CN102663430A - Object clustering method in situation assessment - Google Patents

Object clustering method in situation assessment Download PDF

Info

Publication number
CN102663430A
CN102663430A CN2012101128499A CN201210112849A CN102663430A CN 102663430 A CN102663430 A CN 102663430A CN 2012101128499 A CN2012101128499 A CN 2012101128499A CN 201210112849 A CN201210112849 A CN 201210112849A CN 102663430 A CN102663430 A CN 102663430A
Authority
CN
China
Prior art keywords
bunch
mini
target
attribute
threshold value
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.)
Granted
Application number
CN2012101128499A
Other languages
Chinese (zh)
Other versions
CN102663430B (en
Inventor
覃征
江子能
卢正才
张海生
李凤翔
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to CN2012101128499A priority Critical patent/CN102663430B/en
Publication of CN102663430A publication Critical patent/CN102663430A/en
Application granted granted Critical
Publication of CN102663430B publication Critical patent/CN102663430B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

An object clustering method in situation assessment includes the steps: firstly, generating a property priority list and a clustering factor matrix; secondly, segmenting objects in a battlefield into clusters according to the first priority selected from the property priority list and deleting the selected priority from the property priority list; thirdly, judging whether the object quantity in the clusters is greater than the preset threshold value of the object quantity or not, if so, then continuously selecting a priority from the property priority list to segment the clusters, if not or the property priority list is empty, then performing clustering analysis on object data in the bottom clusters to generate mini-clusters according to the object location relationship; and finally, utilizing a fusion algorithm to fuse the mini-clusters, and then acquiring an object group. The object clustering method in the situation assessment overcomes defects in the object clustering and solves the key problem that accuracy and speed of object clustering are low, so that the battlefield situation assessment and policies made by the commander in the battlefield can be effectively supported.

Description

Target grouping method in a kind of situation assessment
Technical field
The present invention relates to battlefield target grouping method, be specifically related to target grouping method in a kind of situation assessment.
Background technology
Owing to the variation of expression forms of information in the battlefield, caused the huge property of complicacy, uncertainty and information content between the information; And the battlefield commander can't handle the information of flood tide like this in real time, is difficult to from " noise " of flood tide, obtain Useful Information, and this will cause the appearance of " war is dazzling " and " Fog of War ".In order to solve this two problems, people have proposed the situation assessment.Wherein to hive off be one of core technology in the situation assessment to target, and the speed that target is hived off and accuracy have a strong impact on battlefield commander's decision-making, thereby target is hived off and become the hot issue of being badly in need of solution.
Summary of the invention
In order to overcome the deficiency of above-mentioned prior art, the object of the present invention is to provide target grouping method in a kind of situation assessment, satisfy the demand under the complex environment of battlefield, have the characteristic of accurate of hiving off.
To achieve these goals, the technical scheme of the present invention's employing is:
Target grouping method in a kind of situation assessment comprises the steps:
Step 1; Generate attribute priority chained list and clustering factor matrix; Attribute priority chained list is meant the chained list of arranging sequentially of selected objective attribute target attribute; The clustering factor matrix is the matrix that the clustering factor between different attribute forms, clustering factor be meant a certain target on the single attribute or on a plurality of attributes with the polymerization expectation value of other target;
Step 2 is selected first attribute that the target in the battlefield is cut apart and is generated type bunch in the dependency priority chained list, and with deleting in the selected attribute dependency priority chained list;
Whether step 3, the destination number in judging type bunch greater than predefined destination number threshold value, minimum destination number in destination number threshold value type of being meant bunch;
If step 4 greater than the destination number threshold value, then continues in the dependency priority chained list to select attribute that class bunch is cut apart, and be that destination number in the empty perhaps bottom class bunch is less than the destination number threshold value until attribute priority chained list;
Step 5 if be empty less than threshold value or attribute priority chained list, then carried out cluster analysis according to the position relation of target to the target data in the bottom class bunch and is generated mini bunch;
Step 6 adopts blending algorithm that said mini bunch is merged, and obtains target complex; Corresponding clustering factor need in the clustering factor matrix, be inquired about during fusion, if clustering factor is greater than the clustering factor threshold value; Then can merge; If clustering factor less than the clustering factor threshold value, then can not merge, wherein the clustering factor threshold value is meant the minimum polymerization expectation value between two dissimilar mini bunch.
Select the attribute of limit priority to carry out recurrence in the said step 2 in the dependency priority chained list and cut apart, the said concrete grammar of cutting apart is:
For the attribute of nonumeric type, different types is divided into different classes bunch;
For the attribute of numeric type, cutting procedure is described below: at first at class bunch LC iIn target chained list Objlist in select a target data O at random sGenerate subclass bunch LC as seed i_ 1, and this subclass bunch LC i_ 1 joins above-mentioned type of bunch LC iSubclass bunch chained list LClist in, and then a type bunch LC iIn other target datas O iWith seed O sCompare, if the attribute P that these two targets are being selected iOn difference less than pre-determined threshold value, then with this target data O iAdd the affiliated class bunch LC of seed iIn _ 1; If its difference is greater than threshold value, then with this destination object as new seed O s' the new class bunch LC of generation i_ n; As all target data O iWhen all being assigned to subclass bunch, select the attribute of time priority again, the above-mentioned algorithm of recurrence is until all properties P iBe selected or divided type bunch in the target data that comprised be less than predetermined threshold value, attribute of every selection is cut apart the data in the class bunch and is equivalent to the LCMC tree and increases one deck.
The destination number threshold value generally gets 3 with reference to the DBScan algorithm in the said step 3.
When attribute priority chain table be not empty and type bunch in destination number greater than the destination number threshold value, carry out step 2 in the step 4 continued.
The step that generates mini bunch in the step 5 is: computation core radius C oreR at first select a point as seed, and to put with this is that the center of circle is with CoreR *ScaleFactor is that radius generates one mini bunch, and wherein ScaleFactor is the adjustment factor, and initial value is made as 1.0; Dynamically adjust at feedback stage, whether other target datas in relatively type bunch are positioned at mini bunch then, if; In then adding this mini bunch to this target data, if not, then generate new mini bunch to this target data as new seed; When all data all got in mini bunch, process finished.
Mini bunch of fusion rule is divided into three kinds of situation according to the distance between mini bunch in the step 6:
Distance between two mini bunch of first kind of situation is less than zero, and then this two mini bunch can merge;
Distance between two mini bunch of second kind of situation is greater than the ultimate range between the predefined target complex, can not merge this two mini bunch of at this moment;
Distance between two mini bunch of the third situation is between zero-sum ultimate range, and the point midway between calculating two mini bunch earlier this moment is the center of circle with this point midway then; With in two mini bunch bigger one be radius; How many target datas circle of paintings has drop in the new circle in calculating two mini bunch, if target data falls within the new circle more than or equal to 3; Then this two mini bunch belongs to a crowd, can merge.
Compared with prior art; Through the target grouping method in the situation of battlefield assessment of the present invention; Solve deficiency and key issue during target is hived off, promptly improved the speed that degree of accuracy that target hives off and target are hived off, therefore; Can support situation of battlefield assessment and battlefield commander's decision-making effectively, have practicality.
Description of drawings
Fig. 1 is the process flow diagram of target grouping method in the situation of battlefield assessment according to the invention.
Fig. 2 is the simple case of target grouping method in the situation of battlefield assessment according to the invention.
Fig. 3 is described mini bunch of crossing situation.
Fig. 4 be described mini bunch of non-intersect but its distance less than bunch between the situation of ultimate range.
Between described mini bunch of Fig. 5 distance greater than bunch between the situation of ultimate range.
Fig. 6 is an example of mini bunch of fusion.
Embodiment
Below in conjunction with accompanying drawing and embodiment the present invention is explained further details.
The target grouping method is a kind of target grouping method based on the hierarchical clustering technology in the situation of battlefield assessment of the present invention; Degree of accuracy that it hives off and speed are that the tracking that has earlier is all incomparable, support situation of battlefield assessment and battlefield commander's decision-making effectively through the method for the invention.
With reference to shown in Figure 1, target is hived off and is comprised the steps: in the situation of battlefield assessment according to the invention
At first, generate attribute priority chained list and clustering factor matrix.Attribute priority chained list is meant the chained list at the sequencing of selected objective attribute target attribute.Suppose to have two attribute V of attribute priority chained list iAnd A i, V iPriority greater than A iPriority, can the preferential V of selection in the process that target hives off following carrying out i, just select A then iThe clustering factor matrix is the matrix that the clustering factor between different attribute forms.Clustering factor be meant a certain target on the single attribute or on a plurality of attributes with the polymerization expectation value of other target.This is a predetermined value in algorithm, can derive from empirical value or prior method acquisition through machine learning.For example: the data in the battlefield have " type (type) " attribute; Suppose that wherein the type of a target OBJ_1 is " tank "; And the type of another target OBJ_2 is " opportunity of combat ", and opportunity of combat and tank hardly maybe be in same crowds, thereby on " type " this attribute; The polymerization expectation value of " tank " and " aircraft ", promptly clustering factor approaches 0 (step 101).
Next is that target data is cut apart, and the purpose of " cutting apart " is bunch to be divided into less class bunch according to the class that different attributes will be bigger, makes that all data contacts of cutting apart in the back type bunch are tightr, and more trend belongs to same target complex.Data in the battlefield be divided into numeric type with nonumeric type, need separate processes.For example " type " attribute of target data is non-numeric type in the battlefield, and " speed " attribute is a numeric type.
Select the attribute P of limit priority in the partitioning algorithm dependency priority chained list (PPL) i(each attribute is selected once, after the selection with this attribute deletion) carried out recurrence and cut apart (step 102,103,104 and 105).If attribute is non-numeric type, " type " attribute for example, different " type " are divided into different class bunch.For the attribute of numeric type, cutting procedure is described below: at first at class bunch LC iIn target chained list Objlist in select a target data O at random sGenerate subclass bunch LC as seed i_ 1 and this subclass bunch LC i_ 1 joins above-mentioned type of bunch LC iSubclass bunch chained list LClist in, and then a type bunch LC iIn other target datas O iWith seed O sCompare, if the attribute P that these two targets are being selected iOn difference less than pre-determined threshold value, then with this target data O iAdd the affiliated class bunch LC of seed iIn _ 1; If its difference is greater than threshold value, then with this destination object as new seed O s' the new class bunch LC of generation i_ n.As all target data O iWhen all being assigned to subclass bunch, select the attribute of time priority again, the above-mentioned algorithm of recurrence is until all properties P iBe selected or divided type bunch in the target data that comprised be less than predetermined threshold value.Attribute of every selection is cut apart the data in the class bunch and is equivalent to the LCMC tree and increases one deck.
Shown in (1-1) and formula (1-2), O iAnd O jRepresent two targets, and T iBe O iA property value, T jBe O jA property value; Function f (O i, O j) expression target O iWhether can be assigned in same type bunch.
Nonumeric type data:
f ( O i , O j ) = True T i = T j False T i ! = T j (formula 1-1)
The numeric type data:
f ( O i , O j ) = True ( T i - T j ) / T j < = &mu; False ( T i - T j ) / T j > &mu; (formula 1-2)
When attribute priority chain table for empty or type bunch in destination number when reaching institute's pre-set threshold (step 105); Data all can be divided into less class bunch; But the target data in these types bunch just expectation belongs to certain a group; Still be not the part of a crowd or certain a group, thereby need the target data in the class bunch be divided into mini bunch once more, and the data in mini bunch are the parts that constitute a crowd or certain a group.
Its main algorithm that generates mini bunch is computation core radius C oreR at first, selects a point as seed, and to put with this be that the center of circle is with CoreR *ScaleFactor is that (wherein ScaleFactor is the adjustment factor in one mini bunch of radius generation; Initial value is made as 1.0, can dynamically adjust at feedback stage), whether be positioned at mini bunch to other target datas in relatively type bunch then; If; In then adding this mini bunch to this target data, if not, then generate new mini bunch to this target data as new seed.So carry out, when all data all got in mini bunch, process finished (step 106).
After cutting procedure is accomplished, be mini bunch fusion, comprise LCMC tree middle period subclass bunch inside mini bunch fusion and type bunch between mini bunch fusion, its main algorithm thought is the distance (step 107) between two mini bunch of the comparison.
Distance definition between mini bunch is suc as formula shown in (1-3).
Dis (MC i, MC j)=dist (O i, O j)-R i-R j(formula 1-3)
Dist ( O i , O j ) = ( x i - x j ) 2 + ( y i - y j ) 2 + ( z i - z j ) 2 (formula 1-4)
Shown in (3-9) and formula (3-10), MC iAnd MC jRepresent two mini bunch, and O iAnd O jRepresent mini bunch initial point; Dist (MC i, MC j) two MC of expression i, MC jBetween distance, and dist (O i, O j) distance between two initial points of expression.x i, y i, z iBe MC iD coordinates value, and x j, y j, z jBe MC jD coordinates value.
Mainly contain three kinds of situation by distance, the distance between two mini bunch of first kind of situation that is to say that less than zero two mini bunch intersects, and this two mini bunch can permeate mini bunch; Distance between two mini bunch of second kind of situation is greater than the ultimate range between the predefined target complex, and at this moment this two mini bunch can not merge; Distance between two mini bunch of the third situation is between zero-sum ultimate range, and the point midway between at this moment calculating two mini bunch earlier is the center of circle with this point midway then; With in two mini bunch bigger one be radius, how many target datas circle of paintings has drop in the new circle in calculating two mini bunch; If target data falls within the new circle more than or equal to 3; Then we think that this two mini bunch belongs to a crowd, can merge, and vice versa.At fusing stage; We need carry out cut operator to the LCMC tree according to clustering factor; When carrying out mixing operation to mini bunch between the class bunch; At first search the class that need to merge bunch is cut apart attribute at this layer clustering factor; If clustering factor CF is zero perhaps less than the threshold value of input, represent that then LC_1 and LC_2 do not belong to same crowd, do not need to merge; Can reduce amount of calculation like this, also just be equivalent to carry out cut operator in the LCMC tree; If CF then need merge greater than the threshold value of input.Every fusion finishes one deck, and then the LCMC tree will reduce one deck, until obtaining needed result.
Fig. 2 is the simple case of target grouping method in the situation of battlefield assessment according to the invention.Suppose three targets, PL are arranged in the battlefield 1, PL 2, TA 1And in the hypothesis attribute priority chained list three attribute P are arranged 1And P 2Its concrete steps are following:
(1) disposable all data is read in the internal memory, and deposit in the root node;
(2) all data in the root node are pressed attribute P 1Cut apart, generate two class bunch LC (P 1) 1 and LC (P 1) _ 2.
(3) data in two classes bunch are pressed attribute P 2Cut apart, generate three class bunch LC (P 2) _ 1, LC (P 2) _ 2 and LC (P 2) _ 3.
(4) data in the leaf class bunch are divided into mini bunch according to aforementioned algorithm, generate three mini bunch, be respectively MC_1, MC_2 and MC_3.
(5), obtain last result with mini bunch of fusion.
Fig. 3, Fig. 4 and shown in Figure 5 be three kinds of relations of mini bunch.Fig. 3 is described mini bunch of crossing situation.Fig. 4 be described mini bunch non-intersect but its distance less than bunch between ultimate range.Between described mini bunch of Fig. 5 distance greater than bunch between ultimate range.
Mini bunch (MC, Mini-Cluster): mini bunch be meant that cutting procedure is accomplished after, according between the distance distance obtained one group bunch, it is the part of the final target complex that forms, and promptly representes target complex when mini bunch that merges after accomplishing.
MC i={ ID, T i, P i, R i, ObjList i, MCIdList i(formula 1-5)
Formula (1-5) is represented one mini bunch, R iBe mini bunch radius, T iRepresent classification, P iRepresent position (center of circle), R iRepresent radius, ObjList iRepresent this mini bunch of target data that is comprised, MCIdList iStoring Fused mini bunch ID, from MCIdList iIn can find out that this mini bunch is formed by which mini bunch of fusion.
Suppose MC 1And MC 2It is two mini bunch.And supposition
MC 1={ ID 1, T 1, P 1, R 1, ObjList 1, MCIdList 1(formula 1-5)
MC 2={ ID 2, T 2, P 2, R 2, ObjList 2, MCIdList 2(formula 1-6)
If MC 1And MC 2Can merge, then
MC 1+ MC 2={ ID 3, T 3, P 3, R 3, ObjList 3, MCIdList 3(formula 1-7)
T wherein 3=T 1+ T 2, P 3=(P 1+ P 2)/2,, R 3=(|| P 1-P 2||+R 1+ R 2)=2, ObjList 3=ObjList 1+ ObjList 2, MCIdList 3=MCIdList 1+ MCIdList 2+ ID 1+ ID 2If the type on "=" the right is a numeric type, "+" is exactly the numerical value addition with them; If the numerical value on "=" the right is character string type, "+" just is equivalent to merge character string; If the numerical value on "=" the right is the chained list type, "+" just representative adds all data in the chained list on "=" the right in the chained list on "=" left side.
Shown in Figure 6 is an example of mini bunch of fusion.
LC (P 1) _ 1 and LC (P 1The attribute of cutting apart of) _ 2 is P 1Be " type " LC (P 1" type " of) _ 1 is " tank ", LC (P 1" type " of) _ 2 is " aircraft ", and then they CF is 0, then representes LC (P 1) _ 1 and LC (P 1) _ 2 can not belong to same crowd, then LC (P 1) _ 1 and LC (P 1) _ 2 do not need to merge, i.e. LC (P 1) _ 1 and LC (P 1The descendants of) _ 2 need not merged, can be directly LC (P 1) _ 1 and LC (P 1The child of) _ 2 be mini bunch as LC (P 1) _ 1 and LC (P 1Mini bunch of) _ 2 father node reduced the effect that mixing operation has obtained beta pruning.

Claims (7)

1. target grouping method during a situation is assessed is characterized in that, comprises the steps:
Step 1; Generate attribute priority chained list and clustering factor matrix; Attribute priority chained list is meant the chained list of arranging sequentially of selected objective attribute target attribute; The clustering factor matrix is the matrix that the clustering factor between different attribute forms, clustering factor be meant a certain target on the single attribute or on a plurality of attributes with the polymerization expectation value of other target;
Step 2 is selected first attribute that the target in the battlefield is cut apart and is generated type bunch in the dependency priority chained list, and with deleting in the selected attribute dependency priority chained list;
Whether step 3, the destination number in judging type bunch greater than predefined destination number threshold value, minimum destination number in destination number threshold value type of being meant bunch;
If step 4 greater than the destination number threshold value, then continues in the dependency priority chained list to select attribute that class bunch is cut apart, and be that destination number in the empty perhaps bottom class bunch is less than the destination number threshold value until attribute priority chained list;
Step 5 if be empty less than threshold value or attribute priority chained list, then carried out cluster analysis according to the position relation of target to the target data in the bottom class bunch and is generated mini bunch;
Step 6 adopts blending algorithm that said mini bunch is merged, and obtains target complex; Corresponding clustering factor need in the clustering factor matrix, be inquired about during fusion, if clustering factor is greater than the clustering factor threshold value; Then can merge; If clustering factor less than the clustering factor threshold value, then can not merge, wherein the clustering factor threshold value is meant the minimum polymerization expectation value between two dissimilar mini bunch.
2. according to the said target grouping method of claim 1, it is characterized in that, select the attribute of limit priority to carry out recurrence in the said step 2 in the dependency priority chained list and cut apart.
3. according to claim 1 or 2 said target grouping methods, it is characterized in that the said concrete grammar of cutting apart is:
For the attribute of nonumeric type, different types is divided into different classes bunch;
For the attribute of numeric type, cutting procedure is described below: at first at class bunch LC iIn target chained list Objlist in select a target data O at random sGenerate subclass bunch LC as seed i_ 1, and this subclass bunch LC i_ 1 joins above-mentioned type of bunch LC iSubclass bunch chained list LClist in, and then a type bunch LC iIn other target datas O iWith seed O sCompare, if the attribute P that these two targets are being selected iOn difference less than pre-determined threshold value, then with this target data O iAdd the affiliated class bunch LC of seed iIn _ 1; If its difference is greater than threshold value, then with this destination object as new seed O s' the new class bunch LC of generation i_ n; As all target data O iWhen all being assigned to subclass bunch, select the attribute of time priority again, the above-mentioned algorithm of recurrence is until all properties P iBe selected or divided type bunch in the target data that comprised be less than predetermined threshold value, attribute of every selection is cut apart the data in the class bunch and is equivalent to the LCMC tree and increases one deck.
4. according to the said target grouping method of claim 1, it is characterized in that the destination number threshold value gets 3 in the said step 3.
5. according to the said target grouping method of claim 1, it is characterized in that, when attribute priority chain table be not empty and type bunch in destination number greater than the destination number threshold value, carry out step 2 in the step 4 continued.
6. according to the said target grouping method of claim 1, it is characterized in that the step that generates mini bunch is: computation core radius C oreR at first select a point as seed, and to put with this is that the center of circle is with CoreR *ScaleFactor is that radius generates one mini bunch, and wherein ScaleFactor is the adjustment factor, and initial value is made as 1.0; Dynamically adjust at feedback stage, whether other target datas in relatively type bunch are positioned at mini bunch then, if; In then adding this mini bunch to this target data, if not, then generate new mini bunch to this target data as new seed; When all data all got in mini bunch, process finished.
7. according to target grouping method in the said situation assessment of claim 1, it is characterized in that said mini bunch of fusion rule is divided into three kinds of situation according to the distance between mini bunch:
Distance between two mini bunch of first kind of situation is less than zero, and then this two mini bunch can merge;
Distance between two mini bunch of second kind of situation is greater than the ultimate range between the predefined target complex, can not merge this two mini bunch of at this moment;
Distance between two mini bunch of the third situation is between zero-sum ultimate range, and the point midway between calculating two mini bunch earlier this moment is the center of circle with this point midway then; With in two mini bunch bigger one be radius; How many target datas circle of paintings has drop in the new circle in calculating two mini bunch, if target data falls within the new circle more than or equal to 3; Then this two mini bunch belongs to a crowd, can merge.
CN2012101128499A 2012-04-17 2012-04-17 Object clustering method in situation assessment Active CN102663430B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2012101128499A CN102663430B (en) 2012-04-17 2012-04-17 Object clustering method in situation assessment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2012101128499A CN102663430B (en) 2012-04-17 2012-04-17 Object clustering method in situation assessment

Publications (2)

Publication Number Publication Date
CN102663430A true CN102663430A (en) 2012-09-12
CN102663430B CN102663430B (en) 2013-11-27

Family

ID=46772914

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2012101128499A Active CN102663430B (en) 2012-04-17 2012-04-17 Object clustering method in situation assessment

Country Status (1)

Country Link
CN (1) CN102663430B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106251004A (en) * 2016-07-22 2016-12-21 中国电子科技集团公司第五十四研究所 The Target cluster dividing method divided based on room for improvement distance
CN107830865A (en) * 2017-10-16 2018-03-23 东软集团股份有限公司 A kind of vehicle target sorting technique, device, system and computer program product
CN109544082A (en) * 2017-09-21 2019-03-29 成都紫瑞青云航空宇航技术有限公司 A kind of system and method for digital battlefield confrontation
CN111783020A (en) * 2020-07-22 2020-10-16 中国人民解放军海军航空大学 Multidimensional characteristic battlefield entity target grouping method and system
CN117633563A (en) * 2024-01-24 2024-03-01 中国电子科技集团公司第十四研究所 Multi-target top-down hierarchical grouping method based on OPTICS algorithm

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106599927B (en) * 2016-12-20 2019-05-17 中国电子科技集团公司第五十四研究所 The Target cluster dividing method divided based on Fuzzy ART

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1087200A1 (en) * 1999-09-03 2001-03-28 BODENSEEWERK GERÄTETECHNIK GmbH Missile mission unit

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1087200A1 (en) * 1999-09-03 2001-03-28 BODENSEEWERK GERÄTETECHNIK GmbH Missile mission unit

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
张明远等: "态势觉察中目标分群技术的实现", 《电光与控制》 *
龙真真等: "基于层次聚类态势估计中的目标分群算法", 《弹箭与制导学报》 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106251004A (en) * 2016-07-22 2016-12-21 中国电子科技集团公司第五十四研究所 The Target cluster dividing method divided based on room for improvement distance
CN106251004B (en) * 2016-07-22 2019-10-29 中国电子科技集团公司第五十四研究所 The Target cluster dividing method divided based on room for improvement distance
CN109544082A (en) * 2017-09-21 2019-03-29 成都紫瑞青云航空宇航技术有限公司 A kind of system and method for digital battlefield confrontation
CN109544082B (en) * 2017-09-21 2023-07-07 成都紫瑞青云航空宇航技术有限公司 System and method for digital battlefield countermeasure
CN107830865A (en) * 2017-10-16 2018-03-23 东软集团股份有限公司 A kind of vehicle target sorting technique, device, system and computer program product
CN111783020A (en) * 2020-07-22 2020-10-16 中国人民解放军海军航空大学 Multidimensional characteristic battlefield entity target grouping method and system
CN111783020B (en) * 2020-07-22 2024-01-05 中国人民解放军海军航空大学 Battlefield entity target grouping method and system with multidimensional features
CN117633563A (en) * 2024-01-24 2024-03-01 中国电子科技集团公司第十四研究所 Multi-target top-down hierarchical grouping method based on OPTICS algorithm
CN117633563B (en) * 2024-01-24 2024-05-10 中国电子科技集团公司第十四研究所 Multi-target top-down hierarchical grouping method based on OPTICS algorithm

Also Published As

Publication number Publication date
CN102663430B (en) 2013-11-27

Similar Documents

Publication Publication Date Title
CN102663430B (en) Object clustering method in situation assessment
US11829887B2 (en) Progressive objective addition in multi-objective heuristic systems and methods
CN104679743B (en) A kind of method and device of the preference pattern of determining user
Aghdaie et al. Synergies of data mining and multiple attribute decision making
Carpentier et al. Trade-offs between timber production, carbon stocking and habitat quality when managing woodlots for multiple ecosystem services
Avila et al. SUSSING MERGER TREES: the influence of the halo finder
CN102881039B (en) Based on the tree three-dimensional vector model construction method of laser three-dimensional scanning data
CN104266656A (en) Method and device for searching shortest path of road network
KR101675778B1 (en) Decision making system corresponding to volcanic disaster
CN110046810A (en) A kind of Shop Floor Multiobjective Scheduling method based on Timed Petri nets
CN110097529A (en) A kind of farmland Grading unit division methods and system based on semantic rules
CN106203494A (en) A kind of parallelization clustering method calculated based on internal memory
CN104820684A (en) Fast on-line analysis processing method based on spatial position
Heinonen et al. The use of cellular automaton approach in forest planning
CN102193998A (en) Arc scanning type construction scheme of triangular irregular network containing edge topological information
CN105550368A (en) Approximate nearest neighbor searching method and system of high dimensional data
CN103926879B (en) Aero-engine casing characteristic recognition method
CN110375759A (en) Multi-robots Path Planning Method based on ant group algorithm
CN106488401B (en) Generate the method and device of seamless adjacent geography fence
CN109035363A (en) A kind of line circle optimal modeling method of iteratively faster
Zouadi et al. Hybrid manufacturing/remanufacturing lot-sizing problem with returns supplier’s selection under, carbon emissions constraint
CN105930531A (en) Method for optimizing cloud dimensions of agricultural domain ontological knowledge on basis of hybrid models
CN107818338A (en) A kind of method and system of building group pattern-recognition towards Map Generalization
CN103778195B (en) Sorting reverse skyline query method in spatial database
CN103823843B (en) Gauss mixture model tree and incremental clustering method thereof

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant