Radar pulse sorting method based on inverse distance weighted online clustering

Abstract

The invention discloses a radar pulse sorting method based on inverse distance weighted online clustering, which belongs to the field of radar signal sorting, and comprises the following steps of: 1, generating a large number of pulse samples with typical radar signal diversity by using python according to the characteristics of modern radar signals; step 2, performing full-amount pre-clustering on the first 1000 pulse data sets of the generated pulse data set; step 3, carrying out real-time signal sorting of single-sample inverse distance weighted similarity on the residual pulse data; step 4, carrying out real-time verification on a sorting result; 5, repeating the step 3 and the step 4 until all data are sorted; and step 6, repeating the step 3 and the step 4 by using an incremental Leader algorithm and a non-incremental DBSCAN and BGMM algorithm, and performing comparative test verification with the method of the invention, the method of the invention performs clustering association of pulses based on a discriminant criterion of weighted sum, overcomes the trouble of priori knowledge dependence, dynamically updates a cluster center through a moving average technology, and realizes higher sorting accuracy.

Description

Technical Field

[0001] The invention relates to the technical field of radar signal sorting, and in particular to a radar pulse sorting method based on inverse distance weighted online clustering. Background Art

[0002] Radar pulse sorting algorithm is a technology for extracting each radar pulse sequence from a randomly interwoven pulse stream. It is an important technical basis for radar signal parameter analysis and extraction, radar type identification, the core step in sensing the electronic battlefield situation, and also the key link in restricting radar confrontation. However, with the increasing complexity of the electromagnetic signal environment, radar signals are characterized by high density, diversification, and concealment, which brings unprecedented challenges to radar signal sorting.

[0003] The existing radar pulse sorting algorithms are mainly developed from two directions:

[0004] The first is a classification method based on intra-pulse features. By extracting combined features in multiple domains (time domain, frequency domain, time-frequency domain), the radar signal sorting problem is converted into a supervised classification problem. However, this method ignores the complex characteristics of the electromagnetic environment. For example, a large number of signal samples cannot be manually labeled, and it is difficult to find features with good generalization and high separability in the face of complex modulation, variable parameters, agile or even random signals. At the same time, the time cost of feature extraction and classifier training is unbearable under the high real-time requirements, so it is difficult to adapt to the needs of modern radar signal sorting.

[0005] The second is a method based on pulse-to-pulse features, which are usually described by pulse description words (PDW). Template matching is an early sorting method. By establishing a database containing characteristic parameters of radar radiation sources, the received signal is compared with the template in the database to achieve rapid signal sorting. However, with the development of radar technology and the increase in signal types, the template matching method faces challenges in database completeness and updating in practical applications.

[0006] Sorting methods based on pulse repetition interval (PRI) are another major type of inter-pulse modulation feature sorting technology. PRI describes the time interval between adjacent pulses in a pulse sequence, has relative stability, and is the key to the sorting algorithm. Technologies such as PRI search method, PRI histogram method, PRI transformation method, and plane transformation method are effective in processing scenarios with low pulse density, few radiation sources, and fixed PRI parameters. However, with the development of radar signal modulation technology, the non-constant characteristics of PRI (such as PRI agility and random modulation) and the increase in pulse density, traditional sorting methods based on PRI face challenges and their processing performance declines.

[0007] Unsupervised clustering-based methods have become an important research direction due to the difficulty in obtaining prior information on the target radiation source. Traditional clustering methods, such as K-Means and fuzzy C-means clustering, can process clusters of arbitrary shapes, but they require prior information such as the number of clusters to be set in advance, which limits their scope of use. Density-based clustering algorithms can mine clusters of arbitrary shapes, but they are highly sensitive to parameters and their performance may degrade or fail in dynamic scenes. In addition, these clustering algorithms process data in batches, cannot meet real-time requirements, and ignore the time-varying information behind the data points.

[0008] In recent years, incremental algorithms have begun to be used for radar emitter signal sorting, processing each pulse in sequence to achieve real-time sorting. The improved incremental DBSCAN algorithm and incremental Leader algorithm significantly reduce the computational complexity and optimize the processing of noise pulses through real-time clustering of single sample increments. However, when the data distribution of radar emitters overlaps, the performance of these algorithms will be seriously affected. Summary of the invention

[0009] In view of the shortcomings of the prior art, the present invention discloses a radar pulse sorting method based on inverse distance weighted online clustering, which solves the problems of insufficient real-time performance and insufficient prior information in the existing radar sorting technology.

[0010] To achieve the above object, the present invention provides the following technical solution: a radar pulse sorting method based on inverse distance weighted online clustering, comprising the following steps:

[0011] Step 1: Construct radar signal samples. According to the characteristics of modern radar signals, use Python to generate several pulse sequence samples with typical radar signal diversity.

[0012] Step 2: Perform full clustering on the first 1000 pulse sequences to obtain the initial cluster center list and category number;

[0013] Step 3: Based on the pre-obtained initial cluster center list and category serial number, the remaining data is sorted according to the time beat by single-sample inverse distance weighted similarity in real-time signal sorting;

[0014] Step 4: Real-time verification of the sorting results of step 3;

[0015] Step 5: Repeat steps 3 and 4 until all data are sorted;

[0016] Step 6: Repeat steps 3, 4, and 5 using the incremental Leader algorithm and the non-incremental DBSCAN and BGMM algorithms for comparative experimental verification.

[0017] Preferably, in step 1, firstly, it is determined that the radar signal sample includes five-dimensional features of the radar pulse description word:

[0018] TOA: Time of arrival, which indicates the time it takes for a pulse to reach the receiver, in seconds;

[0019] CF: Center frequency, indicating the frequency center of the pulse, in megahertz;

[0020] PW: Pulse width, which indicates the duration of the pulse in microseconds;

[0021] PA: Amplitude, which indicates the strength of the pulse, in volts;

[0022] DOA: Angle of arrival, indicating the direction of the pulse, in degrees;

[0023] 1) Determine the arrival time based on the three modulation modes of radar,

[0024] For a simple pulse radar, the first-order difference of arrival times is a constant:

[0025] ΔTOA(t)=C;

[0026] For frequency modulated continuous wave radar, the first-order difference of arrival time is a linear function of time:

[0027] ΔTOA(t)=t;

[0028] For nonlinear FM radar, the first-order difference of arrival time is a complex function of time:

[0029] ΔTOA(t)=f(t);

[0030] 2) The angle of arrival varies from 0 to 360 degrees

[0031] The continuously varying angle of arrival is simulated by taking a combination of intercepted sine, cosine, polynomial, exponential, and linear functions:

[0032]

[0033] By giving different a, b, c, d, e, f, g and DOA(t) value threshold ranges, the arrival angle changes of different radar signals in the real electromagnetic world are simulated;

[0034] 3) The center frequency range is from 1 GHz to 15 GHz, and the stable center frequency is described by a combination of a constant function and Gaussian noise:

[0035]

[0036] 4) The pulse width ranges from 0.05 microseconds to 35 microseconds, and the stable center frequency is described by a combination of a constant function and Gaussian noise:

[0037]

[0038] 5) The range of pulse amplitude variation. According to the definition of pulse amplitude, P is specified as a random number within a certain range:

[0039]

[0040] According to the above rules, 12 radar signal sources with different pulse numbers and scales were simulated. The pulse width, center frequency, arrival angle and amplitude arrival time were plotted on the plane, and each point represents a pulse.

[0041] Preferably, in step 2, using the first 1000 radar signal samples for full pre-clustering specifically includes the following:

[0042] For the first 1000 pulse trains x n (n=1,2,…,K), perform full clustering and obtain the initial cluster center list x i And category number i (1,2,…,C).

[0043] Preferably, in step 3, according to the pre-obtained initial cluster center list and category serial number, the remaining data is subjected to real-time signal sorting of single-sample inverse distance weighted similarity according to the time rhythm, which specifically includes the following steps:

[0044] Given a pulse query point x j , a list of cluster centers of known classes x i , i = 1, 2, ..., C, where C is the maximum index of all known clusters and a pulse stream T:

[0045] T={(CF1, DO stop 1), (CF2, DO stop 2), ..., (DF N , DO stop N )},

[0046] Where X j =(CF j , DOA j ) is the feature vector of the instance, j = 1, 2, ..., N;

[0047] Step 3.1: Calculate x j With the currently known cluster center x i The CF and DOA distance between:

[0048]

[0049] Step 3.2: Select the k cluster centers closest to each pulse feature, k is n / 2, n is the current number of clusters, recorded as:

[0050] N k1 (CF j )=min k distance CF ;

[0051] N k2 (DOA j )=min k distance DOA ;

[0052] Step 3.3: Based on the inverse distance weighted similarity, calculate each x i In N k (x q ) in the reciprocal s(x j , x i ),

[0053]

[0054] Where ∈ is a small positive number to avoid the situation where the denominator is zero. The calculated distance is converted into a reliable probability threshold. i In N k (x j ) is normalized so that the probability of each set of features sums to 1, which is provided for subsequent sorting, that is,

[0055]

[0056] Step 3.4: Get each x i The weight w(x j , x i ) after N k1 (CF j )∩N k2 (DOA j ) perform weighted sum of similarities to obtain the total weight W K =W(N k1 (CF j )∩N k2 (DOA j )), according to W K Make a judgment and get x j Category y q ,Right now

[0057]

[0058] n represents the current number of categories, and threshold represents the threshold set by the similarity function.

[0059] Preferably, in step 4, the sorting results in step 3 are verified in real time, and the experimental results are measured from three aspects: ARI, AMI, and ACC;

[0060] 1) ARI algorithm. The value range of ARI is -1 to 1. The larger the value, the better the clustering result. The calculation formula of ARI is:

[0061]

[0062] in, It represents the number of combinations of selecting k elements from n elements, n ij Indicates the number of samples contained in the i-th category in the clustering result and the j-th category in the true category, a i Indicates the number of samples in the i-th category in the clustering results, b j represents the number of samples of the jth class in the true category, and n represents the total number of samples;

[0063] 2) AMI algorithm. The value range of AMI is 0 to 1. The larger the value, the better the clustering result. The calculation formula of AMI is:

[0064]

[0065] Among them, MI represents the mutual information between the clustering result and the true category, H(U) and H(V) represent the entropy of the clustering result and the true category respectively, and E[MI] represents the expected mutual information between the clustering result and the true category when randomly assigned;

[0066] 3) The ACC algorithm is based on the Hungarian algorithm and is used to solve the maximum weight matching problem of bipartite graphs. The idea of ​​the ACC algorithm is to find a maximum weight matching so that each category in the clustering result corresponds to a category in the real category:

[0067]

[0068] Where w is a D×D matrix, D=max(max(y),max(Y))+1, and L is the Hungarian matching algorithm;

[0069] Finally, we get y for each time step in step 3. q Compare with the actual label and calculate the ARI, AMI, ACC scores, and take their average as the metric, as follows:

[0070]

[0071] The above score represents the average value of ARI, AMI and ACC scores.

[0072] Compared with the prior art, the present invention has the following beneficial effects:

[0073] 1. The present invention proposes a similarity measurement algorithm based on inverse distance, and designs a weighted sum-based discrimination criterion for clustering and associating pulses. The algorithm does not require a preset number of clusters, does not rely on the selection of sensitive parameters, can adapt to the overlap of data distribution, and overcomes the problem of reliance on prior knowledge. In addition, the algorithm uses incremental processing, only stores a small amount of data in memory, and can deinterleave pulses in real time, meeting the real-time requirements for data processing in modern electromagnetic environments. The simulation experimental results on the simulated data set show that the invention has a high sorting accuracy.

[0074] 2. The present invention proposes a radar pulse sorting method of inverse distance weighted online clustering. In order to dynamically learn the representation of pulses, the present invention adopts a method of dynamically updating the cluster center using a sliding average technology. The cluster center is represented as the average value of nearly one hundred pulse data streams, which is continuously and dynamically updated, so as to better track the changing law of pulses, automatically learn the pulse feature representation, and achieve higher sorting accuracy.

[0075] 3. In the present invention, in order to alleviate the pulse clustering phenomenon of different radars caused by the overlap of pulse features, the concept of k nearest neighbors is introduced to calculate the similarity of each pulse feature; by selecting the k cluster centers closest to the pulse feature, the algorithm can more accurately evaluate the similarity between the pulse and each cluster center; using the inverse distance weighted method, the algorithm converts the similarity into a probability form, avoiding complex threshold selection and ensuring the rationality of the similarity probability. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The accompanying drawings are used to provide further understanding of the present invention and constitute a part of the specification. They are used to explain the present invention together with the embodiments of the present invention and do not constitute a limitation of the present invention.

[0077] In the attached picture:

[0078] Figure 1 is a distribution diagram of a simulated data set of the present invention;

[0079] Figure 2 It is a flow chart of the algorithm flow chart of the present invention. DETAILED DESCRIPTION

[0080] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0081] A radar pulse sorting method based on inverse distance weighted online clustering comprises the following steps:

[0082] Step 1: Construct a radar signal sample. First, determine the five-dimensional features of the radar signal sample including the radar pulse description word:

[0083] TOA: Time of arrival, which indicates the time it takes for a pulse to reach the receiver, in seconds;

[0084] CF: Center frequency, indicating the frequency center of the pulse, in megahertz;

[0085] PW: Pulse width, which indicates the duration of the pulse in microseconds;

[0086] PA: Amplitude, which indicates the strength of the pulse, in volts;

[0087] DOA: Angle of arrival, indicating the direction of the pulse, in degrees;

[0088] The arrival time (TOA) of the radar signal ranges from 0 to 6×10^11s;

[0089] 1) Determine the time of arrival (TOA) based on the three modulation modes of radar,

[0090] For a simple pulse radar, the first-order difference of arrival times is a constant:

[0091] ΔTOA(t)=C

[0092] For frequency modulated continuous wave radar, the first-order difference of arrival time is a linear function of time:

[0093] ΔTOA(t)=t

[0094] For nonlinear FM radar, the first-order difference of arrival time is a complex function of time:

[0095] ΔTOA(t)=f(t)

[0096] 2) Angle of arrival (DOA) varies from 0 to 360 degrees

[0097] The continuously varying angle of arrival is simulated by taking a combination of intercepted sine, cosine, polynomial, exponential and linear functions:

[0098]

[0099] By giving different value threshold ranges of a, b, c, d, e, f, g and DOA(t), the changes in the arrival angles of different radar signals in the real electromagnetic world are simulated.

[0100] 3) Center frequency (CF) ranges from 1GHz to 15GHz

[0101] The stable center frequency is described by a combination of a constant function and Gaussian noise:

[0102]

[0103] 4) Pulse width (PW) ranges from 0.05 microseconds to 35 microseconds

[0104] The stable center frequency is described by a combination of a constant function and Gaussian noise:

[0105]

[0106] 5) The pulse amplitude (PA) has a large range of variation. According to the definition of pulse amplitude, P is specified as a random number within a certain range.

[0107]

[0108] like Figure 1 As shown in the figure, according to the above rules, 12 radar signal sources with different pulse numbers (from 54722 pulses to more than 976575 pulses) were simulated, and the pulse width (PW), center frequency (CF), arrival angle (DOA) and amplitude (PA) were plotted on the arrival time (TOA) plane, and each point represents a pulse.

[0109] Step 2: Use the first 1000 radar signal samples for full pre-clustering:

[0110] For the first 1000 pulse trains x n (n=1,2,…,K), perform full clustering and obtain the initial cluster center list x i And category number i (1,2,…,C).

[0111] Step 3: Based on the pre-obtained initial cluster center list and category serial number, the remaining data is sorted according to the time beat by single-sample inverse distance weighted similarity in real-time signal sorting:

[0112] Given a pulse query point x j , a list of cluster centers of known classes x i , i = 1, 2, ..., C, where C is the maximum index of all known clusters and a pulse stream T;

[0113] T={(CF1, DOA1), (CF2, DOA2),…, (DF N , DOA N )},

[0114] Where X j =(CF j , DOA j ) is the feature vector of the instance, j = 1, 2, …, N.

[0115] Step 3.1: Calculate x j With the currently known cluster center x i The CF and DOA distance between:

[0116]

[0117] Step 3.2: Since different radiation sources may have very similar distributions in certain features, the overlap of data distribution may lead to misjudgment of clustering results. In order to make the similarity probability more credible, the algorithm introduces the idea of ​​k nearest neighbors and selects the k cluster centers closest to each pulse feature (k is usually n / 2, n is the current number of clusters), recorded as:

[0118] N k1 (CF j )=min k distance CF

[0119] N k2 (DOA j )=min k distance DOA

[0120] Step 3.3: The present invention proposes an inverse distance weighted similarity, which calculates each x i In N k (x q ) in the reciprocal s(x j , x i ),

[0121]

[0122] Where ∈ is a small positive number to avoid the situation where the denominator is zero. The calculated distance is converted into a reliable probability threshold, avoiding the threshold selection of complex distance parameters. For each x i In N k (x j ) is normalized so that the probability of each set of features sums to 1, which is provided for subsequent sorting, that is,

[0123]

[0124] Step 3.4: Get each x i The weight w(x j , x i ), the final judgment is to select from the intersection of the similarity of each group of pulse features, avoiding the misclassification of pulses due to the similarity of a certain feature. k1 (CF j )∩N k2 (DOA j) perform weighted sum of similarities to obtain the total weight W K =W(N k1 (CF j )∩N k2 (DOA j )), according to W K Make a judgment and get x j Category y q ,Right now;

[0125]

[0126] n represents the current number of categories, and threshold represents the threshold set by the similarity function.

[0127] Step 4: Real-time verification of the sorting results in step 3:

[0128] In order to compare the advantages and disadvantages of different algorithms, the present invention measures the experimental results from three aspects: ARI, AMI, and ACC.

[0129] 1) ARI (Adjusted Rand Index) is an indicator used to measure the similarity between clustering results and true categories. It takes into account the impact of random assignment. The larger the value, the more similar the clustering results are to the true categories. The value range of ARI is -1 to 1. The larger the value, the better the clustering results. The calculation formula of ARI is:

[0130]

[0131] in, It represents the number of combinations of selecting k elements from n elements, n ij Indicates the number of samples contained in the i-th category in the clustering result and the j-th category in the true category, a i Indicates the number of samples in the i-th category in the clustering results, b j represents the number of samples of the jth category in the true category, and n represents the total number of samples.

[0132] 2) AMI (Adjusted Mutual Information) is an indicator used to measure the similarity between clustering results and true categories. It takes into account the impact of random assignment and category imbalance. The larger the value, the more similar the clustering results are to the true categories. The value range of AMI is 0 to 1. The larger the value, the better the clustering result. The calculation formula of AMI is:

[0133]

[0134] Among them, MI represents the mutual information between the clustering results and the true categories, H(U) and H(V) represent the entropy of the clustering results and the true categories respectively, and E[MI] represents the expected mutual information between the clustering results and the true categories when randomly assigned.

[0135] 3) ACC (Accuracy) is an indicator used to measure the proportion of correctly classified samples in the clustering results to the total number of samples. It only considers whether the sample classification is correct, and does not consider the classification accuracy and recall. Different from the general ACC, the ACC algorithm of the present invention is based on the Hungarian algorithm (also known as the Kuhn-Munkres algorithm), which is an algorithm for solving the maximum weight matching problem of bipartite graphs. The idea of ​​the ACC algorithm is to find a maximum weight matching so that each class in the clustering result corresponds to a class in the real class as much as possible.

[0136]

[0137] Where w is a D×D matrix, D=max(max(y),max(Y))+1, and L is the Hungarian matching algorithm.

[0138] Obtain y for each time step in step 3 q Compared with the actual label, the ARI, AMI, and ACC scores are calculated and their average is taken as the metric, as follows:

[0139]

[0140] Step 5: Repeat step 3 and step 4 until the data is finished. The algorithm flow chart is as follows: Figure 2 shown.

[0141] Step 6: Repeat steps 3, 4, and 5 using the incremental Leader algorithm and the non-incremental DBSCAN and BGMM algorithms for comparative experimental verification.

[0142] Examples are listed below to illustrate the present invention:

[0143] Embodiment 1:

[0144] The first 1000 pulses of the pulse stream are clustered (DBSCAN), and a total of 6 types of pulses appear. These 6 types of pulses are regarded as known radars, and their center points, CF mean and DOA mean are calculated, denoted as X i , starting from the 1001th pulse, pulse-by-pulse online clustering is performed to sort the remaining pulses. According to the pulse characteristics (center frequency, arrival angle), the pulses are processed non-incrementally and incrementally respectively.

[0145] In order to evaluate the performance of the present invention (IDP algorithm) and its comparative algorithm, three indicators are used: ARI (Adjusted Rand Index), AMI (Adjusted Mutual Information), and ACC (Accuracy). ARI and AMI can more comprehensively measure the degree of fit between the clustering results and the true category labels, and have been widely used in clustering algorithm evaluation; ACC directly reflects the proportion of clustering labels given by the algorithm that are the same as the true labels. The value range of the three is 0 to 1, and the closer to 1, the better the performance. Since the non-incremental algorithm is sensitive to initialization, that is, different initial clustering centers will affect the final clustering results, in order to ensure the uniqueness of the experiment, the random seed is fixed to 42, that is, the same initial clustering center is used each time the non-incremental algorithm is run. The experimental results show that in the complex conditions of signal interweaving, overlap, parameter changes and noise, the IDP algorithm is significantly better than other algorithms in the above three indicators. The presented ARI, AMI, and ACC values ​​are the average values ​​of 10 scenarios, as shown in Table 1;

[0146] Table 1 Comparison results of algorithms

[0147]

[0148] It can be seen that the ARI value of the IDP algorithm is 0.924, which is significantly higher than other algorithms; the AMI value is 0.935, which is also the highest; the ACC value is 0.948, while other algorithms are around 0.9. This fully verifies that IDP has better clustering accuracy and robustness. Among them, static DBSCAN and BGMM, as non-incremental algorithms, need to load all data at once, ignoring the influence of time parameters. The DBSCAN algorithm has a good ability to distinguish noise pulses, but it needs to reasonably select parameters such as the neighborhood radius, otherwise it is easy to merge different clusters or split the same cluster. The clustering quality of BGMM depends largely on the number of Gaussian distributions estimated in advance.

[0149] Algorithms usually have some user parameters that may affect the results and need to be fine-tuned. These parameters usually involve aspects such as the sensitivity, accuracy, and efficiency of the algorithm. As shown in Table 2, different algorithms have different parameters, for example:

[0150] IDP: threshold. As mentioned above, this is a similarity threshold used to determine whether two data points belong to the same class. If the distance between two data points is less than or equal to the threshold, they are considered to be of the same class. The smaller the threshold value, the more sensitive the algorithm is and the more clusters there are, but it may also lead to overfitting or noise.

[0151] Leader: Feature distance, which is a measure used to calculate the similarity between data points. Feature distance can be Euclidean distance, Manhattan distance, cosine distance, etc. The choice of feature distance depends on the distribution of the data and the type of features. The larger the value of feature distance, the easier it is for the algorithm to classify different data points into one category, and the fewer the number of clusters, but it may also lead to underfitting or loss of information.

[0152] DBSCAN: eps radius, which is a radius used to define the neighborhood. If there are enough data points in the neighborhood of a data point (greater than or equal to min_samples), then it is considered a core point, otherwise it is a boundary point or a noise point. The larger the value of eps radius, the easier it is for the algorithm to classify data points that are far apart into one category, and the fewer the number of clusters, but it may also lead to overfitting or noise.

[0153] BGMM: n_components, which is the number of components used to specify the Gaussian mixture model. Each component represents a Gaussian distribution that describes a subset of the data. The larger the value of n_components, the better the algorithm can capture the details of the data and the greater the number of clusters, but it may also lead to overfitting or increased computational complexity.

[0154] The average values ​​of ARI, AMI, and ACC are used to observe these effects. In this validation set, there are 11 categories in total;

[0155] Table 2 User parameter impact

[0156]

[0157] The IDP algorithm can obtain clustering results close to 11 categories under different threshold values, indicating that the algorithm is not very sensitive to the choice of parameters, and the number of clusters is consistent with the actual number of categories, indicating that the algorithm can divide the data well. The average evaluation index (ARI, AMI, ACC) of the algorithm reaches the highest value of 0.942 when the threshold is 0.6. The algorithm can better maintain the internal similarity and external difference of the data, and the clustering accuracy is also very high.

[0158] The Leader algorithm obtains clustering results of more than 11 categories under different eps values, indicating that the algorithm is sensitive to the choice of parameters. In addition, there is a gap between the number of clusters and the actual number of categories, indicating that the algorithm may separate some data points that belong to the same category.

[0159] The DBSCAN algorithm obtains clustering results of more than 11 categories under different eps values, and the algorithm is also sensitive to the choice of parameters. The number of clusters does not match the actual number of categories, and there is a more obvious batch phenomenon in this data set.

[0160] The BGMM algorithm has relatively good results under different n_components values. However, the BGMM algorithm needs to specify the number of clusters. When the number of clusters is greater than the actual number of categories, the evaluation index will drop significantly. The BGMM calculation complexity is also high because it needs to estimate the parameters of the Gaussian distribution of each component, and it assumes that the data follows a Gaussian distribution, which may not be in line with the actual situation.

[0161] Finally, it should be noted that the above description is only a preferred example of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A radar pulse sorting method based on inverse distance weighted online clustering, characterized in that: The following steps are involved: Step 1: Construct radar signal samples. According to the characteristics of modern radar signals, use Python to generate several pulse sequence samples with typical radar signal diversity. Step 2: Perform full clustering on the first 1000 pulse sequences to obtain the initial cluster center list and category number; Step 3: Based on the pre-obtained initial cluster center list and category serial number, the remaining data is sorted according to the time beat by single-sample inverse distance weighted similarity in real-time signal sorting; Step 4: Real-time verification of the sorting results of step 3; Step 5: Repeat steps 3 and 4 until all data are sorted; Step 6: Repeat steps 3, 4, and 5 using the incremental Leader algorithm and the non-incremental DBSCAN and BGMM algorithms for comparative experimental verification.

2. The radar pulse sorting method based on inverse distance weighted online clustering according to claim 1 is characterized in that: In the step 1, the five-dimensional features of the radar signal sample including the radar pulse description word are first determined: TOA: Time of arrival, which indicates the time it takes for a pulse to reach the receiver, in seconds; CF: Center frequency, indicating the frequency center of the pulse, in megahertz; PW: Pulse width, which indicates the duration of the pulse in microseconds; PA: Amplitude, which indicates the strength of the pulse, in volts; DOA: Angle of arrival, indicating the direction of the pulse, in degrees; 1) Determine the arrival time based on the three modulation modes of radar, For a simple pulse radar, the first-order difference of arrival times is a constant: ΔTOA(t)=C; For frequency modulated continuous wave radar, the first-order difference of arrival time is a linear function of time: ΔTOA(t)=t; For nonlinear FM radar, the first-order difference of arrival time is a complex function of time: ΔTOA(t)=f(t); 2) The angle of arrival varies from 0 to 360 degrees The continuously varying angle of arrival is simulated by taking a combination of intercepted sine, cosine, polynomial, exponential, and linear functions: By giving different value threshold ranges of a, b, c, d, e, f, g and DOA(t), the changes in the arrival angles of different radar signals in the real electromagnetic world are simulated; 3) The center frequency range is from 1 GHz to 15 GHz, and the stable center frequency is described by a combination of a constant function and Gaussian noise: 4) The pulse width ranges from 0.05 microseconds to 35 microseconds, and the stable center frequency is described by a combination of a constant function and Gaussian noise: 5) The range of pulse amplitude variation. According to the definition of pulse amplitude, P is specified as a random number within a certain range: According to the above rules, radar signal sources with different pulse numbers, pulse widths, center frequencies, arrival angles, and amplitudes are simulated and their arrival times are plotted on a plane, with each point representing a pulse.

3. The radar pulse sorting method based on inverse distance weighted online clustering according to claim 1 is characterized in that: In the step 2, the full amount of pre-clustering using the first 1000 radar signal samples specifically includes: the following: For the first 1000 pulse trains x n (n=1,2,…,K), perform full clustering and obtain the initial cluster center list x i And category number i (1,2,…,C).

4. The radar pulse sorting method based on inverse distance weighted online clustering according to claim 1 is characterized in that: In the step 3, according to the pre-obtained initial cluster center list and category serial number, the remaining data is subjected to real-time signal sorting of single-sample inverse distance weighted similarity according to the time rhythm, which specifically includes the following steps: Given a pulse query point x j , a list of cluster centers of known classes x i , i = 1, 2, ..., C, where C is the maximum index of all known clusters and a pulse stream T: T={(CF1,DOA1),(CF2,DOA2),…,(DF N ,DOA N )}, Where X j =(CF j ,DOA j ) is the feature vector of the instance, j = 1, 2, ..., N; Step 3.1: Calculate x j With the currently known cluster center x i The CF and DOA distance between: Step 3.2: Select the k cluster centers closest to each pulse feature, k is n / 2, n is the current number of clusters, recorded as: N k1 (CF j )=min k distance CF ; N k2 (PRAYER j )=me k distance DOA ; Step 3.3: Based on the inverse distance weighted similarity, calculate each x i In N k (x q ) in the reciprocal s(x j ,x i ), Where ∈ is a small positive number to avoid the situation where the denominator is zero. The calculated distance is converted into a reliable probability threshold. i In N k (x j ) is normalized so that the probability of each set of features sums to 1, which is provided for subsequent sorting, that is, Step 3.4: Get each x i The weight w(x j ,x i ) after N k1 (CF j )∩N k2 (DOA j ) perform weighted sum of similarities to obtain the total weight W K =W(N k1 (CF j )∩N k2 (DOA j )), according to W K Make a judgment and get x j Category y q ,Right now Among them, n represents the number of current categories, and threshold represents the threshold set by the similarity function.

5. The radar pulse sorting method based on inverse distance weighted online clustering according to claim 1, characterized in that: In the step 4, the sorting results in step 3 are verified in real time, and the experimental results are measured from three aspects: ARI, AMI, and ACC; 1) ARI algorithm. The value range of ARI is from -1 to 1. The larger the value, the better the clustering result. The calculation formula of ARI is: in, It represents the number of combinations of selecting k elements from n elements, n ij Indicates the number of samples contained in the i-th category in the clustering result and the j-th category in the true category, a i Indicates the number of samples in the i-th category in the clustering results, b j represents the number of samples of the jth class in the true category, and n represents the total number of samples; 2) AMI algorithm. The value range of AMI is 0 to 1. The larger the value, the better the clustering result. The calculation formula of AMI is: Among them, MI represents the mutual information between the clustering result and the true category, H(U) and H(V) represent the entropy of the clustering result and the true category respectively, and E[MI] represents the expected mutual information between the clustering result and the true category when randomly assigned; 3) The ACC algorithm is based on the Hungarian algorithm and is used to solve the maximum weight matching problem of bipartite graphs. The idea of ​​the ACC algorithm is to find a maximum weight matching so that each category in the clustering result corresponds to a category in the real category: Where w is a D×D matrix, D=max(max(y),max(Y))+1, and L is the Hungarian matching algorithm; Finally, we get y for each time step in step 3. q Compare with the actual label and calculate the ARI, AMI, ACC scores, and take their average as the metric, as follows: The above score represents the average value of ARI, AMI and ACC scores.

Images