CN105447315A - D-S evidence law fusion improved method for evidence conflict - Google Patents
D-S evidence law fusion improved method for evidence conflict Download PDFInfo
- Publication number
- CN105447315A CN105447315A CN201510845459.6A CN201510845459A CN105447315A CN 105447315 A CN105447315 A CN 105447315A CN 201510845459 A CN201510845459 A CN 201510845459A CN 105447315 A CN105447315 A CN 105447315A
- Authority
- CN
- China
- Prior art keywords
- fusion
- evidence
- fusion results
- fusion result
- rule
- 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
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 present invention discloses a D-S evidence law fusion improved method for evidence conflict. The method comprises the steps of: fusing according to a D-S evidence law fusion rule so as to obtain a fusion result m*(Aj); adding a positive number N to each probability distribution value mi (Aj) so as to obtain Mi (Aj)=mi (Aj)+N, then fusing the Mi(Aj) according to the D-S evidence law fusion rule so as to obtain M(Aj), then subtracting the highest order term Ns of N from each M(Aj) so as to obtain M0(Aj), and performing normalization to obtain a fusion result m^(Aj); comparing the fusion result m*(Aj) with the fusion result m^(Aj), if the fusion result m*(Aj) and the fusion result m^(Aj) point to different fault types, taking the fusion result m^(Aj) as a final fusion result; if the fusion result m*(Aj) and the fusion result m^(Aj) do not point to different fault types, setting a formula as shown in the description, subtracting Epsilon from each mj(Sj), then fusing according to the D-S evidence law fusion rule so as to obtain a fusion result m'(Aj); and comparing the fusion result m*(Aj), m^(Aj) and m'(Aj), and selecting the fusion result having the greatest directivity as the final fusion result. The D-S evidence law fusion improved method for evidence conflict has higher accuracy.
Description
Technical field
The present invention relates to and fault diagnosis technology field, relate in particular to a kind of D-S evidence act fusion for evidences conflict and improve one's methods.
Background technology
Fault Diagnosis of Gear Case, due to the Multi-sensor Fusion diagnostic techniques have employed based on D-S evidence act, improves diagnosis efficiency. Its fusion rule: suppose gearbox fault type framework of identification Θ={ S1,S2,…,Sn, namely there is n kind fault, have s evidence: mi:mi(Sj), (i=1,2 ..., s; J=1,2 ..., n), the probability assignments after then mergingThe method is identification gearbox fault effectively.
But, when particular sensor is out of order, according to D-S fusion rule, likely derive a wrong conclusion. Therefore D-S fusion rule is necessary to improve.
Current, the evidence Weighted Fusion method based on Jousselme distance that Yin Xuezhong proposes, utilize the evidences conflict degree structure evidence mutual support degree matrix based on Jousselme distance, and then calculating evidence weight, D-S fusion rule is finally utilized to merge the revised evidence of weighting, efficiently solve " veto by one vote " problem, but accuracy is not high; Yager proposes conflict to distribute to framework of identification, causes have lost the effective information of a large amount of scripts in combination the uncertainty that merges rear evidence is increased; Dubois proposes conflict to distribute to union, but accuracy is not high; The evidence Weighted Fusion method based on angle similarity that Wang Yucheng proposes, utilize the similitude between the included angle cosine value metric evidence of evidence vector, and construct similarity matrix, therefrom obtain the confidence level of evidence and new evidence is obtained to former evidence correction, and combine according to D-S fusion rule, efficiently solve " veto by one vote " problem, but accuracy is not high; The discount computing that Shafer proposes, when knowing the prior information of some uncertain factor, it is joined in the probability assignments m function of evidence, be converted into suitable discount factor, suitable correction is carried out to m function, and according to the combination of D-S fusion rule, the subject matter of the method is how to determine the discount rate of prior information; Murphy proposes a kind of evidence average combined rule, first basic evidence credits assigned is carried out on average, and then carry out information fusion with D-S fusion rule, multi-source information just carries out simple average by the method, does not consider interrelated between each evidence; Deng utilizes improved Euclidean distance to obtain the correlation between evidence, and obtains evidence weight, then revises former evidence, then carries out information fusion with D-S fusion rule, efficiently solve " veto by one vote " problem, but accuracy is not high.
Summary of the invention
Technical problem to be solved by this invention is, a kind of D-S evidence act fusion for evidences conflict is provided and improves one's methods, accuracy is high, efficiently solve " veto by one vote " problem.
In order to solve the problems of the technologies described above, to the invention provides a kind of D-S evidence act fusion for evidences conflict and improve one's methods, comprising:
Merge according to D-S evidence act fusion rule, obtain fusion results m*(Aj); Wherein, framework of identification is Θ={ S1,S2,…,Sn, Boolean algebra subset is { Φ, A1,A2,…,A2 n -1, there is s evidence: mi:mi(Aj),i=1,2,…,s,j=1,2,…,2n-1;
By each probability assignments value mi(Aj) all add a positive number N, obtain Mi(Aj)=mi(Aj)+N, then by Mi(Aj) carry out fusion according to D-S evidence act fusion rule and obtain M (Aj), afterwards by each M (Aj) all deduct the most high-order term N of Ns, obtain M0(Aj), then after being normalized, obtain fusion results
Contrast fusion results m*(Aj) and m^(Aj), if fusion results m*(Aj) and m^(Aj) point to different fault types, then fusion results m^(Aj) be final fusion results;
If fusion results m*(Aj) and m^(Aj) point to same fault type, then establishBy each mi(Sj) all deduct ε after, then merge according to D-S evidence act fusion rule, obtain fusion results m'(Aj);
Contrast fusion results m*(Aj)、m^(Aj) and m'(Aj), choosing the fusion results that wherein directive property is the strongest is final fusion results m (Aj)。
Further, N is the positive number much larger than 1.
Implement the present invention, there is following beneficial effect: accuracy of the present invention is high, efficiently solve " veto by one vote " problem, when being applied to fault diagnosis, can ensure under the prerequisite that fault diagnosis result is correct, improve diagnosis efficiency when sensing system failsafe; Also diagnosis efficiency may be improved when particular sensor is out of order, avoid causing because particular sensor is inaccurate whole decision to make mistakes, effectively solve " veto by one vote " problem.
Detailed description of the invention
Technical scheme in the embodiment of the present invention will be clearly and completely described below, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiment. Based on the embodiment in the present invention, the every other embodiment that those of ordinary skill in the art obtain under not making creative work prerequisite, belongs to the scope of protection of the invention.
Embodiments provide a kind of D-S evidence act fusion for evidences conflict to improve one's methods, comprise step:
Merge according to D-S evidence act fusion rule, obtain fusion results m*(Aj); Wherein, framework of identification is Θ={ S1,S2,…,Sn, Boolean algebra subset is { Φ, A1,A2,…,A2 n -1, there is s evidence: mi:mi(Aj),i=1,2,…,s,j=1,2,…,2n-1;
By each probability assignments value mi(Aj) all add a positive number N, obtain Mi(Aj)=mi(Aj)+N, then by Mi(Aj) carry out fusion according to D-S evidence act fusion rule and obtain M (Aj), afterwards by each M (Aj) all deduct the most high-order term N of Ns, obtain M0(Aj), then after being normalized, obtain fusion resultsN is the positive number much larger than 1.
Contrast fusion results m*(Aj) and m^(Aj), if fusion results m*(Aj) and m^(Aj) point to different fault types, then fusion results m^(Aj) be final fusion results;
If fusion results m*(Aj) and m^(Aj) point to same fault type, then establishBy each mi(Sj) all deduct ε after, then merge according to D-S evidence act fusion rule, obtain fusion results m'(Aj);
Contrast fusion results m*(Aj)、m^(Aj) and m'(Aj), choosing the fusion results that wherein directive property is the strongest is final fusion results m (Aj)。
Under regard to the present invention and obtain effect and carry out concrete enforcement and verify.
Verify example 1
Apply the present invention to, in gearbox fault differentiation, establish gearbox fault and be divided into 4 kinds: S1=intact, S2=broken teeth, S3The spot corrosion of=bearing outer ring, S4=tooth root crackle, i.e. framework of identification Θ={ S1,S2,S3,S4, Boolean algebra subset is { Φ, A1,A2,…,A15, suppose A1=S1,A2=S2,A3=S3,A4=S4, Φ is empty set,
Broken teeth fault gear-box has s=3 sensor evidence to be:
m1:m1(A1)=0.020819,m1(A2)=0.953965,m1(A3)=0.018060,m1(A4)=0.015571, the probability assignments value m of other burnt units1(A5)~m1(A15) be all 0;
m2:m2(A1)=0.041925,m2(A2)=0.928404,m2(A3)=0.034728,m2(A4)=0.031065, the probability assignments value m of other burnt units2(A5)~m2(A15) be all 0;
m3:m3(A1)=0.035601,m3(A2)=0.947966,m3(A3)=0.011717,m3(A4)=0.011533, the probability assignments value m of other burnt units3(A5)~m3(A15) be all 0;
Fusion results is obtained: m according to D-S fusion rule*(A1)=0.000037,m*(A2)=0.999943,m*(A3)=0.000008,m*(A4)=0.000012, the probability assignments value of other burnt units is 0, and fusion results show that gearbox fault is S2The correct conclusion of=broken teeth.
Merge according to of the present invention improving one's methods, final fusion results is m (A1)=0.000009,m(A2)=0.999991>m*(A2),m(A3)=0.000000,m(A4)=0, the probability assignments value m (A of other burnt units5)~m(A15) be 0, fusion results show that gearbox fault is S2The result of=broken teeth.
Can find out, the present invention compares according to D-S fusion rule and obtains fusion results more accurately, and diagnosis efficiency is higher.
Verify example 2
Other parts of this checking example are identical with checking example 1, only unlike sensor evidence m3For:
m3:m3(A1)=0.035601,m3(A2)=0.000017,m3(A3)=0.959666,m3(A4)=0.011533, the probability assignments value m of other burnt units3(A5)~m3(A15) be all 0;
Obtaining fusion results according to D-S fusion rule is: m*(A1)=0.047537,m*(A2)=0.023042,m*(A3)=0.920884,m*(A4)=0.008537, the probability assignments value of other burnt units is 0, and fusion results show that gearbox fault is S3The wrong conclusion of=bearing outer ring spot corrosion.
The fusion results that adopts the weighting method based on Jousselme distance to obtain is: m (A1)=0.017252,m(A2)=0.504743,m(A3)=0.115020,m(A4)=0.011034, m (Θ)=0.351951, the probability assignments value of other burnt units is 0, and still can correctly pick out is broken teeth fault, but accuracy not high (other weighting methods roughly the same).
The fusion results of calculating according to the present invention is: when getting N=10, m (A1)=0.032782,m(A2)=0.627462,m(A3)=0.337485,m(A4)=0.019390, the probability assignments value of other burnt units is 0, show that gearbox fault is S2The correct conclusion of=broken teeth.
In sum, it is higher that the present invention compares the fusion results accuracy rate obtaining according to D-S fusion rule and additive method, and diagnosis efficiency is higher.
Verify example 3
Other parts of this checking example are identical with checking example 1, only unlike sensor evidence m3For:
m3:m3(A1)=0.035601,m3(A2)=0.011717,m3(A3)=0.947966,m3(A4)=0.011533, the probability assignments value m of other burnt units3(A5)~m3(A15) be all 0;
Fusion results is obtained: m according to D-S fusion rule*(A1)=0.002816,m*(A2)=0.942592,m*(A3)=0.054047,m*(A4)=0.000545, the probability assignments value of other burnt units is 0, draws the correct conclusion of gearbox fault broken teeth.
According to the present invention, obtain result consistent with D-S fusion rule.
Verify example 4
Other parts of this checking example are identical with checking example 1, only unlike sensor evidence m3For:
m3:m3(A1)=0.035601,m3(A2)=0.111717,m3(A3)=0.847966,m3(A4)=0.011533, the probability assignments value m of other burnt units3(A5)~m3(A15) be all 0;
Obtain according to D-S fusion rule: m*(A1)=0.000312,m*(A2)=0.994282,m*(A3)=0.005346,m*(A4)=0.000060, the probability assignments value of other burnt units is 0, draws the correct conclusion of gearbox fault broken teeth.
The fusion results of calculating according to the present invention is: m (A1)=0.000081,m(A2)=0.998454,m(A3)=0.001465,m(A4)=0, the probability assignments value of other burnt units is 0, draws the correct conclusion of gearbox fault broken teeth.
Can find out, it is higher that the present invention compares the fusion results accuracy rate obtaining according to D-S fusion rule and additive method, and diagnosis efficiency is higher.
Implement the present invention, there is following beneficial effect: accuracy of the present invention is high, efficiently solve " veto by one vote " problem, when being applied to fault diagnosis, can ensure under the prerequisite that fault diagnosis result is correct, improve diagnosis efficiency when sensing system failsafe; Also diagnosis efficiency may be improved when particular sensor is out of order, avoid causing because particular sensor is inaccurate whole decision to make mistakes, effectively solve " veto by one vote " problem.
It should be noted that, in this article, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus the process, method, article or the device that make to comprise a series of key elements not only comprise those key elements, but also comprise other key elements of clearly not listing, or be also included as the intrinsic key element of this process, method, article or device. When more restrictions not, the key element being limited by statement " comprising ... ", and be not precluded within the process, method, article or the device that comprise this key element and also have other identical element.
The invention described above embodiment sequence number, just to describing, does not represent the quality of embodiment.
To the above-mentioned explanation of the disclosed embodiments, professional and technical personnel in the field are realized or use the present invention. To be apparent for those skilled in the art to the multiple amendment of these embodiment, General Principle as defined herein can without departing from the spirit or scope of the present invention, realize in other embodiments. Therefore, the present invention can not be restricted to these embodiment shown in this article, but the widest scope consistent with principle disclosed herein and features of novelty will be met.
Claims (2)
1. the fusion of the D-S evidence act for evidences conflict is improved one's methods, and it is characterized in that, comprising:
Merge according to D-S evidence act fusion rule, obtain fusion results m*(Aj); Wherein, identification frameFrame is Θ={ S1,S2,…,Sn, Boolean algebra subset is { Φ, A1,A2,…,A2 n -1, there is s evidence: mi:mi(Aj),i=1,2,…,s,j=1,2,…,2n-1;
By each probability assignments value mi(Aj) all add a positive number N, obtain Mi(Aj)=mi(Aj)+N, then willMi(Aj) carry out fusion according to D-S evidence act fusion rule and obtain M (Aj), afterwards by each M (Aj) all deduct NMost high-order term Ns, obtain M0(Aj), then after being normalized, obtain fusion results
Contrast fusion results m*(Aj) and m^(Aj), if fusion results m*(Aj) and m^(Aj) point to different faultsType, then fusion results m^(Aj) be final fusion results;
If fusion results m*(Aj) and m^(Aj) point to same fault type, then establishBy each mi(Sj) all deduct ε after, then merge according to D-S evidence act fusion rule, obtain fusion resultsm'(Aj);
Contrast fusion results m*(Aj)、m^(Aj) and m'(Aj), choosing the fusion results that wherein directive property is the strongest isFinal fusion results m (Aj)。
2. the D-S evidence act fusion for evidences conflict is improved one's methods as claimed in claim 1, its featureBe, N is the positive number much larger than 1.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510845459.6A CN105447315B (en) | 2015-11-27 | 2015-11-27 | A kind of D-S evidence acts for evidences conflict merge improved method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510845459.6A CN105447315B (en) | 2015-11-27 | 2015-11-27 | A kind of D-S evidence acts for evidences conflict merge improved method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105447315A true CN105447315A (en) | 2016-03-30 |
CN105447315B CN105447315B (en) | 2018-05-08 |
Family
ID=55557485
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510845459.6A Active CN105447315B (en) | 2015-11-27 | 2015-11-27 | A kind of D-S evidence acts for evidences conflict merge improved method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105447315B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108446458A (en) * | 2018-03-01 | 2018-08-24 | 西北工业大学 | A kind of Weighted Fusion method for diagnosing faults based on DS evidence theories |
CN109540520A (en) * | 2018-11-29 | 2019-03-29 | 中国船舶重工集团海装风电股份有限公司 | A kind of rolling bearing fault fusion diagnosis method based on improvement D-S evidence theory |
CN110674950A (en) * | 2019-09-25 | 2020-01-10 | 国网山东省电力公司日照供电公司 | Power distribution network fault positioning method based on multi-source information |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7424466B2 (en) * | 2002-07-24 | 2008-09-09 | Northrop Grumman Corporation | General purpose fusion engine |
CN101576604A (en) * | 2009-01-04 | 2009-11-11 | 湖南大学 | Method for diagnosing failures of analog circuit based on heterogeneous information fusion |
US7778943B2 (en) * | 2007-02-09 | 2010-08-17 | Honeywell International Inc. | Stochastic evidence aggregation system of failure modes utilizing a modified dempster-shafer theory |
CN104408324A (en) * | 2014-12-11 | 2015-03-11 | 云南师范大学 | D-S evidence theory based multi-sensor information fusion method |
-
2015
- 2015-11-27 CN CN201510845459.6A patent/CN105447315B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7424466B2 (en) * | 2002-07-24 | 2008-09-09 | Northrop Grumman Corporation | General purpose fusion engine |
US7778943B2 (en) * | 2007-02-09 | 2010-08-17 | Honeywell International Inc. | Stochastic evidence aggregation system of failure modes utilizing a modified dempster-shafer theory |
CN101576604A (en) * | 2009-01-04 | 2009-11-11 | 湖南大学 | Method for diagnosing failures of analog circuit based on heterogeneous information fusion |
CN104408324A (en) * | 2014-12-11 | 2015-03-11 | 云南师范大学 | D-S evidence theory based multi-sensor information fusion method |
Non-Patent Citations (3)
Title |
---|
HONGHUA WANG: "《A study on Fault Diagnosis of hydroeelectric Generator Based on D-S Evidence Theory》", 《INTERNATIONAL CONFERENCE ON ELECTRICAL MACHINE AND SYSTEMS 2008》 * |
张岳: "《多传感器信息融合技术在电机故障诊断中的应用研究》", 《中国优秀硕士学位论文全文数据库 工程科技II辑》 * |
苗红霞 等;: "《基于数据融合的高压断路器故障诊断方法研究》", 《工矿自动化》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108446458A (en) * | 2018-03-01 | 2018-08-24 | 西北工业大学 | A kind of Weighted Fusion method for diagnosing faults based on DS evidence theories |
CN108446458B (en) * | 2018-03-01 | 2019-10-22 | 西北工业大学 | A kind of Weighted Fusion rotor method for diagnosing faults based on DS evidence theory |
CN109540520A (en) * | 2018-11-29 | 2019-03-29 | 中国船舶重工集团海装风电股份有限公司 | A kind of rolling bearing fault fusion diagnosis method based on improvement D-S evidence theory |
CN110674950A (en) * | 2019-09-25 | 2020-01-10 | 国网山东省电力公司日照供电公司 | Power distribution network fault positioning method based on multi-source information |
Also Published As
Publication number | Publication date |
---|---|
CN105447315B (en) | 2018-05-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Egawa | Will Income Inequality Cause a Middle-income Trap in Asia?. | |
CN105447315A (en) | D-S evidence law fusion improved method for evidence conflict | |
CN104408324B (en) | Multiple sensor information amalgamation method based on D S evidence theories | |
CN102495919B (en) | Extraction method for influence factors of carbon exchange of ecosystem and system | |
CN104766175A (en) | Power system abnormal data identifying and correcting method based on time series analysis | |
Wan et al. | A preference degree for intuitionistic fuzzy values and application to multi-attribute group decision making | |
CN103116811A (en) | Patent value evaluation method based on value delivery | |
CN104021392A (en) | Conflict evidence fusion method based on vector measurement | |
Vieira et al. | Integrated assessment of water reservoir systems performance with the implementation of ecological flows under varying climatic conditions | |
CN105119876A (en) | automatically-generated domain name | |
CN108171545A (en) | A kind of conversion ratio predictor method based on level of hierarchy data | |
CN105976099A (en) | Fuzzy information-based aerospace model scientific research and production management level evaluation method | |
Zhang et al. | Indirect impacts of future urban expansion on ecosystem services in the transnational area of China, North Korea, and Russia | |
CN103970651A (en) | Software architecture safety assessment method based on module safety attributes | |
Diaz Nieto et al. | Weather indices for designing micro-insurance products for small-holder farmers in the tropics | |
Stoltmann | Application of AHP method for comparing the criteria used in locating wind farms | |
Novkovska | Relationship between shadow and green economy: Less shadow more green | |
Danehkar et al. | Degradation assessment of Jajrood protected area using landscape degradation model | |
CN102663562A (en) | Method for grading land resource evaluation factors based on clonal selection algorithm | |
CN105528735A (en) | Abnormal data point correction method based on measured wind speed and spatial correlation | |
CN104462826A (en) | Method for detecting and measuring multi-sensor evidence conflicts on basis of matrix singular value decomposition | |
CN107506867B (en) | Photovoltaic output prediction method based on dynamic modeling | |
CN106682980A (en) | Method for designing probability generator | |
Morimoto | Study on Links between Cerebral Infarction and Climate Change Based on Hidden Markov Models | |
Jie et al. | The Decision-making Model of Harden Grey Target Based on Interval Number with Preference Information on Alternatives. |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |