CN108345864B - Random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering - Google Patents
Random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering Download PDFInfo
- Publication number
- CN108345864B CN108345864B CN201810182612.5A CN201810182612A CN108345864B CN 108345864 B CN108345864 B CN 108345864B CN 201810182612 A CN201810182612 A CN 201810182612A CN 108345864 B CN108345864 B CN 108345864B
- Authority
- CN
- China
- Prior art keywords
- radiation source
- radar radiation
- random set
- random
- similarity
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Signal Processing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a random aggregation type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering, which is characterized in that a similarity distance matrix is constructed aiming at signal parameter data of a random aggregation type airborne radar radiation source target, the similarity between every two random aggregation type measured values is compared, random aggregation type clustering is iteratively fused on the basis of a hierarchical clustering method on the basis of the similarity distance matrix until clusters meeting a random aggregation distance threshold cannot be found, the support degree and the random aggregation value of each new cluster are iteratively calculated at the same time, and finally, the random aggregation value corresponding to the clusters meeting the support degree threshold is mined and used as high-frequency mode output. The invention has the advantages that: (1) a similarity comparison method for random set type radar radiation source signal parameters is provided; (2) based on the similarity distance matrix, a high-frequency signal parameter mode can be further mined; (3) the calculation cost is low, and the implementation method is engineered.
Description
Technical Field
The invention relates to the field of analysis and processing of random set type radar radiation source signal parameter data, in particular to a weighted clustering-based random set type radar radiation source signal parameter high-frequency mode mining method.
Background
As is known, the signal characteristics of radar radiation sources are increasing nowadays, and airborne radar radiation sources exhibit, in addition to the traditional type, a random set of signal parameters consisting of several random measured values. The conventional signal feature types are mostly continuous type and discrete type. Nowadays, random collective type signal characteristics also become an important type of radar radiation source data. This is because, as the complexity of the radar radiation source increases, the variety and form of the signal characteristics also becomes more and more diverse. Furthermore, due to the advancement of signal characteristic measurement techniques and the influence of the measurement environment (e.g., noise), the measured value of the signal characteristic is usually no longer a fixed value, but rather a set of several random values. The existing radar radiation source signal characteristic analysis method and the existing radar radiation source type identification method almost aim at the signal characteristics of fixed values, and cannot be specially used for describing and analyzing random set type signal characteristics. With the improvement of science and technology, the data resource of the random set type radar radiation source also increases, and a corresponding processing method is urgently needed.
Disclosure of Invention
The purpose of the invention is as follows: aiming at the problems, the invention provides a random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering.
The technical scheme is as follows: in order to realize the purpose of the invention, the technical scheme adopted by the invention is as follows: a random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering is characterized in that a similarity distance matrix is constructed according to signal parameter data of a random set type airborne radar radiation source target, similarity between two random set type measurement values is compared, random set type clustering is fused iteratively on the basis of a hierarchical clustering method on the basis of the similarity distance matrix until clusters meeting a random set distance threshold cannot be found, meanwhile, the support degree and the random set value of each new cluster are calculated iteratively, and finally, the random set value corresponding to the clusters meeting the support degree threshold is mined to serve as high-frequency mode output.
A random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering comprises the following steps:
(1) radar radiation source target according to random set size SORDSorting from big to small;
(2) initializing a similarity distance matrix with the size of n multiplied by n;
(3) setting a target index s of a current radar radiation source;
(4) setting a target index t of a radar radiation source to be compared;
(5) judging the target index to be compared, if t is not more than n and S (ORD [ t ])/S (ORD [ S ]) is more than or equal to n, continuing the step 6, and if t is more than or equal to n, continuing the step 8;
(6) updating the current target index value;
(7) judging the current target index, if s is more than or equal to n, continuing the step 10, otherwise, returning to the step 4;
(8) calculating a similarity distance dist between radar radiation sources and updating a similarity matrix;
(9) updating the radar radiation source index to be compared, and returning to the step 5;
(10) constructing n initial clusters;
(11) judging two clusters C with minimum similarity distance in distance matrixuAnd Cv(u<v) whether dist (C) is satisfiedu,Cv) If yes, continuing to step 12, if not, jumping to step 14;
(12) merged cluster CuAnd CvAs a new cluster CwAnd updating;
(13) updating the distance matrix based on the new cluster, defining the distance between the new cluster and other clusters as the minimum value of the distances between the original two clusters forming the new cluster and other clusters, and returning to the step 11;
(14) and finding out clusters with all support degrees meeting the threshold minsup, and outputting a corresponding random set as a high-frequency mode of the radar parameter.
Further, the random set distance threshold is used for comparing the similarity between the random measurement value sets, and the support threshold min is used for judging whether the signal parameter mode belongs to the high frequency.
Further, the measured value of the airborne radar radiation source signal parameter is a random set consisting of a plurality of random measured values.
Has the advantages that: the invention has the advantages that: (1) a similarity comparison method for random set type radar radiation source signal parameters is provided; (2) based on the similarity distance matrix, a high-frequency signal parameter mode can be further mined; (3) the calculation cost is low, and the implementation method is engineered.
Drawings
FIG. 1 is a flow chart of the operation of the present invention.
Detailed Description
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
The method comprises the steps of constructing a similarity distance matrix aiming at signal parameter data of a random set type airborne radar radiation source target, comparing the similarity between every two random set type measured values, fusing random set type clusters in an iterative mode on the basis of a hierarchical clustering method on the basis of the similarity distance matrix until a cluster meeting a random set distance threshold cannot be found, calculating the support degree and the random set value of each new cluster in an iterative mode, and finally mining the random set value corresponding to the cluster meeting the support degree threshold as a high-frequency mode to be output.
Assume that the data set of airborne radar radiation source signal parameters consists of a number of airborne radar radiation source targets { o1, o2, …, on }, each of which consists of a random set of measurements of the signal parameters a (the measurements are a set consisting of a number of random numbers). In particular, a radar radiation source target oiThe measured values on the signal parameter a may be represented as a random set Si={Sip}pWherein S isipRepresenting an object oiOne of several random measurements on the signal parameter a. In addition, assume that the similarity threshold of the random measurement value is, the distance threshold of the random set is (between 0 and 1), the support threshold is minsup, and the minimum weight threshold of the measurement value is minw.
As shown in fig. 1, the method of the present invention specifically includes the following steps:
(1) target sorting;
sequencing radar radiation source targets from large to small according to the size of a random set, and recording as ORD (order of arrival) so as to enable S to beORD[1]≥SORD[2]≥…≥SORD[n]Wherein S isORD[1]To representThe random set size of the first radar source target in the ORD sequence, and so on.
(2) Initializing a similarity distance matrix;
the initialized similarity distance matrix DistAlrray has the size of n multiplied by n, the row index value corresponds to the index value of the ORD sequence, the diagonal line and the lower unit take values of null, and the upper triangular unit takes a value of infinity.
(3) Setting a current target;
and setting the current radar radiation source target index s to be 1.
(4) Setting a target to be compared;
and setting the target index t of the radar radiation source to be compared as s + 1.
(5) Judging a target to be compared;
and judging the target index value t to be compared, if t is not more than or equal to n and size (ORD [ t ])/size (ORD [ s ]) is more than or equal to n, continuing the step 6, and if t is more than or equal to n, continuing the step 8.
(6) Modifying the current target;
and updating the current target index value s to be s + 1.
(7) Judging a current target;
and if the current target index value s is larger than or equal to n, jumping to the step 10, otherwise, returning to the step 4.
(8) Calculating a similarity distance;
calculating the similarity distance between the radar radiation sources ORD [ s ] and ORD [ t ] and updating the similarity matrix DistAlrray, wherein the calculation method of the similarity distance value comprises the following steps:
wherein S isORD[s]And SORD[t]Respectively representing radar radiation sources ORD [ s ]]And ORD [ t]Random set size of (a), MatchSet (o)ORD[s],oORD[t]) Representing radar radiation sources ORD s]And ORD [ t]And (3) a matching set between the random sets, wherein the matching set is defined as a set consisting of measured value matching pairs which are respectively from the two random sets and have measured value difference not exceeding a measured value similarity threshold, and the matching set comprises the following steps:
MatchSet(oi,oj)={<Sip,Sjq>|Sip∈Si,Sjq∈Sj,|Sip-Sjq|≤} (2)
(9) updating the target to be compared;
and updating the index t +1 of the radar radiation source to be compared, and returning to the step 5.
(10) Initializing a cluster;
n initial clusters are constructed according to n random sets of radar radiation sources, and the random set of each cluster is respectively initialized to be the random set corresponding to the radar radiation sources, namely C1=S1,C2=S2,......,Cn=Sn(ii) a Each cluster CkIs initialized to 1 and denoted as Wk1=Wk2=……=Wk|Ck|1, where | Ck | represents cluster CkThe random set size of (a); the support degree of each cluster is initialized to 1, Sup1=Sup2=…=Supn1 is ═ 1; membership set Memset of each clusterkInitialisation to a set of corresponding radar radiation source targets, i.e. Memsetk={ok}。
(11) Judging clustering;
two clusters C with minimum similarity distance in the distance matrix DistAlrray are judgeduAnd Cv(u<v) whether dist (C) is satisfiedu,Cv) ≦ continue step 12 if satisfied, and jump to step 14 if not satisfied.
(12) Clustering and fusing;
merged cluster CuAnd CvAs a new cluster CwCluster of CwThe random set elements and their weights are updated according to equations (3) and (4), respectively:
updating the support degree of the new cluster to Supw=Supu+Supv(ii) a Member set Memset of new clusterkUpdated to Memsetw=Memsetu+Memsetv。
(13) Updating the distance matrix;
and updating the distance matrix DistAlrray based on the new cluster, defining the distance between the new cluster and other clusters as the minimum value of the distances between the original two clusters forming the new cluster and other clusters, and returning to the step 11.
(14) Outputting in a high-frequency mode;
and finding out clusters with all support degrees meeting the threshold minsup, and outputting a corresponding random set as a high-frequency mode of the radar parameter.
The method can enhance the analysis capability of the random set type signal characteristics and can better complete the radar radiation source identification task. The method for mining the high-frequency mode of the radiation source signal parameters of the random aggregation type airborne radar based on the weighted clustering is described by an example.
Assuming that 3 samples of a type of radar radiation source are detected, the PRI signal for each sample is a random set of signal parameters, whose values are shown in table 1, in MHz.
TABLE 1
Sample(s) | PRI Signal parameters (MHz) |
o1 | {60,69,85,100} |
o2 | {61,70,84,99} |
o3 | {70,86,101} |
The similarity threshold value of the random measurement value is 0.1, the distance threshold value of the random set is 0.6, the support degree meets the threshold minsup value of 2, the minimum weight threshold minw of the measurement value is 0.3, and the method comprises the following steps:
step 1, radar radiation source targets are sorted from large to small according to random set size, and ORD (ORD-o)1<o2<o3;
TABLE 2
Sample(s) | o1 | o2 | o3 |
o1 | - | ∞ | ∞ |
o2 | - | - | ∞ |
o3 | - | - | - |
TABLE 3
Sample(s) | o1 | o2 | o3 |
o1 | - | 0 | ∞ |
o2 | - | - | ∞ |
o3 | - | - | - |
TABLE 4
Sample(s) | o1 | o2 | o3 |
o1 | - | 0 | 0.25 |
o2 | - | - | ∞ |
o3 | - | - | - |
step 6, modifying the current target, and updating the index value s + 1-2 of the current target;
TABLE 5
Sample(s) | C1 | C2 | C3 |
C1 | - | 0 | 0.25 |
C2 | - | - | 0.25 |
C3 | - | - | - |
step 6, modifying the current target, and updating the index value s + 1-3 of the current target;
TABLE 6
TABLE 7
Sample(s) | PRI Signal parameters (MHz) | Sup | MemSet |
C1’ | {60.5(1),69.5(1),85(1),100(1)} | 2 | {o1,o2} |
C3 | {70(1),86(1),101(1)} | 1 | {o3} |
TABLE 8
Clustering | C1 | C3 |
C1 | - | 0.25 |
C3 | - | - |
TABLE 9
Sample(s) | PRI Signal parameters (MHz) | Sup | MemSet |
C1” | {60.5(0.67),69.67(1),85.3(1),100.3(1)} | 3 | {o1,o2,o3} |
watch 10
Sample(s) | C1” |
C1” | - |
The research result of the invention is beneficial to improving the analysis capability of the signal characteristics of the random set type radar radiation source and further improving the type identification capability of the radar radiation source.
The research work of the present invention was funded by the national science foundation (No. 61771177).
Claims (1)
1. A random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering is characterized in that: the method comprises the following steps:
(1) aiming at signal parameter data of a random set type airborne radar radiation source target, a measured value of the airborne radar radiation source signal parameter is a random set formed by a plurality of random measured values, and a similarity distance matrix is constructed;
(1.1) radar radiation source targets are grouped into random set size SORDSorting from big to small;
(1.2) initializing a similarity distance matrix with the size of n multiplied by n;
(1.3) setting a current radar radiation source target index s;
(1.4) setting a target index t of a radar radiation source to be compared;
(1.5) judging the target index to be compared, if t is not more than n and S (ORD [ t ])/S (ORD [ S ]) is more than or equal to n, continuing the step (1.6), and if t is more than or equal to n, continuing the step (1.8); the random set distance threshold is used for comparing the similarity between the random measurement value sets;
(1.6) updating the current target index value;
(1.7) judging the current target index, if s is more than or equal to n, continuing the step (3.1), and if not, returning to the step (1.4);
(1.8) calculating a similarity distance dist between radar radiation sources and updating a similarity matrix;
(1.9) updating the radar radiation source index to be compared, and returning to the step (1.5);
(2) comparing the similarity between two random set type measured values;
(3) iteratively fusing random set type clusters on the basis of a hierarchical clustering method on the basis of a similarity distance matrix until a cluster meeting a random set distance threshold cannot be found;
(3.1) constructing n initial clusters;
(3.2) judging two clusters C with minimum similarity distance in the distance matrixuAnd Cv(u<v) whether dist (C) is satisfiedu,Cv) ≦ if step (3.3) is satisfied, if not, jump to step (3.4);
(3.3) Merge clustering CuAnd CvAs a new cluster CwAnd updating;
(3.4) updating the distance matrix based on the new cluster, defining the distance between the new cluster and other clusters as the minimum value of the distances between the original two clusters forming the new cluster and other clusters, and returning to the step (3.2);
(4) meanwhile, the support degree and the random set value of each new cluster are calculated in an iterative mode, and finally, the random set value corresponding to the cluster meeting the threshold of the support degree is mined out and is used as the high-frequency mode output; and finding out clusters with all support degrees meeting a threshold minsup, judging whether the signal parameter mode belongs to high frequency, and outputting a corresponding random set as a high frequency mode of the radar parameter.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810182612.5A CN108345864B (en) | 2018-03-06 | 2018-03-06 | Random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810182612.5A CN108345864B (en) | 2018-03-06 | 2018-03-06 | Random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108345864A CN108345864A (en) | 2018-07-31 |
CN108345864B true CN108345864B (en) | 2020-09-08 |
Family
ID=62957435
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810182612.5A Active CN108345864B (en) | 2018-03-06 | 2018-03-06 | Random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108345864B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110135450B (en) * | 2019-03-26 | 2020-06-23 | 中电莱斯信息系统有限公司 | Hot spot path analysis method based on density clustering |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2002014896A1 (en) * | 2000-08-11 | 2002-02-21 | Alenia Marconi Systems Limited | Method of interference suppression in a radar system |
CN102749616A (en) * | 2012-06-29 | 2012-10-24 | 北京市遥感信息研究所 | Fuzzy-clustering-based Aegis system signal sorting method |
CN102930255A (en) * | 2012-11-13 | 2013-02-13 | 中国电子科技集团公司第二十八研究所 | Signal analysis method based on radiation source category pairs |
CN104794431A (en) * | 2015-03-25 | 2015-07-22 | 中国电子科技集团公司第二十八研究所 | Radar radiation source pulse-to-pulse mode excavation method based on fuzzy matching |
CN105005029A (en) * | 2015-07-17 | 2015-10-28 | 哈尔滨工程大学 | Multi-mode radar signal sorting method based on data field hierarchical clustering |
CN106056098A (en) * | 2016-06-23 | 2016-10-26 | 哈尔滨工业大学 | Pulse signal cluster sorting method based on class merging |
CN106405518A (en) * | 2016-12-07 | 2017-02-15 | 中国船舶重工集团公司第七二四研究所 | Complex system radar signal grade correlating, clustering and sorting method |
-
2018
- 2018-03-06 CN CN201810182612.5A patent/CN108345864B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2002014896A1 (en) * | 2000-08-11 | 2002-02-21 | Alenia Marconi Systems Limited | Method of interference suppression in a radar system |
CN102749616A (en) * | 2012-06-29 | 2012-10-24 | 北京市遥感信息研究所 | Fuzzy-clustering-based Aegis system signal sorting method |
CN102930255A (en) * | 2012-11-13 | 2013-02-13 | 中国电子科技集团公司第二十八研究所 | Signal analysis method based on radiation source category pairs |
CN104794431A (en) * | 2015-03-25 | 2015-07-22 | 中国电子科技集团公司第二十八研究所 | Radar radiation source pulse-to-pulse mode excavation method based on fuzzy matching |
CN105005029A (en) * | 2015-07-17 | 2015-10-28 | 哈尔滨工程大学 | Multi-mode radar signal sorting method based on data field hierarchical clustering |
CN106056098A (en) * | 2016-06-23 | 2016-10-26 | 哈尔滨工业大学 | Pulse signal cluster sorting method based on class merging |
CN106405518A (en) * | 2016-12-07 | 2017-02-15 | 中国船舶重工集团公司第七二四研究所 | Complex system radar signal grade correlating, clustering and sorting method |
Non-Patent Citations (3)
Title |
---|
Radar Emission Sources Identification Based on Hierarchical Agglomerative Clustering for Large Data Sets;Janusz Dudczyk;《Hindawi Publishing Corporation Journal of Sensors》;20161231;第1-10页第3-4节 * |
基于PRI熵的雷达信号聚类方法研究;孙盼杰 等;《电子信息对抗技术》;20080131;第23卷(第1期);第22-25页 * |
基于距离度量方法的雷达辐射源信号识别研究;郭强 等;《中国电子科学研究学报》;20101231;第5卷(第6期);第609-611页 * |
Also Published As
Publication number | Publication date |
---|---|
CN108345864A (en) | 2018-07-31 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109409400A (en) | Merge density peaks clustering method, image segmentation system based on k nearest neighbor and multiclass | |
CN112526513B (en) | Millimeter wave radar environment map construction method and device based on clustering algorithm | |
CN113160274A (en) | Improved deep sort target detection tracking method based on YOLOv4 | |
CN106951915B (en) | One-dimensional range profile multi-classifier fusion recognition method based on category confidence | |
CN105960021A (en) | Improved position fingerprint indoor positioning method | |
CN110083665A (en) | Data classification method based on the detection of improved local outlier factor | |
CN113344019A (en) | K-means algorithm for improving decision value selection initial clustering center | |
Wishart | K-means clustering with outlier detection, mixed variables and missing values | |
CN111062425B (en) | Unbalanced data set processing method based on C-K-SMOTE algorithm | |
Zheng et al. | Improvement of grayscale image 2D maximum entropy threshold segmentation method | |
CN116539167B (en) | High-voltage power supply working temperature distribution data analysis method | |
CN112508963B (en) | SAR image segmentation method based on fuzzy C-means clustering | |
CN113435108B (en) | Battlefield target grouping method based on improved whale optimization algorithm | |
CN108345864B (en) | Random set type radar radiation source signal parameter high-frequency mode mining method based on weighted clustering | |
Velden et al. | Resolving author name homonymy to improve resolution of structures in co-author networks | |
CN111125470A (en) | Method for improving abnormal data mining and screening | |
CN107564008A (en) | Rapid SAR image segmentation method based on crucial pixel fuzzy clustering | |
CN114241233A (en) | Nonlinear class group sparse representation true and false target one-dimensional range profile identification method | |
CN110988856B (en) | Target detection trace agglomeration algorithm based on density clustering | |
CN113128584A (en) | Mode-level unsupervised sorting method of multifunctional radar pulse sequence | |
Aslam et al. | A comprehensive survey on optimization techniques in image processing | |
Liu et al. | A weight-incorporated similarity-based clustering ensemble method | |
CN115292303A (en) | Data processing method and device | |
CN110377798B (en) | Outlier detection method based on angle entropy | |
CN109800384B (en) | Basic probability assignment calculation method based on rough set information decision table |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |