CN106919800A - The conflicting evidence management method of low similarity collision in a kind of evidence theory - Google Patents
The conflicting evidence management method of low similarity collision in a kind of evidence theory Download PDFInfo
- Publication number
- CN106919800A CN106919800A CN201710133553.8A CN201710133553A CN106919800A CN 106919800 A CN106919800 A CN 106919800A CN 201710133553 A CN201710133553 A CN 201710133553A CN 106919800 A CN106919800 A CN 106919800A
- Authority
- CN
- China
- Prior art keywords
- evidence
- matrix
- burnt
- generation
- designated
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
- G16Z99/00—Subject matter not provided for in other main groups of this subclass
Abstract
The conflicting evidence management method of low similarity collision in a kind of evidence theory, to all subset sorts of framework of identification, subset quantity is designated as P, then several evidences are generated by framework of identification, each evidence includes at least one burnt unit, after evidence generation, burnt metasequence generation phase and evidence weight allocated phase are carried out to evidence successively；Burnt metasequence generation phase includes generation extension evidence, generation sequence evidence and the generation step of ordinal matrix three；Evidence weight allocated phase is including calculating the poor matrix of sequence, calculating ranking factor and sequence number, calculating evidence similarity and support, calculate evidence credit worthiness, calculating evidence combined support and seek several steps such as evidence weight.The present invention can carry out more preferable weight distribution to the evidence of each in evidence set, reduce the weight distribution ratio of conflicting evidence, reduce its influence to fusion results, so as to lift the noise resisting ability of decision system work judgement system.
Description
Technical field
The invention belongs to information fusion technology field, the conflict of low similarity collision in specifically a kind of evidence theory
Evidence management method.
Background technology
Multi-sensor information fusion improve intelligence system decision science, reaction correctness, target positioning it is accurate
Property and Information locating accuracy aspect play conclusive effect, be widely used in automatic target detection, aircraft navigation,
The aspects such as tactical Situation, threat estimating, fault detect and positioning.However, due to by sensor performance and Data capture environment etc.
, there is conflict between causing sensors for data, it is difficult to directly carry out the effective integration of data in influence.Therefore, data are carried out
The committed step of collision detection and measurement as guide data fusion.D-S evidence theory is as the conventional means of data fusion
Detection and the main method of measurement sensor data collision, are widely used in the side such as target identification, optimization, fail-safe analysis
Face.
In evidence theory, comprising multiple burnt units in an evidence, each burnt unit corresponds to a probability, represents this
Event is genuine probability, such as m:M (A)=0.1, m (B)=0.3, m (C)=0.2, m (AB)=0.4 represents the general of A generations
Rate is that 0.1, AB at least one occurs but do not know to be that the probability which occurs is 0.4.And all burnt units are all contained in one
In set, the set is referred to as framework of identification, and framework of identification also corresponds to the son that probability are 0 in addition to comprising burnt unit including some
Collection.By taking evidence above as an example, framework of identification is { A, B, C }, and all of Jiao unit is all the subset of framework of identification in evidence.
In use, the data from sensor can be converted into evidence, then completed by the fusion between evidence
To the processing procedure of information collected by sensor.But, it is necessary to conflict to the evidence that conversion is completed before evidence fusion
Management, to reduce influence of the wrong data to fusion results.In existing method for collision management, evidence two-by-two is mainly based upon
Between similarity calculating complete evidence weight determination.However, similarity has collision, can cause to work as similarity
When collision occurs, weight distribution is unreasonable.
Similarity Measure aspect between different evidences, related scholar provides two kinds of computational methods, respectively：Method
First, by Wen C, WangY, Xu X. et al. is open on International Symposium on Neural Networks
" Fuzzy Information Fusion Algorithm of Fault Diagnosis Based on Similarity
Computational methods described in Measure of Evidence [C] " texts；Method two, by Jousselme AL, Grenier D,Boss é et al. in《Information Fusion》(2001,2(2):" the Anew distance delivered on 91-101)
Computational methods described between two bodies of evidence [J] " texts.
The content of the invention
In order to solve deficiency of the prior art, the present invention provides a kind of conflict card of low similarity collision in evidence theory
According to management method, on the basis of Similarity Measure, the calculating of burnt metasequence is introduced, reduce similarity collision true to weight
Fixed influence, makes the weight distribution of evidence in information fusion process more reasonable.
To achieve these goals, the concrete scheme that uses of the present invention for：
The conflicting evidence management method of low similarity collision in a kind of evidence theory, to all subset sorts of framework of identification, son
Collection quantity is designated as p, then generates several evidences by framework of identification, and each evidence includes at least one burnt unit, in evidence generation
Afterwards, burnt metasequence generation phase and evidence weight allocated phase are carried out to evidence successively；
The burnt metasequence generation phase is comprised the following steps：
Step S1, generation extension evidence is extended to each evidence, the method for extension is：Concentrated in p son of framework of identification,
Each is not included in the subset of the evidence to be added in the evidence, several new Jiao units are formed, each new Jiao unit correspondence
Probability be designated as 0, record the position i of each burnt unit in evidence is extended；
Step S2, all burnt unit that will be extended in evidence sort and generate sequence evidence from big to small by corresponding probability, and record is every
Position j of the individual burnt unit in the evidence that sorts；
Step S3, generation p rank matrixes, referred to as ordinal matrix, each one (i, j) coordinate of burnt unit's correspondence, the seat in ordinal matrix
Matrix element value at mark is 1, is 0 without burnt first corresponding matrix element value；
The weight distribution stage comprises the following steps：
Step N1, the ordinal matrix for extracting an evidence are designated as focus ordinal matrix, then calculate remaining sequence on evidence
The mean matrix of matrix, calculates the sequence difference matrix that focus ordinal matrix is calculated as the evidence with the difference of mean matrix afterwards, and will
The value of the second order normal form of sequence difference matrix is calculated as the norm that sorts；
Step N2, to performing step N1 on evidence；
Step N3, the sequence difference matrix to each evidence, calculate the negative sequence norm power of natural logrithm e, are calculated as ranking factor,
And calculate the ranking factor of the evidence account for ranking factor sum on evidence proportion, evidence number is multiplied by with rate of specific gravity, tied
Fruit is calculated as the number that sorts；
Step N4, the similarity for calculating each evidence and other evidences are simultaneously sued for peace, and are designated as the support of the evidence；
Step N5, calculate each evidence support account for support sum on evidence proportion, be designated as the credit worthiness of the evidence；
Step N6, the credit worthiness for calculating each evidence and the product of sequence number, are designated as the joint support of the evidence；
Step N7, the joint support for calculating each evidence account for the proportion of the sum that all evidence combineds are supported, are designated as the evidence
Weight.
In the step S1, Jiao unit of new addition in each evidence is ranked up according to the order of framework of identification subset.
In the step S2, when several burnt first corresponding probability are equal, carried out according to the order of framework of identification subset
Sequence.
Beneficial effect：The present invention can carry out more preferable weight distribution to the evidence of each in evidence set.Traditional is logical
The method for crossing Similarity Measure to realize conflicting evidence management, easily influences conflicting evidence because of the problem of similarity collision
Judge, and then influence the weight distribution of each evidence, disturb final information fusion process.And in the process of Similarity Measure
In, the size that can only calculate burnt unit is calculated without the order of focusing unit, therefore by generating sequence evidence and then calculating
Sequence number, using sequence number an as foundation for determining conflicting evidence, by row in the case of being collided there is similarity
Ordinal number carrys out the accurate weight distribution ratio for finding conflicting evidence, realizing reducing conflicting evidence, reduces its influence to fusion results,
So as to can more accurately make correct decision-making when sensor collection is to containing noisy data, the anti-noise of decision system is lifted
Acoustic energy power.
Brief description of the drawings
Fig. 1 is the simplified flowchart of management method.
Specific embodiment
Embodiments of the present invention are illustrated below according to accompanying drawing.
As shown in figure 1, the conflicting evidence management method that low similarity is collided in a kind of evidence theory, to the institute of framework of identification
There is subset sort, subset quantity is designated as p, then generates several evidences by framework of identification, each evidence includes that at least one is burnt
Unit, after evidence generation, carries out burnt metasequence generation phase and evidence weight allocated phase to evidence successively；
The burnt metasequence generation phase is comprised the following steps：
Step S1, generation extension evidence is extended to each evidence, the method for extension is：Concentrated in p son of framework of identification,
Each is not included in the subset of the evidence to be added in the evidence, several new Jiao units are formed, each new Jiao unit correspondence
Probability be designated as 0, record the position i of each burnt unit in evidence is extended；
Step S2, all burnt unit that will be extended in evidence sort and generate sequence evidence from big to small by corresponding probability, and record is every
Position j of the individual burnt unit in the evidence that sorts；
Step S3, generation p rank matrixes, referred to as ordinal matrix, each one (i, j) coordinate of burnt unit's correspondence, the seat in ordinal matrix
Matrix element value at mark is 1, is 0 without burnt first corresponding matrix element value；
The weight distribution stage comprises the following steps：
Step N1, the ordinal matrix for extracting an evidence are designated as focus ordinal matrix, then calculate remaining sequence on evidence
The mean matrix of matrix, calculates the sequence difference matrix that focus ordinal matrix is calculated as the evidence with the difference of mean matrix afterwards, and will
The value of the second order normal form of sequence difference matrix is calculated as the norm that sorts；
Step N2, to performing step N1 on evidence；
Step N3, the sequence difference matrix to each evidence, calculate the negative sequence norm power of natural logrithm e, are calculated as ranking factor,
And calculate the ranking factor of the evidence account for ranking factor sum on evidence proportion, evidence number is multiplied by with rate of specific gravity, tied
Fruit is calculated as the number that sorts；
Step N4, the similarity for calculating each evidence and other evidences are simultaneously sued for peace, and are designated as the support of the evidence；
Step N5, calculate each evidence support account for support sum on evidence proportion, be designated as the credit worthiness of the evidence；
Step N6, the credit worthiness for calculating each evidence and the product of sequence number, are designated as the joint support of the evidence；
Step N7, the joint support for calculating each evidence account for the proportion of the sum that all evidence combineds are supported, are designated as the evidence
Weight.
In the step S1, Jiao unit of new addition in each evidence is ranked up according to the order of framework of identification subset.
In the step S2, when several burnt first corresponding probability are equal, carried out according to the order of framework of identification subset
Sequence.
The calculating of conflicting evidence management method of the present invention is illustrated using an example for evidence treatment below
Process.
One framework of identification being made up of p subset, five evidence m are generated by framework of identification1,m2,m3,m4,m5, while really
The order of all subsets is calmlyJiao unit of five evidences is as follows：
Then with evidence m1As a example by calculated.
To evidence m1It is extended, generation evidence m1Extension evidenceNote
Record the position i of each burnt unit:iA=1, iB=2, iC=3, iAB=4, iAC=5, iBC=6, iABC=7,
The sequence evidence of step S2, generation extension evidence m'Record each
The position j of burnt unit:jA=2, jB=3, jC=1, jAB=4, jAC=5, jBC=6, jABC=7,
Generation evidence m1Ordinal matrix
To other evidences m2,m3,m4,m5Repeat step S1~S3, respectively obtain their ordinal matrix Mx：
Step N1, calculating evidence m1Sequence difference matrix, be designated as MM1：
Step N2, the sequence difference matrix MM for calculating remaining evidencex, it is specific as follows：
Step N3, the sequence number for calculating each evidence, are designated as Fx：
Calculate the similarity sim of each evidence and other evidences and sue for peace, be designated as the support sup of the evidencex, similarity
Computational methods be preferably method two, concrete outcome is as follows；
Calculate each evidence support account for support sum on evidence proportion, be designated as the credit worthiness of the evidence
Creditx, concrete outcome is as follows：
Step N6, the credit worthiness for calculating each evidence and the product of sequence number, are designated as the joint support Usup of the evidencex,
Concrete outcome is as follows；
Step N7, the joint support for calculating each evidence account for the proportion of the sum that all evidence combineds are supported, are designated as this
The weight w of evidencex, concrete outcome is as follows：
So it can be seen that conflicting evidence m2Final weight is minimum such that it is able to reduce it to fusion results
Influence, lifts decision system or judges the noise resisting ability of system.
The present invention by generating sequence evidence and then calculating sequence number, using sequence number as determine one of conflicting evidence according to
According to, conflicting evidence is found by sequence number is accurate in the case of being collided there is similarity, realize reducing conflicting evidence
Weight distribution ratio, reduce its influence to fusion results, so as to sensor collection to contain noisy data when can be more
Correct decision-making is accurately made, the noise resisting ability of decision system is lifted.
Claims (3)
1. the conflicting evidence management method that low similarity is collided in a kind of evidence theory, to all subset sorts of framework of identification,
Subset quantity is designated as p, then generates several evidences by framework of identification, and each evidence includes at least one burnt unit, and its feature exists
In：After evidence generation, burnt metasequence generation phase and evidence weight allocated phase are carried out to evidence successively；
The burnt metasequence generation phase is comprised the following steps：
Step S1, generation extension evidence is extended to each evidence, the method for extension is：Concentrated in p son of framework of identification,
Each is not included in the subset of the evidence to be added in the evidence, several new Jiao units are formed, each new Jiao unit correspondence
Probability be designated as 0, record the position i of each burnt unit in evidence is extended；
Step S2, all burnt unit that will be extended in evidence sort and generate sequence evidence from big to small by corresponding probability, and record is every
Position j of the individual burnt unit in the evidence that sorts；
Step S3, generation p rank matrixes, referred to as ordinal matrix, each one (i, j) coordinate of burnt unit's correspondence, the seat in ordinal matrix
Matrix element value at mark is 1, is 0 without burnt first corresponding matrix element value；
The weight distribution stage comprises the following steps：
Step N1, the ordinal matrix for extracting an evidence are designated as focus ordinal matrix, then calculate remaining sequence on evidence
The mean matrix of matrix, calculates the sequence difference matrix that focus ordinal matrix is calculated as the evidence with the difference of mean matrix afterwards, and will
The value of the second order normal form of sequence difference matrix is calculated as the norm that sorts；
Step N2, to performing step N1 on evidence；
Step N3, the sequence difference matrix to each evidence, calculate the negative sequence norm power of natural logrithm e, are calculated as ranking factor,
And calculate the ranking factor of the evidence account for ranking factor sum on evidence proportion, evidence number is multiplied by with rate of specific gravity, tied
Fruit is calculated as the number that sorts；
Step N4, the similarity for calculating each evidence and other evidences are simultaneously sued for peace, and are designated as the support of the evidence；
Step N5, calculate each evidence support account for support sum on evidence proportion, be designated as the credit worthiness of the evidence；
Step N6, the credit worthiness for calculating each evidence and the product of sequence number, are designated as the joint support of the evidence；
Step N7, the joint support for calculating each evidence account for the proportion of the sum that all evidence combineds are supported, are designated as the evidence
Weight.
2. the conflicting evidence management method that low similarity is collided in a kind of evidence theory as claimed in claim 1, its feature exists
In：In the step S1, Jiao unit of new addition in each evidence is ranked up according to the order of framework of identification subset.
3. the conflicting evidence management method that low similarity is collided in a kind of evidence theory as claimed in claim 1, its feature exists
In：In the step S2, when several burnt first corresponding probability are equal, it is ranked up according to the order of framework of identification subset.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710133553.8A CN106919800A (en) | 2017-03-08 | 2017-03-08 | The conflicting evidence management method of low similarity collision in a kind of evidence theory |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710133553.8A CN106919800A (en) | 2017-03-08 | 2017-03-08 | The conflicting evidence management method of low similarity collision in a kind of evidence theory |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106919800A true CN106919800A (en) | 2017-07-04 |
Family
ID=59460570
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710133553.8A Pending CN106919800A (en) | 2017-03-08 | 2017-03-08 | The conflicting evidence management method of low similarity collision in a kind of evidence theory |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106919800A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111814278A (en) * | 2020-08-31 | 2020-10-23 | 深圳领威科技有限公司 | Data processing method, data processing device and terminal equipment |
-
2017
- 2017-03-08 CN CN201710133553.8A patent/CN106919800A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111814278A (en) * | 2020-08-31 | 2020-10-23 | 深圳领威科技有限公司 | Data processing method, data processing device and terminal equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106650699A (en) | CNN-based face detection method and device | |
| CN110111113B (en) | Abnormal transaction node detection method and device | |
| CN112541532B (en) | Target detection method based on dense connection structure | |
| CN103150546B (en) | video face identification method and device | |
| WO2010037030A1 (en) | Evaluating loan access using online business transaction data | |
| CN106250442A (en) | The feature selection approach of a kind of network security data and system | |
| CN113011973B (en) | Method and equipment for financial transaction supervision model based on intelligent contract data lake | |
| CN106776884A (en) | A kind of act of terrorism Forecasting Methodology that multi-categorizer is combined based on multi-tag | |
| CN111292008A (en) | Privacy protection data release risk assessment method based on knowledge graph | |
| CN103473786A (en) | Gray level image segmentation method based on multi-objective fuzzy clustering | |
| CN109935337A (en) | A kind of medical record lookup method and system based on similarity measurement | |
| CN106250913B (en) | A kind of combining classifiers licence plate recognition method based on local canonical correlation analysis | |
| CN106021671B (en) | In conjunction with the circuit health grading evaluation method of correlative relationship and grey cluster technology | |
| CN107101829A (en) | A kind of intelligent diagnosing method of aero-engine structure class failure | |
| CN108460336A (en) | A kind of pedestrian detection method based on deep learning | |
| CN114266289A (en) | Complex equipment health state assessment method | |
| CN103150476B (en) | A kind of system efficiency evaluation method based on data station field | |
| CN111639706A (en) | Personal risk portrait generation method based on image set and related equipment | |
| CN108287964B (en) | Gray cloud reasoning structure damage identification method based on acceleration inner product vector | |
| CN106228132A (en) | Target identification method and Target Identification Unit | |
| CN106919800A (en) | The conflicting evidence management method of low similarity collision in a kind of evidence theory | |
| Wang et al. | Temperature forecast based on SVM optimized by PSO algorithm | |
| CN111310121A (en) | New energy output probability prediction method and system | |
| CN115204475A (en) | Drug rehabilitation place security incident risk assessment method | |
| CN113989014A (en) | Local government debt risk assessment method and device and storage medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170704 |