zero, and calculate the window centers of the equal-interval ranges;

[0026] Perform pulse loss suppression processing on the window center sequence of the equal-interval ranges;

[0027] Statistically count the maximum and minimum values of the window center sequence of the processed window centers of the equal-interval ranges, and calculate the maximum and minimum values of the expected range of the pulse repetition interval.

[0028] Further, perform pulse repetition interval transformation using the preset parameters of the pulse repetition interval transformation method to obtain the target pulse repetition intervals corresponding to each pulse descriptor class, including:

[0029] Determine the transformation range according to the maximum and minimum values of the expected range of the pulse repetition interval;

[0030] Perform pulse repetition interval transformation within the transformation range, and determine whether there is a pulse repetition interval exceeding the threshold. If so, record the corresponding pulse repetition interval as the target pulse repetition interval, otherwise proceed to the next step;

[0031] Generate an equal-interval sequence from the minimum value to the maximum value of the pulse repetition interval sequence according to the set step size;

[0032] Statistically count the frequency of each pulse repetition interval falling into the corresponding equal-interval range according to the equal-interval sequence, and calculate the scoring probability of the equal-interval range based on the ratio of the frequency of the equal-interval range to the length of the pulse repetition interval sequence;

[0033] Select the equal-interval ranges with a scoring probability greater than zero, and calculate the window centers of the equal-interval ranges;

[0034] Select the window center with the highest scoring probability in the window center sequence of the equal-interval ranges, and record the corresponding pulse repetition interval as the target pulse repetition interval.

[0035] Further, retrieve the matching pulse sequences from the pulse descriptor data using the target pulse repetition intervals corresponding to each pulse descriptor class, including:

[0036] Use the direct search method to find the starting pulse within the pulse descriptor class;

[0037] According to the starting pulse, use the curve fitting method for searching, calculate the distance between the remaining pulses and the fitting curve, and extract the optimal pulses.

[0038] Further, finding the starting pulse within the pulse descriptor class using the direct search method includes:

[0039] Traverse the pulse descriptor data, set the first pulse descriptor as the starting pulse, and calculate the theoretical arrival time for each subsequent pulse;

[0040] Determine whether there is a difference between the theoretical arrival times of a continuously set number of pulses that is less than or equal to the set tolerance compared to the actual arrival times; if so, use the current starting pulse as the starting pulse obtained by the search; otherwise, update the starting pulse and continue the search.

[0041] Further, perform a search using the curve fitting method based on the starting pulse, calculate the distance between the remaining pulses and the fitting curve, and extract the optimal pulse, including:

[0042] B1. Fit an initial linear model based on several actual pulses matched to the starting pulse;

[0043] B2. Calculate the theoretical arrival time of the target pulse; if the theoretical arrival time is greater than the maximum pulse arrival time carried in the pulse descriptor to be retrieved, jump to step B5; otherwise, jump to step B3;

[0044] B3. Match according to the theoretical arrival time in the pulse descriptor to be retrieved within a certain tolerance based on the actual pulse arrival time. If no match is found, it means the pulse is lost, and return to step B2 for re-search; otherwise, classify the pulse descriptor closest to the theoretical arrival time as the target pulse and jump to step B4;

[0045] B4. Perform linear fitting through the pulse arrival time point of the currently retrieved successful target pulse, update the parameters in the linear model, and return to step B2 for re-search;

[0046] B5. The theoretical arrival time has exceeded the range of the pulses to be retrieved, and it is considered that this search ends.

[0047] Further, the initial linear model is specifically:

[0048]

[0049] where, is the pulse arrival time, is the pulse sequence number, is the pulse repetition interval, is the starting time.

[0050] Further, calculating the error between the theoretical arrival time and the actual arrival time is specifically:

[0051]

[0052] Wherein, is the error between the theoretical arrival time and the actual arrival time, is the pulse arrival time of the i-th pulse descriptor word, is the pulse sequence number of the i-th pulse descriptor word, is the pulse repetition interval, is the starting time, is the number of pulse descriptor words.

[0053] The present invention has the following beneficial effects:

[0054] By introducing the collaborative sparse processing of phase, frequency, and pulse width characteristics, the present invention effectively reduces the pulse density in the PRI transformation. Combining with the parameter estimation method of the adaptive PRI transformation, the dependence of the algorithm on the initial parameters is reduced. At the same time, through the PRI value iterative optimization and curve fitting in sequence retrieval, especially in the case of a high pulse loss rate, the robustness and accuracy of sequence retrieval are significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 is a schematic flow chart of a pulse descriptor word sorting method based on the pulse repetition interval transformation method in an embodiment of the present invention;

[0056] Figure 2 is a schematic flow chart of a multi-feature collaborative sparse method in an embodiment of the present invention;

[0057] Figure 3 is a schematic flow chart of an adaptive PRI transformation parameter estimation method in an embodiment of the present invention;

[0058] Figure 4 is a schematic flow chart of the PRI transformation and target PRI extraction in an embodiment of the present invention;

[0059] Figure 5 is a schematic flow chart of a sequence retrieval method in an embodiment of the present invention;

[0060] Figure 6 is a PDW data diagram of simulation in an embodiment of the present invention;

[0061] Figure 7 is a processing result diagram of the simulation data after corresponding processing in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0062] The specific embodiments of the present invention will be described below to facilitate those skilled in the art to understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions made using the concept of the present invention are within the scope of protection.

[0063] As Figure 1 shown, an embodiment of the present invention provides a pulse description word sorting method based on the pulse repetition interval transformation method, including the following steps S1 to S4:

[0064] S1. For the input pulse description word data, use the phase, pulse width, and frequency carried by the pulse description word to perform clustering to obtain pre-sorted pulse description word classes;

[0065] In an alternative embodiment of the present invention, step S1 uses the phase, pulse width, and frequency carried by the pulse description word to perform clustering on the input pulse description word data to obtain pre-sorted pulse description word classes, including:

[0066] Input the pulse description word data, and each pulse description word data includes the corresponding phase, pulse width, frequency, and arrival time;

[0067] Perform the first clustering on the phase features carried by the pulse description word to obtain the phase clustering center and the clustered pulse description word data;

[0068] Perform the second clustering jointly on the pulse width features and frequency features carried by the clustered pulse description word data to obtain the pulse width clustering center, the frequency clustering center, and the clustered pulse description word data;

[0069] Filter each clustered pulse description word data using the effective number threshold to obtain the pulse description word data of effective clustering.

[0070] Among them, the clustering method includes:

[0071] A1. Initialize the clustering center, the number of clustering centers, and the number within the corresponding class;

[0072] A2. Calculate the error between the features carried by the pulse description word data and the clustering center;

[0073] A3. Determine whether the calculated error is less than the set feature clustering threshold; if so, classify the pulse description word as the clustering center, and at the same time update the clustering center and jump to step A5; otherwise, jump to step A4;

[0074] A4. Determine whether the cluster center serial number is equal to the number of cluster centers; if so, generate a new cluster center and jump to A5; otherwise, update the cluster center serial number and jump to step A3;

[0075] A5. Update the pulse descriptor serial number and initialize the cluster center serial number, then jump to step A3.

[0076] As Figure 2 shown, step S1 in this embodiment specifically includes the following sub-steps:

[0077] S11. Input pulse descriptor data , and each pulse descriptor data contains the corresponding phase Phase, pulse width PW, frequency RF, and time of arrival TOA;

[0078] S12. Use the carried phase information to perform the first clustering based on the improved K-Means method;

[0079] Initialize the corresponding phase cluster center , the number of cluster centers , and the number within the corresponding class :

[0080]

[0081]

[0082]

[0083] Among them represents the phase carried by the first , and then traverse all data and perform the following calculations:

[0084] 1. Calculate the error between the current and :

[0085]

[0086] 2. If is satisfied, then classify into , and at the same time update and jump to step 4:

[0087]

[0088]

[0089] 3. If is not satisfied, judge Is equal to (i.e., all cluster centers fail to match). If satisfied, generate new cluster centers and jump to step 4:

[0090]

[0091]

[0092]

[0093] Otherwise, update , and jump to step 2;

[0094] 4. Update and and jump to step 1. When all are classified, exit.

[0095] S13. Based on the first clustering, for each class including perform a second clustering on the combination of PW and RF to obtain the center of each class after clustering and . The clustering idea is similar to that of S12 and will not be elaborated here;

[0096] At this time, for any class , the following conditions should be satisfied:

[0097]

[0098]

[0099]

[0100] Among them, represents the Euclidean distance between and , and are the clustering thresholds for Phase, RF, and PW features.

[0101] S14. Let the number of PDWs in each clustering result be , and the threshold of the effective number be . Then the output effective clustering results satisfy the following conditions:

[0102]

[0103] S2. Adaptively estimate the preset parameters of the pulse repetition interval transformation method for the pre-sorted pulse description word classes;

[0104] In an alternative embodiment of the present invention, the preset parameters of the step S2 for adaptively estimating the pulse repetition interval transformation method for pre-sorted pulse description words include:

[0105] Extract the arrival time sequence of the pre-sorted pulse description words, perform differencing on the arrival time sequence to obtain a pulse repetition interval sequence, and statistically calculate the minimum and maximum values of the pulse repetition interval sequence;

[0106] Generate an equidistant sequence from the minimum value to the maximum value of the pulse repetition interval sequence according to a set step size;

[0107] According to the equidistant sequence, statistically calculate the frequency of each pulse repetition interval falling into the corresponding equidistant interval for the pulse repetition interval sequence, and calculate the score probability of the equidistant interval according to the ratio of the frequency of the equidistant interval to the length of the pulse repetition interval sequence;

[0108] Select the equidistant intervals with a score probability greater than zero, and calculate the window center of the equidistant intervals;

[0109] Perform pulse loss suppression processing on the window center sequence of the equidistant intervals;

[0110] Statistically calculate the maximum and minimum values of the window center sequence of the processed equidistant intervals, and calculate the maximum and minimum values of the expected range of the pulse repetition interval.

[0111] As Figure 3 shown, the step S2 in this embodiment specifically includes the following sub-steps:

[0112] S21. Extract the TOA sequence of the effective clustering result output in step S1, denoted as , and further obtain the PRI sequence:

[0113]

[0114] S22. Statistically calculate the maximum and minimum values of the obtained PRI sequence, denoted as and ;

[0115] S23. With a step size of , generate to equidistant sequence :

[0116]

[0117] where is the rounding symbol;

[0118] Use the histogram statistical method to statistically analyze the distribution of the PRI sequence according to this equidistant sequence. That is, statistically analyze each PRI value Fall into the corresponding interval The frequency count within it is obtained to get the frequency count of each window :

[0119]

[0120] Where is the total length of the PRI sequence, is an indicator function, indicating that when falls into the interval the value is 1, otherwise 0. The scoring probability of each interval is defined as the ratio of the frequency count of this interval to the length of the PRI sequence :

[0121]

[0122] Among them, is the frequency count of the PRI sequence falling into the window , is the length of the total PRI sequence.

[0123] S24. Select the windows with a scoring probability greater than 0 and calculate their window centers, denoted as :

[0124]

[0125] S25. For the pulse loss situation existing in the actual data, perform pulse loss suppression on the PRI sequence:

[0126] For each ( ) check whether there exists such that is times, where :

[0127]

[0128] If it exists, discard the corresponding ;

[0129] S26. Statistically obtain its and and for the PRI sequence after pulse loss suppression, and further obtain and :

[0130]

[0131]

[0132] wherein 、 and are respectively the minimum value, the maximum value and the step size of the sequence. In this embodiment, the step size .

[0133] S3. Perform pulse repetition interval (PRI) transformation using the preset parameters of the PRI transformation method to obtain the target PRI corresponding to each pulse descriptor class;

[0134] In an alternative embodiment of the present invention, step S3 performs PRI transformation using the preset parameters of the PRI transformation method to obtain the target PRI corresponding to each pulse descriptor class, including:

[0135] Determine the transformation range according to the maximum value and the minimum value of the expected range of the PRI;

[0136] Perform PRI transformation within the transformation range, and determine whether there is a PRI exceeding the threshold. If so, record the corresponding PRI as the target PRI, otherwise proceed to the next step;

[0137] Generate an equidistant sequence from the minimum value to the maximum value of the PRI sequence according to the set step size;

[0138] Count the frequency of each PRI falling into the corresponding equidistant interval in the PRI sequence according to the equidistant sequence, and calculate the score probability of the equidistant interval according to the ratio of the frequency of the equidistant interval to the length of the PRI sequence;

[0139] Select the equidistant interval with a score probability greater than zero, and calculate the window center of the equidistant interval;

[0140] Select the window center with the maximum score probability in the window center sequence of the equidistant interval, and record the corresponding PRI as the target PRI.

[0141] As Figure 4 shown, step S3 in this embodiment specifically includes the following sub-steps:

[0142] S31. Read in data;

[0143] S32. According to and , perform PRI transformation within this range. If there is a PRI exceeding the threshold, record the corresponding PRI as the target PRI and jump to S34, otherwise jump to S33;

[0144] S33. When the PRI transformation fails, use the histogram statistics method to perform statistics on this PRI sequence to obtain sequence. Then select the corresponding score probability with the largest , denoted as the target PRI;

[0145] S34. Output the target PRI.

[0146] S4. Retrieve the matching pulse sequence from the pulse descriptor data using the target pulse repetition interval corresponding to each pulse descriptor class.

[0147] In an alternative embodiment of the present invention, step S4 retrieves the matching pulse sequence from the pulse descriptor data using the target pulse repetition interval corresponding to each pulse descriptor class, including:

[0148] Use the direct search method to find the starting pulse within the pulse descriptor class;

[0149] According to the starting pulse, use the curve fitting method to search, calculate the distance between the remaining pulses and the fitting curve, and extract the optimal pulse.

[0150] Among them, using the direct search method to find the starting pulse within the pulse descriptor class includes:

[0151] Traverse the pulse descriptor data, set the first pulse descriptor as the starting pulse, and calculate the theoretical arrival time for each subsequent pulse;

[0152] Determine whether there is a difference between the theoretical arrival times of a continuously set number of pulses that is less than or equal to the set tolerance compared to the actual arrival time; if so, use the current starting pulse as the starting pulse obtained by the search; otherwise, update the starting pulse and continue the search.

[0153] Among them, according to the starting pulse, using the curve fitting method to search, calculate the distance between the remaining pulses and the fitting curve, and extract the optimal pulse, including:

[0154] B1. Fit the initial linear model according to the m actual pulses matched by the starting pulse.

[0155] B2. Calculate the theoretical arrival time for the m + k-th pulse after the starting pulse (k starts from 1). If the theoretical arrival time is greater than the maximum value of TOA carried in the PDW to be retrieved, jump to step B5; otherwise, jump to step B3.

[0156] B3. Match according to the theoretical arrival time in the pulse descriptors to be matched within a certain tolerance according to the actual pulse arrival time. If none of them match, it means the pulse is lost. Update k and return to step B2 for re-search; otherwise, update k and jump to step B4.

[0157] B4. Perform linear fitting through n pulse arrival time points, update the parameters in the linear model, and return to step B2 for re-searching.

[0158] B5. If the theoretical arrival time has exceeded the range of the pulses to be retrieved, it is considered that the current search ends.

[0159] As Figure 5 shown, step S4 in this embodiment specifically includes the following sub-steps:

[0160] S41. Read in the current PDW to be retrieved.

[0161] S42. Search for the starting pulse of the sequence.

[0162] Assume that the pulses to be retrieved, each corresponds to an arrival time . Traverse the entire PDW:

[0163] 1. For the starting pulse , with the target PRI being , and setting a certain tolerance , search one by one backward:

[0164] For each subsequent pulse (starting from the one after the ), the theoretical arrival time is:

[0165]

[0166] For each subsequent pulse , its actual arrival time should satisfy:

[0167]

[0168] If there are multiple pulses that all satisfy the above formula, select the one closest to the theoretical TOA:

[0169]

[0170] 2. If no actual TOA can be paired with the theoretical TOA for consecutive times, it is considered that the current is not credible, update the starting point , and go back to step 1:

[0171]

[0172] 3. If , it means that no credible starting pulse is finally found, and the search ends;

[0173] If the current starting pulse successfully matches actual pulses, a reliable starting pulse is considered to be found , and the search is ended.

[0174] If a starting pulse is successfully found , go to S43. Otherwise, go to S44.

[0175] S43. After successfully finding the starting pulse , starting from this pulse, the curve fitting method is used for searching and iterating continuously :

[0176] Assume the starting pulse , the pulse arrival time can be represented by a linear equation as:

[0177]

[0178] where is the pulse arrival time, is the pulse sequence number, represents the pulse repetition interval (PRI), is the starting time. At the same time, define the error function:

[0179]

[0180] This sum of squared errors represents the difference between the actual arrival time and the theoretical arrival time.

[0181] At this time, the optimal PRI is solved by the least squares method. Let and have derivatives equal to zero, and the system of equations is obtained:

[0182]

[0183] Solve the system of equations to obtain the optimal slope (i.e., PRI) and intercept , and the formulas are as follows:

[0184]

[0185]

[0186] The process is as follows:

[0187] 1. According to the starting actual pulses found, fit to obtain the initial linear model , where represents the estimated value of the initial PRI.

[0188] 2. For the k-th pulse to be matched (k starts from 1), calculate its theoretical arrival time according to the current linear model:

[0189]

[0190] If is greater than the maximum value of TOA carried in the PDW to be retrieved, jump to step 5; otherwise, jump to step 3.

[0191] 3. According to Start the matching in the pulses to be retrieved: Calculate the error

[0192] and define the candidate matching set as . If is empty, it means that the k-th pulse is lost. Update and jump to step 2; otherwise, mark the smallest as the target pulse, update and jump to step 4;

[0193] 4. Perform linear fitting through the arrival time points of the currently retrieved pulses, update the current slope and intercept and return to step 2 for re-search.

[0194] 5. The currently retrieved pulse has exceeded the range of the pulses to be retrieved, and it is considered that this search is over.

[0195] S44. Output the result and the retrieval is completed.

[0196] The present invention simulates the PDW data of three radiation sources, and the specific parameters are shown in Table 1, and the corresponding data are as Figure 6 shown.

[0197] Table 1: Parameters of three radiation sources in the simulated PDW data

[0198] Radiation source Frequency (MHz) Pulse width (μs) Pulse repetition interval (μs) Radiation source 1 5 3 1000 Radiation source 2 5 6 1000 Radiation source 3 5 9 1000

[0199] After the processing of S1 to S4, the sorted PDW data is as Figure 7 shown. The results show that the three radiation sources are successfully separated from the mixed PDW, and effective sorting is achieved. In addition, the time consumption of the PRI transformation before and after the use of the adaptive estimation method for different numbers of pulses is also compared, and the results are shown in Table 2.

[0200] ​Table 2: PRI transformation time-consuming results before and after using the parameter adaptive estimation method

[0201] Number of pulses (pcs) Time consumed before use (s) Time consumed after use (s) 2000 0.257470 0.198543 4000 0.638562 0.331082 8000 2.380886 0.826612 16000 8.980357 2.413699

[0202] It can be seen that after adopting the parameter adaptive estimation method in the present invention, the PRI transformation speed under different pulse numbers has been improved, significantly enhancing the real-time performance of the PRI transformation in engineering applications.

[0203] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0204] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implement the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0205] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one process Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0206] Specific embodiments are applied in the present invention to elaborate on the principles and implementation manners of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

[0207] Those of ordinary skill in the art will realize that the embodiments described herein are provided to assist the reader in understanding the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations that do not depart from the essence of the present invention based on these technical revelations disclosed by the present invention, and these deformations and combinations are still within the scope of protection of the present invention.

Claims

1. A pulse description word sorting method based on the pulse repetition interval transformation method, characterized in that It includes the following steps: Cluster the input pulse descriptor word data by using the phase, pulse width, and frequency carried by the pulse descriptor word to obtain pre-sorted pulse descriptor word classes; Adaptively estimate the preset parameters of the pulse repetition interval transformation method for the pre-sorted pulse descriptor word classes; Perform pulse repetition interval transformation by using the preset parameters of the pulse repetition interval transformation method to obtain the target pulse repetition intervals corresponding to each pulse descriptor word class; Retrieve the matching pulse sequences from the pulse descriptor word data by using the target pulse repetition intervals corresponding to each pulse descriptor word class.

2. The pulse descriptor word sorting method based on the pulse repetition interval transformation method according to claim 1, wherein Cluster the input pulse descriptor word data by using the phase, pulse width, and frequency carried by the pulse descriptor word to obtain pre-sorted pulse descriptor word classes, including: Input the pulse descriptor word data, and each pulse descriptor word data includes the corresponding phase, pulse width, frequency, and arrival time; Perform the first clustering on the phase features carried by the pulse descriptor word to obtain the phase clustering center and the clustered pulse descriptor word data; Perform the second clustering jointly on the pulse width features and frequency features carried by the clustered pulse descriptor word data to obtain the pulse width clustering center, frequency clustering center, and the clustered pulse descriptor word data; Filter each clustered pulse descriptor word data by using the effective number threshold to obtain the effectively clustered pulse descriptor word data.

3. A pulse descriptor word sorting method based on the pulse repetition interval transformation method according to claim 2, characterized in that The clustering method includes: A1. Initialize the clustering center, the number of clustering centers, and the corresponding number within the class; A2. Calculate the error between the features carried by the pulse descriptor word data and the clustering center; A3. Determine whether the calculated error is less than the set feature clustering threshold; if so, classify the pulse descriptor word as the clustering center, and at the same time update the clustering center and jump to step A5; otherwise, jump to step A4; A4. Determine whether the clustering center serial number is equal to the number of clustering centers; if so, generate a new clustering center and jump to A5; otherwise, update the clustering center serial number and jump to step A3; A5. Update the pulse descriptor word serial number and initialize the clustering center serial number, and jump to step A3.

4. A pulse description word sorting method based on the pulse repetition interval transformation method according to claim 1, characterized in that Adaptive estimation of the preset parameters of the pulse repetition interval transformation method for the pre-sorted pulse descriptor word classes includes: Extract the arrival time sequence of the pre-sorted pulse descriptor word, perform differencing on the arrival time sequence to obtain the pulse repetition interval sequence, and statistically calculate the minimum value and maximum value of the pulse repetition interval sequence; Generate an equally spaced sequence from the minimum value to the maximum value of the pulse repetition interval sequence according to the set step size; According to the equally spaced sequence, statistically calculate the frequency of each pulse repetition interval falling into the corresponding equally spaced interval for the pulse repetition interval sequence, and calculate the score probability of the equally spaced interval according to the ratio of the frequency of the equally spaced interval to the length of the pulse repetition interval sequence; Select the equally spaced intervals with a score probability greater than zero, and calculate the window center of the equally spaced intervals; Perform pulse loss suppression processing on the window center sequence of the equally spaced intervals; Statistically calculate the maximum value and minimum value of the window center sequence for the processed window center sequence of the equally spaced intervals, and calculate the maximum value and minimum value of the expected range of the pulse repetition interval.

5. A pulse description word sorting method based on the pulse repetition interval transformation method according to claim 1, characterized in that, Perform pulse repetition interval transformation by using the preset parameters of the pulse repetition interval transformation method to obtain the target pulse repetition intervals corresponding to each pulse descriptor word class, including: Determine the transformation range according to the maximum and minimum values of the expected range of the pulse repetition interval; Perform pulse repetition interval transformation within the transformation range, and determine whether there is a pulse repetition interval exceeding the threshold. If so, record the corresponding pulse repetition interval as the target pulse repetition interval; otherwise, proceed to the next step; Generate an equally spaced sequence from the minimum value to the maximum value of the pulse repetition interval sequence according to the set step size; According to the equally spaced sequence, count the frequency of each pulse repetition interval falling into the corresponding equally spaced interval in the pulse repetition interval sequence, and calculate the scoring probability of the equally spaced interval according to the ratio of the frequency of the equally spaced interval to the length of the pulse repetition interval sequence; Select the equally spaced intervals with a scoring probability greater than zero, and calculate the window center of the equally spaced intervals; Select the window center with the highest scoring probability in the window center sequence of the equally spaced intervals, and record the corresponding pulse repetition interval as the target pulse repetition interval.

6. A pulse descriptor word sorting method based on the pulse repetition interval transformation method according to claim 1, characterized in that, Retrieve the matching pulse sequence from the pulse descriptor data using the target pulse repetition interval corresponding to each pulse descriptor class, including: Use the direct search method to find the starting pulse within the pulse descriptor class; According to the starting pulse, use the curve fitting method for search, calculate the distance between the remaining pulses and the fitting curve, and extract the optimal pulse.

7. A pulse description word sorting method based on the pulse repetition interval transformation method according to claim 6, characterized in that, Using the direct search method to find the starting pulse within the pulse descriptor class includes: Traverse the pulse descriptor data, set the first pulse descriptor as the starting pulse, and calculate the theoretical arrival time for each subsequent pulse; Determine whether there is a difference between the theoretical arrival times of a continuous set number of pulses that is less than or equal to the set tolerance compared to the actual arrival times; if so, use the current starting pulse as the starting pulse obtained by the search; otherwise, update the starting pulse and continue the search.

8. A pulse descriptor word sorting method based on the pulse repetition interval transformation method according to claim 6, characterized in that According to the starting pulse, use the curve fitting method for search, calculate the distance between the remaining pulses and the fitting curve, and extract the optimal pulse, including: B1. Fit the initial linear model according to several actual pulses matched by the starting pulse; B2. Calculate the theoretical arrival time of the target pulse; if the theoretical arrival time is greater than the maximum value of the pulse arrival time carried in the pulse descriptor to be retrieved, jump to step B5; otherwise, jump to step B3; B3. Match according to the theoretical arrival time in the pulse descriptors to be retrieved within a certain tolerance according to the actual pulse arrival time. If none of them match, it means the pulse is lost, and return to step B2 for re-search; otherwise, classify the pulse descriptor closest to the theoretical arrival time as the target pulse and jump to step B4; B4. Perform linear fitting through the pulse arrival time point of the currently retrieved successful target pulse, update the parameters in the linear model, and return to step B2 for re-search; B5. If the theoretical arrival time has exceeded the range of the pulses to be retrieved, end this search.

9. A pulse description word sorting method based on the pulse repetition interval transformation method according to claim 8, characterized in that The specific form of the initial linear model is: Among them, is the pulse arrival time, is the pulse sequence number, is the pulse repetition interval, is the starting time.

10. A pulse descriptor word sorting method based on the pulse repetition interval transformation method according to claim 8, characterized in that The specific calculation of the error between the theoretical arrival time and the actual arrival time is: wherein, is the error between the theoretical arrival time and the actual arrival time, is the pulse arrival time of the i-th pulse descriptor, is the pulse sequence number of the i-th pulse descriptor, is the pulse repetition interval, is the starting time, is the number of pulse descriptors.

Images