CN113065655B - Maintainability design expert system fusion reasoning method, device and storage medium - Google Patents

Maintainability design expert system fusion reasoning method, device and storage medium Download PDF

Info

Publication number
CN113065655B
CN113065655B CN202110278604.2A CN202110278604A CN113065655B CN 113065655 B CN113065655 B CN 113065655B CN 202110278604 A CN202110278604 A CN 202110278604A CN 113065655 B CN113065655 B CN 113065655B
Authority
CN
China
Prior art keywords
evidence
knowledge
fusion
design
confidence
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
Application number
CN202110278604.2A
Other languages
Chinese (zh)
Other versions
CN113065655A (en
Inventor
杨志刚
吴昊
法峰
李敏
成威强
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.)
Changsha Chumeng Information Technology Co ltd
Original Assignee
Changsha Chumeng Information Technology Co ltd
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 Changsha Chumeng Information Technology Co ltd filed Critical Changsha Chumeng Information Technology Co ltd
Priority to CN202110278604.2A priority Critical patent/CN113065655B/en
Publication of CN113065655A publication Critical patent/CN113065655A/en
Application granted granted Critical
Publication of CN113065655B publication Critical patent/CN113065655B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/20—Administration of product repair or maintenance

Abstract

The application relates to a fusion reasoning method, a fusion reasoning device and a storage medium for a maintainability design expert system, which can improve the confidence level of the maintainability design decision by carrying out uncertainty reasoning on various different knowledge obtained by original reasoning of the expert system. The method can be applied to fusion of various different knowledge, including various case knowledge, rule knowledge, model knowledge and the like; thirdly, based on the traditional evidence theory, the conflict problem can be well solved by introducing the concept of evidence preference. The method is very suitable for the design of maintenance schemes of complex mechanical products and equipment such as automobiles, ships, airplanes and the like.

Description

Maintainability design expert system fusion reasoning method, device and storage medium
Technical Field
The application relates to the technical field of product maintainability design, in particular to a maintainability design expert system fusion reasoning method, a device and a storage medium.
Background
Maintainability is a quality characteristic of the product itself, which is an inherent characteristic imparted by the design of the product that makes it easy, quick and economical to maintain. The method solves the problem of the fundamental design of maintainability, which is widely accepted and valued in developed countries, and is applied to the design and development of products such as airplanes, ships, automobiles and the like, thereby obtaining considerable economic and social benefits. With the continuous development of product maintainability design ideas and tools, parallel maintainability design becomes a necessary requirement for modern product design and development.
The maintainability design expert system is an auxiliary system capable of optimizing the design process and helping designers to carry out maintainability design, and finally obtains one or a plurality of feasible solutions of the maintainability design problem through reasoning and decision-making according to the description of the maintainability design problem, so as to realize the online support of the product design process. The knowledge and experience data related to ship maintainability design can be converted into a knowledge base in an expert system, so that the accumulation and inheritance of knowledge are realized; the method can express and process the problem information of the ship maintainability design, and carry out reasoning decision of maintainability design knowledge to obtain a suggested maintainability design scheme; the ship designer can be trained with existing knowledge to improve the level of maintainability expertise of the new designer and experience in solving the problem.
The reasoning of the maintainability knowledge is the core of the knowledge base of the maintainability expert system, and the reasonable or reasonable reasoning is related to the performance quality of the whole expert system. An expert system with excellent performance must have a reasonable reasoning mode. For a maintainability expert system, the knowledge is complex due to the large content of the knowledge. A single knowledge representation is not convenient and reasonable for representing all maintainability knowledge. The method of the maintainability reasoning is also necessarily different due to the existence of various knowledge types and different knowledge expression modes. Only in this way, the method can be well combined with maintainability knowledge representation to efficiently complete maintainability reasoning. Therefore, to implement real-time efficient reasoning, a mixed reasoning mechanism combining case-based reasoning (CBR), rule-based reasoning (RBR) and model-based reasoning (MBR) is to be adopted to represent the reasoning process of completing the maintainance expert system according to different types of maintainance knowledge.
In the use process of the product maintainability design expert system, the following two problems are often encountered: (1) Along with the continuous enrichment of the knowledge of maintainer experts, various reasoning results are often obtained according to the conventional reasoning method, however, a designer only needs to give out a final maintainer design scheme, and intelligent filtering is needed for other evidence with low confidence coefficient by using software. (2) The expert in the maintainability field of different professions has different insights on the maintainability design method, so that the occurrence of knowledge conflict is caused, and in this case, how to perform fusion reasoning and comprehensive decision is a difficult problem to be solved in the expert system.
Disclosure of Invention
Based on the above, in order to make the mechanical product easy to maintain, for the maintainability design problem of the mechanical product, a method, a device and a storage medium for fusion reasoning of the maintainability design expert system are provided, so as to enhance the reasoning capability of the expert system on the uncertainty problem and the conflict problem.
A method of business design expert system fusion reasoning, the method comprising:
and acquiring a knowledge body of the maintenance design of the mechanical product from a maintenance expert database.
Constructing an evidence body according to the knowledge body, and constructing a fusion recognition frame according to the evidence body; the fusion recognition framework comprises a plurality of coke elements, and each coke element corresponds to one maintainability design scheme.
And obtaining the confidence degree distribution of the evidence body according to the evidence body and the fusion recognition framework.
Acquiring a preset evidence preference vector; and obtaining conflict correction items of the evidence body according to the evidence preference vector and the confidence allocation.
According to the evidence confidence degree distribution, the conflict correction items and the preset conflict correction item weight coefficients, fusion is carried out to obtain multi-knowledge fusion credibility; normalizing the multi-knowledge fusion reliability to obtain normalized multi-knowledge fusion reliability; the normalized multi-knowledge body fusion reliability comprises: the reliability of the evidence body to all maintainability design schemes in the fusion recognition framework and the uncertainty of the evidence body.
And determining the final maintainability design scheme according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and a preset decision rule.
In one embodiment, constructing a evidence body based on the case knowledge, the rule knowledge, and the model knowledge; constructing a fusion recognition framework according to the evidence body, and further comprising:
each knowledge in the case knowledge is used as a case evidence body.
And obtaining a rule evidence body according to the rule knowledge.
And taking each knowledge in the model knowledge as a model evidence body.
And obtaining the evidence body according to the case evidence body, the rule evidence body and the model evidence body.
Initializing a fusion identification framework as an empty set, namely: Θ = Φ, where Θ is the fusion recognition framework.
Setting a count x, wherein the count x is an integer greater than or equal to zero.
Fusing r cases evidence bodies and s rules evidence bodies, if element h x Is a new element, then Θ=Θ { h } U } x -a }; if element h x Not a new element, then Θ=Θ.
When the count meets preset output conditions, stopping fusion to obtain the fusion identification framework; the preset search output condition is thatWherein jv represents the number of coke elements contained in the jth evidence body; the knowledge element framework includes n focal elements.
In one embodiment, the obtaining the confidence assignment of the evidence body according to the evidence body and the fusion recognition framework further includes:
and carrying out case-based reasoning on the case knowledge to obtain the case comprehensive similarity of each case in the case library.
Obtaining a j-th focal element A in the fusion recognition frame Θ according to the preset design decision attribute of the case, the comprehensive similarity of the case and the preset similarity lower limit j Overall focal length similarity of (2); the j-th focal element A j The total focal length similarity calculation formula is as follows:
wherein: sim' j Overall focal element similarity, sim, representing the j-th focal element k The comprehensive similarity of cases, sim, representing each case in the case library k The delta is greater than delta, delta is a preset similarity lower limit, and DCS_Dcp k Representing design decision attributes of the kth case, A j Represents the j-th focal element;
and carrying out normalization processing on the total similarity of the focus elements to obtain normalized case knowledge confidence assignment.
And carrying out rule reasoning according to the rule knowledge to obtain the rule knowledge confidence degree distribution.
And carrying out normalization processing on the rule knowledge confidence allocation to obtain normalized rule knowledge confidence allocation.
Determining model knowledge confidence allocation according to the number of solutions obtained by model knowledge reasoning; when the number of the solutions is 1, the model knowledge confidence is allocated to be 1; and when the number of the solutions is greater than 1, carrying out average distribution on the opposite confidence level to obtain the model knowledge confidence level.
And obtaining the evidence confidence distribution according to the case knowledge confidence distribution, the rule knowledge confidence distribution and the model knowledge confidence distribution.
In one embodiment, an evidence preference vector is set; obtaining a conflict correction term of the evidence body according to the evidence preference vector and the evidence body confidence distribution, wherein the conflict correction term comprises the following steps:
Setting an evidence preference vector; the evidence preference vector is a row vector having the same number of dimensions as the number of focal elements included in the element frame.
The values of the elements in the evidence preference vector are defined as: the preference corresponding to a given focal element is:
wherein lambda is i Representative evidence body E i Reliability boundaries of (2);representative evidence body E i Evidence body confidence assignment;represents the ith focal element A i Is a preference degree of (2).
Based on the values of the elements in the evidence preference vector, the evidence body E i The evidence confidence degree distribution of the evidence and the coke element number are adopted to obtain conflict correction items of the evidence; the calculation formula of the conflict correction term of the evidence body is as follows:
wherein: q (A) represents a collision correction term of the evidence body;representative evidence body E i Evidence body confidence assignment; />Represents the ith focal element A i Is a preference degree of (2); n represents the number of focal elements.
In one embodiment, according to the evidence body confidence degree distribution, the conflict correction term and a preset conflict correction term weight coefficient, fusion is carried out to obtain multi-knowledge body fusion credibility; normalizing the multi-knowledge body fusion reliability to obtain normalized multi-knowledge body fusion reliability, and further comprising: carrying out knowledge fusion according to the evidence confidence degree distribution, the conflict correction items and preset conflict correction item weight coefficients and combining with an improved evidence theory synthesis rule to obtain multi-knowledge fusion credibility; setting the credibility of the empty set to 0; the multi-knowledge body fusion credibility calculation formula is as follows:
m * (Φ)＝0
Wherein: represents the kth evidence corresponding to the 1 st, 2 nd, … th and n th evidence 1 、k 2 、...、k n A plurality of coke elements; k represents a preset conflict correction term weight coefficient; q (A) represents a conflict correction term corresponding to subset A in the fusion identification framework; q (Θ) represents a conflict correction term corresponding to the fusion recognition framework Θ; m is m * (A) Representing a confidence allocation function corresponding to subset a in the fusion recognition frame; m is m * (Θ) represents the confidence allocation function corresponding to the fusion recognition framework Θ.
Normalizing the fusion credibility of the multiple knowledge bodies to obtain the credibility of the evidence bodies for all propositions in the fusion recognition frame and the uncertainty of the evidence, wherein the calculation formula of the credibility of the evidence bodies for all propositions in the fusion recognition frame and the uncertainty of the evidence is as follows:
m(Φ)＝0
wherein: a represents a subset of the fusion recognition frames; m (a) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to subset a; m (Θ) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to the fusion recognition framework Θ; m is m * (A) Representative of the fusion recognition frameConfidence allocation function for subset a; m is m * (Θ) represents the confidence allocation function corresponding to the fusion recognition framework Θ.
In one embodiment, the preset decision rule includes: rule 1: taking the proposition with the maximum credibility as a design decision conclusion; rule 2: the credibility of the design decision conclusion is larger than the credibility of all other propositions and the uncertainty of the evidence body; the predetermined threshold is greater than 0; rule 3: the uncertainty of the evidence body must be greater than a predetermined uncertainty threshold, which is greater than 0.
Determining a final maintainability design scheme according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and a preset decision rule, and further comprising:
and searching the maximum credibility in the credibility of all maintainability design schemes to obtain the maximum credibility.
A predetermined threshold and a predetermined uncertainty threshold are obtained.
And when the maximum credibility is larger than a preset threshold value than the credibility of all other propositions and the uncertainty of the evidence body, and the uncertainty of the evidence body is larger than the preset uncertainty threshold value, taking the maintenance design scheme corresponding to the maximum credibility as a final maintenance design scheme.
In one embodiment, the method further comprises: and classifying and identifying the final maintainability design scheme to obtain fusion decision results of different maintainability design elements.
A maintainability design expert system fusion inference apparatus, said apparatus comprising:
the knowledge body acquisition module: a knowledge body for acquiring a maintenance design of a mechanical product from a maintenance expert database.
Fusion recognition frame construction module: the method comprises the steps of constructing an evidence body according to the knowledge body, and constructing a fusion recognition frame according to the evidence body; the fusion recognition framework comprises a plurality of coke elements, and each coke element corresponds to one maintainability design scheme.
Evidence body confidence assignment determination module: and obtaining the evidence confidence allocation according to the evidence and the fusion recognition framework.
A conflict correction term determination module: the method comprises the steps of acquiring a preset evidence preference vector; and obtaining conflict correction items of the evidence body according to the evidence preference vector and the confidence allocation.
A multi-knowledge body fusion credibility determination module: the method is used for obtaining multi-knowledge fusion credibility through fusion according to the evidence confidence distribution, the conflict correction items and preset conflict correction item weight coefficients; normalizing the multi-knowledge fusion reliability to obtain normalized multi-knowledge fusion reliability; the normalized multi-knowledge body fusion reliability comprises: the reliability of the evidence body to all maintainability design schemes in the fusion recognition framework and the uncertainty of the evidence body.
Final serviceability design determination module: the method is used for determining the final maintainability design scheme according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and preset decision rules.
In one embodiment, the fusion recognition frame construction module is further configured to: taking each knowledge in the case knowledge as a case evidence body; obtaining a rule evidence body according to rule knowledge; taking each knowledge in the model knowledge as a model evidence body; and obtaining the evidence body according to the case evidence body, the rule evidence body and the model evidence body. Initializing a fusion identification framework as an empty set, namely: Θ = Φ, where Θ is the fusion recognition framework; setting a count x, wherein the count x is an integer greater than or equal to zero; fusing each element appearing in the r cases evidence bodies and the s rules evidence bodies, if the element h x Is a new element, then Θ=Θ { h } U } x -a }; if the element h x Not a new element, then Θ=Θ; when the count meets preset output conditions, stopping fusion to obtain the fusion identification framework; the preset search output condition is that Where jv represents the number of coke elements contained in the jth evidence.
A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor realizes the steps of:
and acquiring a knowledge body of the maintenance design of the mechanical product from a maintenance expert database.
Constructing an evidence body according to the knowledge body, and constructing a fusion recognition frame according to the evidence body; the fusion recognition framework comprises a plurality of coke elements, and each coke element corresponds to one maintainability design scheme.
And obtaining the confidence degree distribution of the evidence body according to the evidence body and the fusion recognition framework.
Acquiring a preset evidence preference vector; and obtaining conflict correction items of the evidence body according to the evidence preference vector and the confidence allocation.
According to the evidence confidence degree distribution, the conflict correction items and the preset conflict correction item weight coefficients, fusion is carried out to obtain multi-knowledge fusion credibility; normalizing the multi-knowledge fusion reliability to obtain normalized multi-knowledge fusion reliability; the normalized multi-knowledge body fusion reliability comprises: the reliability of the evidence body to all maintainability design schemes in the fusion recognition framework and the uncertainty of the evidence body.
And determining the final maintainability design scheme according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and a preset decision rule.
According to the method, the device and the storage medium for fusion reasoning of the maintainability design expert system, the uncertainty reasoning is carried out on various different knowledge obtained by original reasoning of the expert system, so that the confidence level of maintainability design decisions can be improved. The method can be applied to fusion of various different knowledge, including various case knowledge, rule knowledge, model knowledge and the like; thirdly, based on the traditional evidence theory, the conflict problem can be well solved by introducing the concept of evidence preference. The method is very suitable for the design of maintenance schemes of complex mechanical products and equipment such as automobiles, ships, airplanes and the like.
Drawings
FIG. 1 is a flow chart of a method for fusion reasoning of a maintainability design expert system in one embodiment;
fig. 2 is a block diagram of a maintainability design expert system fusion inference apparatus in one embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
In one embodiment, as shown in FIG. 1, a method for collaborative reasoning of a maintainability design expert system is provided, comprising the steps of:
step 100: acquiring a knowledge body of the maintainability design of the mechanical product from a maintainability expert database;
the knowledge body comprises: case knowledge, rule knowledge, and model knowledge.
Step 102: constructing an evidence body according to the knowledge body, and constructing a fusion recognition frame according to the evidence body; the fusion recognition framework comprises a plurality of coke elements, and each coke element corresponds to one maintainability design scheme.
And searching for knowledge consistent with the 'product object' and the 'design element' and the design problem from a product maintainability design knowledge base, and reorganizing to form a evidence body.
The fusion recognition framework is a non-empty set, and comprises a plurality of events which are mutually exclusive.
For example fusion recognition frame Θ= (x 1 ,x 2 ,x 3 ) Wherein x is 1 、x 2 、x 3 For events which are mutually exclusive, fusing and identifying the power set 2 of the framework Θ The expression is as follows: { phi, x 1 ,x 2 ,x 3 ，(x 1 ,x 2 ),(x 1 ,x 3 )(x 2 ,x 3 )(x 1 ,x 2 ,x 3 ) -comprising 8 elements.
In the fusion recognition framework, a basic evidence function is defined to express uncertain information, and a mass function is a power set 2 Θ In interval [0,1 ]]A mapping thereon, the mapping satisfying the following relation: m (Φ) =0,the event corresponding to m (A) > 0 is focal element. Each focal element in the fusion recognition frame corresponds to a maintainability design scheme.
Step 104: and obtaining the confidence degree distribution of the evidence body according to the evidence body and the fusion recognition frame.
The evidence body comprises a case evidence body, a rule evidence body and a model evidence body.
Determining case knowledge confidence degree distribution, rule knowledge confidence degree distribution and model knowledge confidence degree distribution according to the case evidence, rule evidence, model evidence and fusion recognition frame to obtain evidence confidence degree distribution; and obtaining evidence body confidence distribution according to the case knowledge body confidence distribution, the rule knowledge body confidence distribution and the model knowledge body confidence distribution.
Step 106: acquiring a preset evidence preference vector; and obtaining conflict correction items of the evidence body according to the evidence preference vector and the confidence allocation.
The preset evidence preference vector is a vector composed of preference of each evidence body.
The collision correction term is a correction term for eliminating evidence collision interference, and is the average value of the sum of the products of the preference degree of each evidence body and the corresponding confidence assignment.
Step 108: according to evidence body confidence degree distribution, conflict correction items and preset conflict correction item weight coefficients, fusion is carried out to obtain multi-knowledge body fusion credibility; normalizing the multi-knowledge body fusion reliability to obtain normalized multi-knowledge body fusion reliability; normalizing the multi-knowledge body fusion reliability comprises: the confidence of the evidence body for all maintainability design schemes in the fusion recognition frame and the uncertainty of the evidence body.
The preset weight coefficient of the conflict correction term is a preset parameter and can be adjusted according to actual conditions.
The multi-knowledge body fusion credibility is obtained by carrying out fusion reasoning through a plurality of evidence body credibility functions by adopting an improved evidence theory.
Uncertainty of evidence is the probability of being unable to be determined as any one of the serviceable designs.
Step 110: and determining the final maintainability design scheme according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and a preset decision rule.
According to the maintainability design expert system fusion reasoning method, uncertainty reasoning is carried out on various different knowledge obtained through original reasoning of the expert system, so that the confidence level of maintainability design decisions can be improved. The method can be applied to fusion of various different knowledge, including various case knowledge, rule knowledge, model knowledge and the like; thirdly, based on the traditional evidence theory, the conflict problem can be well solved by introducing the concept of evidence preference. The method is very suitable for the design of maintenance schemes of complex mechanical products and equipment such as automobiles, ships, airplanes and the like.
In one embodiment, the knowledge body comprises: a case knowledge body, a rule knowledge body and a model knowledge body; step 102 further comprises: taking each knowledge in the case knowledge as a case evidence body; obtaining a rule evidence body according to rule knowledge; taking each knowledge in the model knowledge as a model evidence body; and obtaining the evidence body according to the case evidence body, the rule evidence body and the model evidence body. Initializing a fusion identification framework as an empty set, namely: Θ = Φ, where Θ is the fusion recognition framework; setting a count x, wherein the count x is an integer greater than or equal to zero; fusing r cases evidence and s rules evidence, if element h x Is a new element, then Θ=Θ { h } U } x -a }; if element h x Not a new element, then Θ=Θ; when the count meets the preset output condition, the fusion is stopped, and a fusion identification frame is obtained; presetting theThe search output condition of (2) isWherein jv represents the number of coke elements contained in the jth evidence body; the knowledge element framework includes n focal elements.
In one embodiment, knowledge consistent with "product objects" and "design elements" consistent with design problems is found in a product maintainability design knowledge base, and the evidence is reorganized into evidence:
for case knowledge, each knowledge is a body of evidence.
For rule knowledge, each design is taken as a focal element of the evidence body.
For model knowledge, each knowledge is a body of evidence.
Initializing a knowledge element frameworkCount x=0. Cycling through the r cases evidence and the s rules evidence: if element h x Is a new element, then Θ '=Θ'. U { h x }. If not, Θ '=Θ'. If->The cycle ends.
The fused knowledge element framework Θ' is thus obtained, containing n coke elements in total.
In one embodiment, step 104 further comprises: carrying out case reasoning on the case knowledge to obtain the case comprehensive similarity of each case in the case library; obtaining a j-th focal element A in the fusion recognition frame Θ according to the preset design decision attribute of the case, the comprehensive similarity of the case and the preset lower limit of the similarity j Overall focal length similarity of (2); the j-th focal element A j The total focal length similarity calculation formula is as follows:
(wherein: sim k ＞δ)
Wherein: sim' j Overall focal element similarity, sim, representing the j-th focal element k DCS_Dcp representing the comprehensive similarity of cases in the case library k Representing design decision attributes of the kth case, A j Represents the j-th focal element, and delta represents a preset similarity lower limit.
Normalizing the total similarity of the focus elements to obtain normalized case knowledge confidence assignment; carrying out rule reasoning according to the rule knowledge to obtain rule knowledge confidence allocation; normalizing the rule knowledge confidence allocation to obtain normalized rule knowledge confidence allocation; determining model knowledge confidence distribution according to the number of solutions obtained by model knowledge reasoning; when the number of solutions is 1, the model knowledge confidence is allocated to be 1; when the number of the solutions is greater than 1, carrying out average distribution on the confidence level to obtain a model knowledge confidence level; and obtaining evidence body confidence distribution according to the case knowledge confidence distribution, the rule knowledge confidence distribution and the model knowledge confidence distribution.
The case library is a case library corresponding to case knowledge.
In one embodiment, the derivation of the case knowledge confidence assignment, rule knowledge confidence assignment, and model knowledge confidence assignment is calculated as follows:
(2) Case knowledge confidence assignment
When case-reasoning is performed on a mechanical product, the comprehensive similarity between the mechanical product and each case in the case library is sim i And obtaining the approximation degree adopting the decision according to the design decision attribute DCS-Dcp of the case.
This identifies the j-th focal element A in the framework Θ for the fusion j Defining the total similarity as decision attribute and A j Sum of case similarities:
to enhance the rapidity of fusion, when the number of cases in the case library is large, a lower limit delta of similarity can be agreed, and partial cases with similarity smaller than the lower limit can be ignored.
This may result in a normalized confidence assignment of:
(2) Rule knowledge body confidence assignment
Since rule reasoning is generally given in terms of various design suggestion possibilities, the confidence probabilities derived by rule reasoning can be directly used as confidence scores, provided that the confidence score of the corresponding rule for each decision is f j The normalized confidence assignment may be obtained as:
(3) Model knowledge body confidence assignment
When the ship is designed in maintainability, model reasoning has two different situations of single solution and multiple solutions. For a single solution, no confidence assignment is needed, directly 1. If multiple solutions occur, specific analysis is required according to actual conditions and other auxiliary knowledge, and if other information support is lacking, an easy case is to perform average distribution.
In one embodiment, step 106 further comprises: setting an evidence preference vector; the evidence preference vector is a row vector, and the dimension is the same as the number of focal elements included in the element frame; the values of the elements in the evidence preference vector are defined as: the preference corresponding to a given focal element is:
wherein lambda is i Representative evidence body E i Reliability boundaries of (2);representative evidence body E i Evidence body confidence assignment;represents the ith focal element A i Is a preference degree of (2).
According to the values of elements in the evidence preference vector, an evidence body E i The evidence confidence degree distribution and the coke element number of the evidence body are obtained, and a conflict correction item of the evidence body is obtained; the calculation formula of the conflict correction term of the evidence body is as follows:
wherein: q (A) represents a collision correction term of the evidence body;representative evidence body E i Evidence body confidence assignment; />Represents the ith focal element A i Is a preference degree of (2); n represents the number of focal elements.
In one embodiment, to solve the problem of traditional evidence theory methods, contradictory information is completely ignored when evidence conflicts, thereby resulting in erroneous decisions. The patent considers that when evidence conflict occurs, the evidence is first identified as true or false, and the decision conclusion of the true evidence is believed as much as possible, but is not influenced by other conflict evidence. But if one or several evidences are fully believed, the meaning of fusion is lost. For practical maintainability design support systems, different evidences may be more sensitive to certain problems, such as layout problems, and decision recognition capability based on mathematical models is stronger, so that more accurate design conclusions can be easily drawn, while expert rule-based methods may suggest other schemes, where the former should be believed, while the latter is abandoned from conflicting interference.
First, define the concept of evidence preference: since evidence conflicts are generated when evidence volumes differ significantly, let lambda be assumed i Is evidence body E i Defining the evidence preference vector as the confidence boundary of (1)Wherein->The preference corresponding to a given focal element is:
in order to make the preference degree play a role in correction when the conflict is large, the conflict correction term is defined as follows:
wherein: q (A) represents a collision correction term of the evidence body;representative evidence body E i Evidence body confidence assignment; />Represents the ith focal element A i Is a preference degree of (2); n represents the number of focal elements.
In one embodiment, step 108 further comprises: carrying out knowledge fusion according to evidence body confidence degree distribution, conflict correction items and preset conflict correction item weight coefficients and combining with an improved evidence theory synthesis rule to obtain multi-knowledge body fusion credibility; setting the credibility of the empty set to 0; the multi-knowledge body fusion credibility calculation formula is as follows:
m * (Φ)＝0
wherein: represents the kth evidence corresponding to the 1 st, 2 nd, … th and n th evidence 1 、k 2 、...、k n A plurality of coke elements; k represents a preset conflict correction term weight coefficient; q (A) represents a conflict correction term corresponding to subset A in the fusion identification framework; q (Θ) represents a conflict correction term corresponding to the fusion recognition framework Θ; m is m * (A) Representing a confidence allocation function corresponding to subset a in the fusion recognition frame; m is m * (Θ) represents a confidence allocation function corresponding to the fusion recognition framework Θ;
carrying out normalization processing on the fusion credibility of the multiple knowledge bodies to obtain the credibility of the evidence bodies on all propositions in the fusion recognition frame and the uncertainty of the evidence, wherein the calculation formulas of the credibility of the evidence bodies on all propositions in the fusion recognition frame and the uncertainty of the evidence are as follows:
m(Φ)＝0
wherein: a represents a subset of the fusion recognition frames; m (a) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to subset a; m (Θ) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to the fusion recognition framework Θ; m is m * (A) Representing a confidence allocation function corresponding to subset a in the fusion recognition frame; m is m * (Θ) represents a confidence allocation function corresponding to the fusion recognition framework Θ;
the multi-knowledge body fusion credibility calculation formula has very clear meaning, when k is smaller, the first term in the formula plays a main role, the effect is basically consistent with the effects of Dempster and Yager synthesis rules, and when k is close to 1, namely evidence is high in conflict, the synthesis result is mainly determined by k.q (), so that the defects of the first two methods are well solved, and the adverse effect of local conflict on global fusion performance is reduced.
In one embodiment, the preset decision rule includes: rule 1: taking the proposition with the maximum credibility as a design decision conclusion; rule 2: the credibility of the design decision conclusion is larger than the credibility and the uncertainty of the evidence body of all other propositions by a preset threshold value; the predetermined threshold is greater than 0; rule 3: the uncertainty of the evidence body must be greater than a predetermined uncertainty threshold, which is greater than 0. Step 110 further includes: searching the maximum credibility in the credibility of all maintainability design schemes to obtain the maximum credibility; acquiring a predetermined threshold value and a predetermined uncertainty threshold value; and when the maximum credibility is larger than a preset threshold value than the credibility of all other propositions and the uncertainty of the evidence body, and the uncertainty of the evidence body is larger than the preset uncertainty threshold value, taking the maintenance design scheme corresponding to the maximum credibility as a final maintenance design scheme.
In one embodiment, the design decision A is determined based on a preset decision rule c ：
Rule 1 m (A) c )＝max{m(A i )}；i＝1,2,…,n
Rule 2 m (A) c )-m(A i )>ε，m(A c )-m(Θ)>Epsilon, where the threshold epsilon>0；
Rule 3 m (Θ) < γ, where the threshold γ >0.
Rule 1 indicates that the design decision conclusion is the proposition with the greatest degree of confidence; rule 2 indicates that the credibility of the design conclusion must be epsilon greater than the credibility and evidence uncertainty of all other propositions; rule 3 indicates that the uncertainty of the evidence must be less than γ, where ε and γ are determined from the actual situation.
In one embodiment, the method further comprises: and classifying and identifying the final maintainability design scheme to obtain fusion decision results of different maintainability design elements.
In one embodiment, after the fusion inference of each element is completed, for each design element, the detailed fusion result is displayed as follows: for "(also) element" design of "(also) object", the following scheme should be used: (corresponding to the scheme with the highest confidence in element 1) and give detailed reasoning information. The results of the reasoning are shown in Table 1:
table 1: confidence after fusion for "(element)" of "(object)"
In an embodiment for explaining the specific process and beneficial effects of the present invention for maintainable design fusion reasoning, for the jet pump of the ship domestic sewage treatment device, the out-of-cabin is a difficult link for maintenance thereof, and is also the key content of maintainable design. In the design, four technical experts give 4 kinds of cabin-taking schemes:
scheme E: the cabin outlet path is longest through the adjacent cabin and deck, but the strong structural part is not damaged, and other equipment is not required to be moved;
scheme F: the cabin outlet path is longer, but the strong structure part is not damaged, and a plurality of devices need to be moved;
Scheme G: the cabin outlet path is shorter, the strong structure part is not damaged, and a plurality of devices with requirements on installation precision are required to be moved;
scheme H: the longitudinal direction is directly taken out of the cabin, the cabin taking-out path is short, the strong structural part is destroyed, and other equipment is not required to be moved.
The established maintainability design fusion framework is Θ= { scheme E, scheme F, scheme G, scheme H }. Four experts as rule knowledge of four different confidence levels to generate E 1 ～E 4 Pushing four evidence bodies, and obtaining a confidence allocation function through expert scoring and normalization, wherein the confidence allocation function is as follows:
E 1 :m 1 (E)＝0.000，m 1 (F)＝0.927，m 1 (G)＝0.060，m 1 (H)＝0.011
E 2 :m 2 (E)＝0.902，m 2 (F)＝0.000，m 2 (G)＝0.042，m 2 (H)＝0.056
E 3 :m 3 (E)＝0.021，m 3 (F)＝0.732，m 3 (G)＝0.131，m 3 (H)＝0.107
E 4 :m 4 (E)＝0.057，m 4 (F)＝0.682，m 4 (G)＝0.012，m 4 (H)＝0.051
visible conflicts are predominantly manifested in m 2 In this regard, its decision is clearly different from other evidence. We are directed to E 2 Two different preferences are given: (1) Corresponds to E 1 ～E 4 Four experts have a evidence preference vector of [1 1 1 1 ]]The credibility boundary is 0.5; (2) Evidence preference vector [1 5 1 1 ]]The confidence boundary is still 0.5. The fusion results obtained by several methods, respectively, are shown in the table. In the decision rule, the threshold epsilon=0.3 and gamma=0.4 is selected.
The fusion rule provided by the patent is adopted for fusion and compared with the traditional Dempster fusion rule and the Yager rule. It can be seen that the result obtained by the Dempster synthesis rule is almost wrong, the scheme C with no evidence is used as a fusion conclusion, the credibility of the scheme a and the scheme B is low, and the result is contrary to normal; the Yager synthesis rule obviously shows a 'conservation' weakness when fusing a plurality of evidence bodies, and cannot judge what design scheme. While the improved fusion method well handles conflicts and preferences, in the case of consistent preference, the higher the degree of evidence supported by other evidence is considered, the greater the impact in the final fusion result, so that evidence 2 can be actually seen as an interference to play a smaller role; while the possibility of E propositions is greatest when evidence 2 clearly favors E, so that conflicting contradictions of different evidences are better resolved.
TABLE 2 comparison of results for different composition rules
It should be understood that, although the steps in the flowchart of fig. 1 are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in fig. 1 may include multiple sub-steps or stages that are not necessarily performed at the same time, but may be performed at different times, nor do the order in which the sub-steps or stages are performed necessarily performed in sequence, but may be performed alternately or alternately with at least a portion of other steps or sub-steps of other steps.
In one embodiment, as shown in FIG. 2, a maintainability design expert system fusion inference apparatus is provided, comprising: the system comprises a knowledge body acquisition module, a fusion identification framework construction module, an evidence body confidence degree distribution determination module, a conflict correction item determination module, a unified multi-knowledge body fusion credibility determination module and a final maintainability design scheme determination module, wherein:
The knowledge body acquisition module: a knowledge body for acquiring a maintenance design of a mechanical product from a maintenance expert database.
Fusion recognition frame construction module: the method comprises the steps of constructing an evidence body according to a knowledge body, and constructing a fusion recognition frame according to the evidence body; the fusion recognition framework comprises a plurality of coke elements, and each coke element corresponds to one maintainability design scheme.
Evidence body confidence assignment determination module: and the method is used for obtaining evidence confidence degree distribution according to the evidence and the fusion recognition framework.
A conflict correction term determination module: the method comprises the steps of acquiring a preset evidence preference vector; and obtaining conflict correction items of the evidence body according to the evidence preference vector and the confidence allocation.
The normalized multi-knowledge body fusion reliability determining module: the method comprises the steps of obtaining multi-knowledge fusion credibility through fusion according to evidence body confidence assignment, conflict correction items and preset conflict correction item weight coefficients corresponding to the conflict correction items; normalizing the multi-knowledge body fusion reliability to obtain normalized multi-knowledge body fusion reliability; normalizing the multi-knowledge body fusion reliability comprises: the credibility of the evidence body on all maintainability design schemes in the fusion recognition frame and the uncertainty of the evidence body;
Final serviceability design determination module: the method is used for determining the final maintainability design scheme according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and preset decision rules.
In one embodiment, the knowledge body comprises: a case knowledge body, a rule knowledge body and a model knowledge body; the fusion recognition frame construction module is further configured to: taking each knowledge in the case knowledge body as a case evidence body; obtaining a rule evidence body according to the rule knowledge body; taking each knowledge in the model knowledge body as a model evidence body; and obtaining the evidence body according to the case evidence body, the rule evidence body and the model evidence body. Initializing a fusion identification framework as an empty set, namely: Θ = Φ, where Θ is the fusion recognition framework; setting a count x, wherein the count x is an integer greater than or equal to zero; fusing r cases evidence and s rules evidence, if element h x Is a new element, then Θ=Θ { h } U } x -a }; if element h x Not a new element, then Θ=Θ; when the count meets the preset output condition, the fusion is stopped, and a fusion identification frame is obtained; the preset searching output condition is that Wherein jv represents the number of coke elements contained in the jth evidence body; the knowledge element framework includes n focal elements.
In one embodiment, the evidence body confidence assignment determination module is further to: carrying out case reasoning on the case knowledge body to obtain comprehensive similarity between the case knowledge body and each case in the case library; according to the design decision attribute of the preset case and the caseSynthesizing the similarity and a preset similarity lower limit to obtain a j-th focal element A in the fusion identification framework Θ j Overall focal length similarity of (2); the j-th focal element A j The total focal length similarity calculation formula is as follows:
(wherein: sim k ＞δ)
Wherein: sim' j Overall focal element similarity, sim, representing the j-th focal element k DCS-Dcp representing comprehensive similarity between individual cases in a case knowledge body and individual cases in a case library k Representing design decision attributes of the kth case, A j Represents the j-th focal element, and delta represents a preset similarity lower limit.
Normalizing the total similarity of the focus elements to obtain normalized case knowledge body confidence assignment; carrying out rule reasoning according to the rule knowledge body to obtain the confidence allocation of the rule knowledge body; normalizing the rule knowledge confidence allocation to obtain normalized rule knowledge confidence allocation; determining confidence degree distribution of the model knowledge body according to the number of solutions obtained by model reasoning; when the number of solutions is 1, the confidence of the model knowledge body is allocated to be 1; when the number of the solutions is greater than 1, carrying out average distribution on the confidence level to obtain the confidence level of the model knowledge body; and obtaining evidence body confidence distribution according to the case knowledge body confidence distribution, the rule knowledge body confidence distribution and the model knowledge body confidence distribution.
In one embodiment, the conflict correction term determination module is further configured to: setting an evidence preference vector; the evidence preference vector is a row vector, and the dimension is the same as the number of focal elements included in the element frame; the values of the elements in the evidence preference vector are defined as: the preference corresponding to a given focal element is:
wherein lambda is i Representative evidence body E i Reliability boundaries of (2);representative evidence body E i Evidence body confidence assignment;represents the ith focal element A i Is a preference degree of (2).
According to the values of elements in the evidence preference vector, an evidence body E i The evidence confidence degree distribution and the coke element number of the evidence body are obtained, and a conflict correction item of the evidence body is obtained; the calculation formula of the conflict correction term of the evidence body is as follows:
/>
wherein: q (A) represents a collision correction term of the evidence body;representative evidence body E i Evidence body confidence assignment; />Represents the ith focal element A i Is a preference degree of (2); n represents the number of focal elements.
In one embodiment, the normalized multi-knowledge body fusion reliability determination module is further configured to: carrying out knowledge fusion according to evidence body confidence degree distribution, conflict correction items and preset conflict correction item weight coefficients and combining with an improved evidence theory synthesis rule to obtain multi-knowledge body fusion credibility; setting the credibility of the empty set to 0; the multi-knowledge body fusion credibility calculation formula is as follows:
m * (Φ)＝0
In the middle of Represents the kth evidence corresponding to the 1 st, 2 nd, … th and n th evidence 1 、k 2 、...、k n A plurality of coke elements; k represents a preset conflict correction term weight coefficient; q (A) represents a conflict correction term corresponding to subset A in the fusion identification framework; q (Θ) represents a conflict correction term corresponding to the fusion recognition framework Θ; m is m * (A) Representing a confidence allocation function corresponding to subset a in the fusion recognition frame; m is m * (Θ) represents the confidence allocation function corresponding to the fusion recognition framework Θ.
Carrying out normalization processing on the fusion credibility of the multiple knowledge bodies to obtain the credibility of the evidence bodies on all propositions in the fusion recognition frame and the uncertainty of the evidence, wherein the calculation formulas of the credibility of the evidence bodies on all propositions in the fusion recognition frame and the uncertainty of the evidence are as follows:
m(Φ)＝0
wherein: a represents a subset of the fusion recognition frames; m (a) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to subset a; m (Θ) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to the fusion recognition framework Θ; m is m * (A) Representing a confidence allocation function corresponding to subset a in the fusion recognition frame; m is m * (Θ) represents the confidence allocation function corresponding to the fusion recognition framework Θ.
In one embodiment, the preset decision rule includes: rule 1: taking the proposition with the maximum credibility as a design decision conclusion; rule 2: the credibility of the design decision conclusion is larger than the credibility and the uncertainty of the evidence body of all other propositions by a preset threshold value; the predetermined threshold is greater than 0; rule 3: the uncertainty of the evidence body must be greater than a predetermined uncertainty threshold, which is greater than 0. The final serviceability design determination module is further for: searching the maximum credibility in the credibility of all maintainability design schemes to obtain the maximum credibility; acquiring a predetermined threshold value and a predetermined uncertainty threshold value; and when the maximum credibility is larger than a preset threshold value than the credibility of all other propositions and the uncertainty of the evidence body, and the uncertainty of the evidence body is larger than the preset uncertainty threshold value, taking the maintenance design scheme corresponding to the maximum credibility as a final maintenance design scheme.
In one embodiment, the maintainability design expert system fusion reasoning device further comprises a decision result display module: the method is used for classifying and identifying the final maintainability design scheme to obtain the fusion decision result of different maintainability design elements.
For specific limitations of the maintainability design expert system fusion inference apparatus, reference may be made to the above limitation of the multi-knowledge fusion method, and no further description is given herein. The modules in the above described maintainability design expert system fusion inference apparatus may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.
In one embodiment, a computer readable storage medium is provided, on which a computer program is stored which, when executed by a processor, implements the steps of the method of the above embodiments.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.

Claims (10)

1. A method for fusion reasoning of a maintainability design expert system, the method comprising:
acquiring a knowledge body of the maintainability design of the mechanical product from a maintainability expert database; the maintenance expert database is a ship maintenance expert database, and the knowledge body of the maintenance design of the mechanical product is the knowledge body of the maintenance design of the jet pump cabin of the ship domestic sewage treatment device; the knowledge body is a set of multiple types of knowledge, wherein the knowledge is selected from a preset product maintainability design knowledge base and is consistent with a product object, and design elements and design problems; the design problem is the cabin outlet design problem of the jet pump of the ship domestic sewage treatment device; the product object is a jet pump of a ship domestic sewage treatment device;
Constructing an evidence body according to the knowledge body, and constructing a fusion recognition frame according to the evidence body; the fusion identification framework comprises a plurality of coke elements, and each coke element corresponds to one maintainability design scheme; the maintainability design is boats and ships domestic sewage treatment device's jet pump cabin design, includes: scheme E: the cabin outlet path is longest through the adjacent cabin and deck, but the strong structural part is not damaged, and other equipment is not required to be moved; scheme F: the cabin outlet path is longer, but the strong structure part is not damaged, and a plurality of devices need to be moved; scheme G: the cabin outlet path is shorter, the strong structure part is not damaged, and a plurality of devices with requirements on installation precision are required to be moved; scheme H: the vertical direction cabin outlet has short cabin outlet path, damages the strong structure part and does not need to move other equipment;
obtaining evidence confidence degree distribution according to the evidence and the fusion recognition framework;
acquiring a preset evidence preference vector; obtaining conflict correction items of the evidence body according to the evidence preference vector and the confidence allocation;
fusing according to the evidence confidence distribution, the conflict correction items and preset conflict correction item weight coefficients to obtain multi-knowledge fusion credibility; normalizing the multi-knowledge fusion reliability to obtain normalized multi-knowledge fusion reliability; the normalized multi-knowledge body fusion reliability comprises: the credibility of the evidence body to all maintainability design schemes in the fusion recognition frame and the uncertainty of the evidence body;
And determining the final maintainability design scheme of the jet pump cabin design of the ship domestic sewage treatment device according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and a preset decision rule.
2. The method of claim 1, wherein the knowledge body comprises: case knowledge, rule knowledge, and model knowledge;
constructing an evidence body according to the knowledge body, and constructing a fusion recognition framework according to the evidence body, wherein the construction comprises the following steps:
taking each knowledge in the case knowledge as a case evidence body;
obtaining a rule evidence body according to the rule knowledge;
taking each knowledge in the model knowledge as a model evidence body;
obtaining an evidence body according to the case evidence body, the rule evidence body and the model evidence body;
initializing a fusion identification framework as an empty set, namely: Θ = Φ, where Θ represents a fusion recognition framework;
setting a count x, wherein the count x is an integer greater than or equal to zero;
fusing r cases evidence bodies and s rules evidence bodies, if element h x Is a new element, then Θ=Θ { h } U } x -a }; if element h x Not a new element, then Θ=Θ;
when the count x meets preset output conditions, stopping fusion to obtain the fusion identification framework; the preset search output condition is thatWhere jv represents the number of coke elements contained in the jth evidence.
3. The method of claim 2, wherein deriving evidence body confidence assignments from the evidence body and the fusion recognition framework comprises:
carrying out case reasoning on the case knowledge to obtain the case comprehensive similarity of each case in the case library;
obtaining a j-th focal element A in the fusion recognition frame according to the design decision attribute of the preset case, the comprehensive similarity of the case and the preset similarity lower limit j Overall focal length similarity of (2); the j-th focal element A j The total focal length similarity calculation formula is as follows:
wherein: sim' j The total focal element similarity of the j-th focal element; sim (sim) k The comprehensive similarity of cases, sim, representing each case in the case library k >Delta, delta represents a preset similarity lower limit; DCS_Dcp k Representing design decision attributes of a kth case; a is that j Represents the j-th focal element;
normalizing the total similarity of the focus elements to obtain normalized case knowledge confidence assignment;
Performing rule reasoning according to the rule knowledge to obtain rule knowledge confidence allocation;
normalizing the rule knowledge confidence allocation to obtain normalized rule knowledge confidence allocation;
determining model knowledge confidence allocation according to the number of solutions obtained by model knowledge reasoning; when the number of the solutions is 1, the model knowledge confidence is allocated to be 1; when the number of the solutions is greater than 1, carrying out average distribution on the opposite confidence level to obtain model knowledge confidence level;
and obtaining the evidence confidence distribution according to the case knowledge confidence distribution, the rule knowledge confidence distribution and the model knowledge confidence distribution.
4. The method of claim 1, wherein a pre-set evidence preference vector is obtained; obtaining a conflict correction term of the evidence body according to the evidence preference vector and the confidence allocation, wherein the conflict correction term comprises the following steps:
acquiring a preset evidence preference vector; the evidence preference vector is a row vector, and the dimension is the same as the number of focal elements included in the element frame;
the values of the elements in the evidence preference vector are defined as: the preference corresponding to a given focal element is:
Wherein: lambda (lambda) i Representative evidence body E i Reliability boundaries of (2);
representative evidence body E i Evidence body confidence assignment;
represents the ith focal element A i Is a preference degree of (2);
based on the values of the elements in the evidence preference vector, the evidence body E i The evidence confidence degree distribution of the evidence and the coke element number are adopted to obtain conflict correction items of the evidence; the calculation formula of the conflict correction term of the evidence body is as follows:
wherein: q (A) represents a collision correction term of the evidence body;
representative evidence body E i Evidence body confidence assignment;
represents the ith focal element A i Is a preference degree of (2);
n represents the number of focal elements.
5. The method of claim 1, wherein the multi-knowledge-body fusion credibility is obtained by fusion according to the evidence body confidence assignment, the conflict correction term and a preset conflict correction term weight coefficient; normalizing the multi-knowledge body fusion reliability to obtain normalized multi-knowledge body fusion reliability, comprising:
carrying out knowledge fusion according to the evidence confidence degree distribution, the conflict correction items and the preset conflict correction item weight coefficients and combining with an improved evidence theory synthesis rule to obtain multi-knowledge fusion credibility; setting the credibility of the empty set to 0; the multi-knowledge body fusion credibility calculation formula is as follows:
m * (Φ)＝0
Wherein: represents the kth evidence corresponding to the 1 st, 2 nd, … th and n th evidence 1 、k 2 、…、k n A plurality of coke elements;
k represents a preset conflict correction term weight coefficient;
q (A) represents a conflict correction term corresponding to subset A in the fusion identification framework;
q (Θ) represents a conflict correction term corresponding to the fusion recognition framework Θ;
m * (A) Representing a confidence allocation function corresponding to subset a in the fusion recognition frame;
m * (Θ) represents a confidence allocation function corresponding to the fusion recognition framework Θ;
normalizing the fusion credibility of the multiple knowledge bodies to obtain the credibility of the evidence bodies for all propositions in the fusion recognition frame and the uncertainty of the evidence, wherein the calculation formula of the credibility of the evidence bodies for all propositions in the fusion recognition frame and the uncertainty of the evidence is as follows:
m(Φ)＝0
wherein: a represents a subset of the fusion recognition frames;
m (a) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to subset a;
m (Θ) represents a normalized multi-knowledge-body fusion reliability allocation function corresponding to the fusion recognition framework Θ;
m * (A) Representing a confidence allocation function corresponding to subset a in the fusion recognition frame;
m * (Θ) represents the confidence allocation function corresponding to the fusion recognition framework Θ.
6. The method of claim 1, wherein the pre-set decision rule comprises:
rule 1: taking the proposition with the maximum credibility as a design decision conclusion;
rule 2: the credibility of the design decision conclusion is larger than the credibility of all other propositions and the uncertainty of the evidence body; the predetermined threshold is greater than 0;
rule 3: the uncertainty of the evidence body must be greater than a predetermined uncertainty threshold, which is greater than 0;
determining a final maintainability design scheme according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and a preset decision rule, wherein the method comprises the following steps of:
searching the maximum credibility in the credibility of all maintainability design schemes to obtain the maximum credibility;
acquiring a predetermined threshold value and a predetermined uncertainty threshold value;
and when the maximum credibility is larger than a preset threshold value than the credibility of all other propositions and the uncertainty of the evidence body, and the uncertainty of the evidence body is larger than the preset uncertainty threshold value, taking the maintenance design scheme corresponding to the maximum credibility as a final maintenance design scheme.
7. The method according to claim 1, wherein the method further comprises: and classifying and identifying the final maintainability design scheme to obtain fusion decision results of different maintainability design elements.
8. A maintainability design expert system fusion inference apparatus, said apparatus comprising:
the knowledge body acquisition module: a knowledge body for acquiring a maintenance design of the mechanical product from a maintenance expert database; the maintenance expert database is a ship maintenance expert database, and the knowledge body of the maintenance design of the mechanical product is the knowledge body of the maintenance design of the jet pump cabin of the ship domestic sewage treatment device; the knowledge body is a set of multiple types of knowledge, wherein the knowledge is selected from a preset product maintainability design knowledge base and is consistent with a product object, and design elements and design problems; the design problem is the cabin outlet design problem of the jet pump of the ship domestic sewage treatment device; the product object is a jet pump of a ship domestic sewage treatment device;
fusion recognition frame construction module: the method comprises the steps of constructing an evidence body according to the knowledge body, and constructing a fusion recognition frame according to the evidence body; the fusion identification framework comprises a plurality of coke elements, and each coke element corresponds to one maintainability design scheme; the maintainability design is boats and ships domestic sewage treatment device's jet pump cabin design, includes: scheme E: the cabin outlet path is longest through the adjacent cabin and deck, but the strong structural part is not damaged, and other equipment is not required to be moved; scheme F: the cabin outlet path is longer, but the strong structure part is not damaged, and a plurality of devices need to be moved; scheme G: the cabin outlet path is shorter, the strong structure part is not damaged, and a plurality of devices with requirements on installation precision are required to be moved; scheme H: the vertical direction cabin outlet has short cabin outlet path, damages the strong structure part and does not need to move other equipment;
Evidence body confidence assignment determination module: the evidence confidence degree distribution is used for obtaining the evidence confidence degree distribution according to the evidence and the fusion recognition framework;
a conflict correction term determination module: the method comprises the steps of acquiring a preset evidence preference vector; obtaining conflict correction items of the evidence body according to the evidence preference vector and the confidence allocation;
the normalized multi-knowledge body fusion reliability determining module: the method is used for obtaining multi-knowledge fusion credibility through fusion according to the evidence confidence distribution, the conflict correction items and preset conflict correction item weight coefficients; normalizing the multi-knowledge fusion reliability to obtain normalized multi-knowledge fusion reliability; the normalized multi-knowledge body fusion reliability comprises: the credibility of the evidence body to all maintainability design schemes in the fusion recognition frame and the uncertainty of the evidence body;
final serviceability design determination module: the final maintainability design scheme is used for determining the jet pump cabin design of the ship domestic sewage treatment device according to the credibility of all maintainability design schemes, the uncertainty of the evidence body and preset decision rules.
9. The apparatus of claim 8, wherein the knowledge body comprises: case knowledge, rule knowledge, and model knowledge;
the fusion recognition frame construction module is further used for:
taking each knowledge in the case knowledge as a case evidence body; obtaining a rule evidence body according to rule knowledge; taking each knowledge in model knowledge as a modelA type evidence body; and obtaining the evidence body according to the case evidence body, the rule evidence body and the model evidence body. Initializing a fusion identification framework as an empty set, namely: Θ = Φ, where Θ is the fusion recognition framework; setting a count x, wherein the count x is an integer greater than or equal to zero; fusing r cases evidence bodies and s rules evidence bodies, if the element hx is a new element, then Θ=Θ { h }, and the rule evidence bodies are fused x -a }; if the element hx is not a new element, Θ=Θ; when the count meets preset output conditions, stopping fusion to obtain the fusion identification framework; the preset search output condition is thatWhere jv represents the number of coke elements contained in the jth evidence.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 7.
CN202110278604.2A 2021-03-16 2021-03-16 Maintainability design expert system fusion reasoning method, device and storage medium Active CN113065655B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110278604.2A CN113065655B (en) 2021-03-16 2021-03-16 Maintainability design expert system fusion reasoning method, device and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110278604.2A CN113065655B (en) 2021-03-16 2021-03-16 Maintainability design expert system fusion reasoning method, device and storage medium

Publications (2)

Publication Number Publication Date
CN113065655A CN113065655A (en) 2021-07-02
CN113065655B true CN113065655B (en) 2024-02-23

Family

ID=76561328

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110278604.2A Active CN113065655B (en) 2021-03-16 2021-03-16 Maintainability design expert system fusion reasoning method, device and storage medium

Country Status (1)

Country Link
CN (1) CN113065655B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116258482B (en) * 2023-05-16 2023-07-18 盐城数融智升科技有限公司 Method for automatically selecting maintenance scheme, server and electronic equipment

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103577707A (en) * 2013-11-15 2014-02-12 上海交通大学 Robot failure diagnosis method achieved by multi-mode fusion inference

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102006017824B4 (en) * 2006-04-13 2018-10-11 Dspace Digital Signal Processing And Control Engineering Gmbh Method for constructing a diagnostic function

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103577707A (en) * 2013-11-15 2014-02-12 上海交通大学 Robot failure diagnosis method achieved by multi-mode fusion inference

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
融合多元信息的武器装备性能评估方法;陈雷雨;周志杰;唐帅文;曹友;;系统工程与电子技术;20201031(第07期);全文 *
证据推理方法在住宅质量评估中的应用;江兵;方军;徐冬玲;;建筑经济;20090630(第S1期);全文 *

Also Published As

Publication number Publication date
CN113065655A (en) 2021-07-02

Similar Documents

Publication Publication Date Title
Liu et al. Combination of classifiers with different frames of discernment based on belief functions
Langville et al. Algorithms, initializations, and convergence for the nonnegative matrix factorization
Solesvik et al. Fuzzy decision support systems in marine practice
CN110751355A (en) Scientific and technological achievement assessment method and device
CN113065655B (en) Maintainability design expert system fusion reasoning method, device and storage medium
EP4332823A1 (en) Method of training sentiment preference recognition model for comment information, recognition method, and device thereof
CN112101042A (en) Text emotion recognition method and device, terminal device and storage medium
CN110245700B (en) Classification model construction method, classification model and object identification method
CN110414005A (en) Intention recognition method, electronic device, and storage medium
CN116415170A (en) Prompt learning small sample classification method, system, equipment and medium based on pre-training language model
CN113343677A (en) Intention identification method and device, electronic equipment and storage medium
CN116861269A (en) Multi-source heterogeneous data fusion and analysis method in engineering field
Iwai et al. Bayesian Network Oriented Transfer Learning Method for Credit Scoring Model
CN113724688B (en) Post-processing method and device for speech recognition and computer equipment
CN115146073A (en) Test question knowledge point marking method for cross-space semantic knowledge injection and application
CN114970526A (en) Continuous learning method and system for text intention understanding
CN112513901A (en) Method for finding unique coordination system code from given text using artificial neural network and system for implementing the method
CN115271826A (en) Logistics line price interval prediction method and device
CN114254622A (en) Intention identification method and device
CN114357177A (en) Knowledge hypergraph generation method and device, terminal device and storage medium
CN111797195A (en) Accessory name identification method and device, computer equipment and storage medium
Trofimov Prototype approach to design of the automated expert systems with multi-structural recognition of complex images
CN113239193A (en) Text mining method and system
Midelfart et al. A rough set framework for learning in a directed acyclic graph
Alam et al. A bayesian network-based tunable image segmentation algorithm for object recognition

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