CN113761032A - Aeroengine fault diagnosis method and system based on extension association rule mining - Google Patents
Aeroengine fault diagnosis method and system based on extension association rule mining Download PDFInfo
- Publication number
- CN113761032A CN113761032A CN202111063095.8A CN202111063095A CN113761032A CN 113761032 A CN113761032 A CN 113761032A CN 202111063095 A CN202111063095 A CN 202111063095A CN 113761032 A CN113761032 A CN 113761032A
- Authority
- CN
- China
- Prior art keywords
- fault
- extension
- engine
- correlation
- aero
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 64
- 238000005065 mining Methods 0.000 title claims abstract description 44
- 238000003745 diagnosis Methods 0.000 title claims abstract description 20
- 238000011156 evaluation Methods 0.000 claims abstract description 69
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 21
- 238000012163 sequencing technique Methods 0.000 claims abstract description 4
- 238000012545 processing Methods 0.000 claims description 11
- 230000009466 transformation Effects 0.000 claims description 10
- 238000011426 transformation method Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000005314 correlation function Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 2
- 238000012423 maintenance Methods 0.000 description 13
- 238000004458 analytical method Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 2
- 230000008030 elimination Effects 0.000 description 2
- 238000003379 elimination reaction Methods 0.000 description 2
- 238000012854 evaluation process Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 238000012502 risk assessment Methods 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000010220 Pearson correlation analysis Methods 0.000 description 1
- 238000012098 association analyses Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000003780 insertion Methods 0.000 description 1
- 230000037431 insertion Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- 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
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2216/00—Indexing scheme relating to additional aspects of information retrieval not explicitly covered by G06F16/00 and subgroups
- G06F2216/03—Data mining
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Databases & Information Systems (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Fuzzy Systems (AREA)
- Development Economics (AREA)
- Probability & Statistics with Applications (AREA)
- Software Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
An aero-engine fault diagnosis method and system based on extension association rule mining belong to the field of aero-engine fault diagnosis. The method solves the problems that the efficiency of detecting the faults of the aero-engine is low and effective inference rules cannot be screened out based on the existing association rule mining algorithm. The invention comprises the following steps: acquiring historical data of the aircraft engine, establishing an aircraft engine fault affair set based on the historical data, and expanding the fault affair type into fault characteristics; calculating the correlation between every two fault characteristics by using a Pearson correlation coefficient, setting a correlation threshold value, and eliminating one fault characteristic in a fault characteristic pair higher than the correlation threshold value to obtain a key fault characteristic; extracting frequent items from key fault features by using a frequent item set mining algorithm, and extracting strong association rules from the frequent items; and sequencing the strong association rules based on an extension evaluation method, and diagnosing the faults according to the sequence of the strong association rules. The method is used for diagnosing the faults of the aircraft engine.
Description
Technical Field
The invention belongs to the field of aeroengine fault diagnosis, and particularly relates to an aeroengine fault diagnosis method and system based on extension association rule mining.
Background
The aircraft engine is the heart of the aircraft, provides a power source for flight tasks, and is of vital importance in reliability and safety. Due to the fact that the aeroengine is complex in structure and different in service life of parts, fault reasons are various, and difficulty is increased for fault diagnosis of the aeroengine. However, a large amount of operation and maintenance data are accumulated in the actual process, and the implicit correlation relationship can be analyzed through a machine learning model, so that effective support is provided for fault diagnosis of the aircraft engine. The operation and maintenance data are wide in source, different in structure and low in value density, and an association rule mining technology is adopted to extract strong association rules for fault association analysis of the aircraft engine.
An association rule mining algorithm is more classical, namely an Apriori algorithm, the algorithm preferentially searches the width direction, a frequent item set is generated in a layer-by-layer traversal mode, but the operation efficiency is low due to the fact that a data set required to be traversed is overlarge; therefore, an FP-Tree algorithm is provided, the algorithm searches the depth direction preferentially, frequent items are generated in a Tree node insertion mode, and the operation efficiency is high. But the FP-Tree algorithm also has certain defects that the algorithm flow needs to traverse the transaction set for multiple times, and the operation efficiency is still to be improved; mining association rules, easily generating invalid rules, and adding constraint conditions for the field of fault diagnosis; and (3) association rule mining based on the FP-Tree, wherein the main evaluation indexes are confidence coefficient and support degree, the frequency characteristic is over emphasized, and the utility value to the field of fault diagnosis is low.
Disclosure of Invention
The invention aims to solve the problems that the existing association rule mining algorithm is low in efficiency of detecting faults of an aero-engine and an effective inference rule cannot be screened out in the aspect of fault diagnosis, and provides an aero-engine fault diagnosis method and system based on extension association rule mining.
An aeroengine fault diagnosis method based on extension association rule mining comprises the following steps:
step (ii) ofThe method comprises the steps of acquiring historical data in the operation process of an aircraft engine, wherein the historical data comprises: life calculation, flight record, fault condition and supply chain information; establishing an aeroengine fault affair set B based on historical data, namely B ═ B1,B2,...,BnIn which B1,B2,...,BnRepresenting different fault transaction types, and expanding the different fault transaction types into different fault characteristics, namely Bi＝{Oi,{(C1,V1),(C2,V2),...,(Cn,Vn) } in which O isiRepresenting the name of the ith fault; ciIs characteristic of the i-th fault condition, ViIs represented by CiN, n represents the number of fault features;
calculating the correlation between every two fault features by using a Pearson correlation coefficient, setting a correlation threshold value, eliminating one fault feature in a fault feature pair higher than the correlation threshold value, and taking the residual fault features as key fault features;
extracting frequent items from key fault features by using a frequent item set mining algorithm, and extracting strong association rules from the frequent items, wherein the specific process comprises the following steps:
step three, setting a minimum support threshold Sup _ min, and combining fault transactions which are greater than or equal to the minimum support threshold into a frequent 1-item set;
step three, sorting the fault characteristics in the frequent 1-item set according to the descending order of the support degree to form an ordered set Bs；
Thirdly, removing the infrequent affair set B in the failure affair set of the aero-engine by using an extension transformation algorithm0To obtain a set B-B0；
Step three and four, according to the ordered set BsSet of sequential pairs of B-B0Each transaction in the list is ordered to generate an ordered frequent item set [ P | P]Wherein P is the set [ P | P]P is the rest item;
step three, taking the T as a root node, embedding nodes ([ P | P ], T) by adopting a recursion method, and generating an FP-Tree;
step three, generating a conditional mode base based on the nodes of the FP-Tree, and mining all frequent item sets by the conditional mode base;
calculating the confidence coefficients of all frequent items in the frequent item set, setting a confidence coefficient threshold value, taking the frequent items with the confidence coefficients higher than the confidence coefficient threshold value, and extracting a strong association rule according to the frequent items with the confidence coefficients higher than the confidence coefficient threshold value;
and fourthly, sequencing the strong association rules based on an extension evaluation method, and diagnosing the faults according to the sequence of the strong association rules.
An aircraft engine fault diagnosis system based on the development of the development association rule comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the processor realizes an aircraft engine fault diagnosis method based on the development of the development association rule when executing the computer program.
The invention has the beneficial effects that:
the method can solve the problems of single evaluation index, easy generation of invalid rules, low mining efficiency and the like of the traditional association rule mining. The invention integrates the extension transformation method, the extension evaluation method and the association rule mining algorithm, comprehensively analyzes from multiple dimensions such as fault frequency characteristic, flight safety, economic applicability, maintainability and the like, effectively sequences the association rules through the extension, greatly improves the efficiency of detecting the faults of the aero-engine, widens the evaluation index of the association rule mining, and improves the effectiveness and the practicability of the fault analysis of the aero-engine.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a flow chart of a modeling of a failure transaction set for an aircraft engine;
FIG. 3 is a diagram of an extensible transaction divergence E-R relationship;
FIG. 4 is a process of constructing an extension transformation FP-Tree;
FIG. 5 is a transaction set B structure composition;
fig. 6 is a process of the extension evaluation method for the risk assessment of engine failure.
Detailed Description
The first embodiment is as follows: the extension discipline is an original transection discipline, wherein an extension primitive model objectively describes data characteristics from the aspects of objects, characteristics, magnitude and the like, and provides powerful support for transaction set modeling; the extension transformation method can effectively transform the domain of discourse, can simplify the traversal space of the FP-Tree and improve the operation efficiency; the extension degree evaluation method is mainly used for multi-level and multi-dimensional evaluation index extension and provides a new idea for utility evaluation of association rule mining.
Therefore, the invention provides an aero-engine fault analysis method based on extension association rule mining, which introduces theories such as an extension primitive model, an extension transformation method and an extension goodness evaluation method, improves mining efficiency of an FP-Tree association rule mining algorithm, ranks priorities of association rules from multi-dimensional evaluation indexes, improves utility value of reasoning results, and better guides quality tracing of fault diagnosis.
The embodiment is specifically described with reference to fig. 1 to 6, and the method for analyzing the failure of the aircraft engine based on the extension association rule mining of the embodiment includes the following steps:
the method comprises the following steps of firstly, obtaining historical data of an aircraft engine operation and maintenance process, wherein the historical data comprises: life calculation, flight records, fault condition, supply chain information, and the like; establishing an aircraft engine fault affair set B, wherein the B is (O, C, V), and O represents the name of a fault part; c is a fault piece characteristic and represents a fault phenomenon; v is a characteristic quantity value used for quantitative analysis of fault transactions; according to a multi-object multi-feature divergence rule, carrying out divergence processing on an aeroengine fault affair set; the specific process comprises the following steps:
as shown in fig. 3, in conjunction with the configuration of the aircraft engine, the set of aircraft engine fault transactions B can be expanded into different fault types, i.e., B ═ { B, using a multi-object divergence method1,B2,...,BnIn which B1,B2,...,BnIndicating different fault types, e.g. high-pressure compressor fault B1Fault B of low pressure compressor2High pressure turbine failure B3Low pressure turbine fault B4Air inlet passage fault B5And accessory failure B6Etc.; then according to the neutral bill of materials, the engine maintenance record and the fault information record, the fault affair can be expanded into different fault characteristics by using a multi-characteristic divergence method, namely Bi＝{Oi,{(C1,V1),(C2,V2),...,(Cn,Vn) } in which O isiRepresenting the name of the ith fault; o isi,{(C1,V1),(C2,V2),...,(Cn,Vn) Denotes the failure transaction details of the ith failed piece, CiIs characteristic of the i-th fault condition, ViIs represented by CiN, n represents the number of fault features; for example, the number of influences C1Consequence of failure C2Engine type C3Fault location C4Failed component lot C5Time length of operation of a faulty element C6Nature of failure C7Etc.;
calculating linear correlation between every two fault characteristics by adopting a Pearson correlation coefficient (Pearson), setting a correlation threshold value, and taking the fault characteristics higher than the correlation threshold value as key fault characteristics;
the method comprises the steps that fault affairs of the aircraft engine contain a large amount of multi-source heterogeneous data, different expression modes may exist in the same characteristic information, such as fault consequences and fault properties, in order to solve the problem of consistency of characteristic expression, a correlation analysis method is adopted to prune different fault characteristics, namely, Pearson correlation coefficients (Pearson) are adopted to calculate linear correlation among fault characteristics, the fault characteristics with the Pearson coefficients higher than a minimum threshold value are high linear correlation, and the characteristic quantity needs to be reduced in dimension so as to facilitate subsequent correlation rule mining;
by means of a multi-object multi-feature divergence method, aeroengine fault affairs are described in detail from the qualitative and quantitative aspects, and multi-source heterogeneous original fault data are effectively integrated;
thirdly, frequent item set mining is carried out on key fault features, the high-frequency relation among the fault features of the aircraft engine is analyzed, each fault feature is regarded as an item, a historical record containing a plurality of fault features is a fault affair, the plurality of fault affairs form an aircraft engine fault element, and the recessive relation among the fault affairs can be found through the frequent item set mining; firstly, establishing a mapping relation between an aircraft engine fault affair and a frequent pattern Tree (FP-Tree), then, integrating an extension transformation method into a traditional FP-Tree structure, and optimizing a frequent item search process so as to improve the algorithm mining efficiency, and finally, setting a confidence threshold value, and screening out a strong association rule from the frequent item according to the confidence; the FP-Tree algorithm is used for storing frequent item information without traversing layer by layer along the width direction, but generating Tree nodes along the depth direction, and has good mining efficiency when facing a large-scale fault data set;
and in the third step, frequent item set mining is carried out on the key fault characteristics, and the specific process comprises the following steps:
step three, setting a minimum support threshold Sup _ min, and scanning an aircraft engine fault affair set B; the support degree characterizes the occurrence frequency of the transaction combination, the higher the support degree is, the more easily the transactions occur, and the high-frequency transactions are screened out according to the minimum support degree;
step two, taking the fault characteristics higher than Sup _ min as a frequent 1-item set, and generating an ordered set B in descending order according to the support degrees；
Thirdly, optimizing the search space of the aeroengine failure affair set B, namely adopting an extension transformation method to eliminate the non-frequent affair set B0To obtain a set B-B0；
Step three and four, the optimized transaction set B-B0Scan again and follow set BsThe order of (2) sorts the contained items of each transaction to generate a set [ P | P [ ]]Wherein P is the set [ P | P]P is the rest item;
step three, taking T as a root node, and constructing an FP-Tree, namely an embedded node ([ P | P ], T) in a mode of recursively calling a child node (recursively calling is a common known method for computer programming);
step three, generating a conditional mode base based on the FP-Tree node path, and mining all frequent item sets by the conditional mode base;
calculating the confidence coefficients of all frequent items in the frequent item set, setting a confidence coefficient threshold value, taking the frequent items with the confidence coefficients higher than the confidence coefficient threshold value, and extracting a strong association rule according to the frequent items with the confidence coefficients higher than the confidence coefficient threshold value; the association rule represents the probability of pushing down another frequent item set under the condition of a certain frequent item set, and if the confidence coefficient of the association rule is greater than or equal to the minimum confidence coefficient, the association rule is a strong association rule);
according to the process, the fault association rule is generated based on the extension transformation FP-Tree, and the invalid search space of the original transaction set is removed by adopting the extension transformation method, so that the construction efficiency of the FP-Tree is improved;
and fourthly, sequencing the strong association rules based on an extension evaluation method, and diagnosing the faults according to the sequence of the strong association rules.
The second embodiment is as follows: the difference between the present embodiment and the specific embodiment is that the linear correlation between each two fault features is calculated by using Pearson correlation coefficient (Pearson), and the calculation formula is as follows:
wherein,
represents a feature CXAnd feature CYThe correlation of (c);
represents a feature CXStandard deviation of (d);
express characterSign CYStandard deviation of (d); cov denotes covariance; x, Y is 1 to n, X.noteq.Y.
Pruning is carried out by a Pearson correlation analysis method, fault features with high linear correlation are removed, and invalid feature items are reduced.
Other steps and parameters are the same as those in the first embodiment.
The third concrete implementation mode: the difference between this embodiment and the first or second embodiment is that step three adopts an extension transformation method to eliminate the infrequent transaction set B0To obtain a set B-B0The specific process comprises the following steps:
when a traditional FP-Tree scans a failure transaction set B, all transactions are traversed one by one, and a large amount of time is consumed, so that the invention adds a detachable transformation algorithm on the basis of the traditional FP-Tree structure;
assume that subset B exists in transaction set B0It can be expressed as:
B0＝{T01,T02…T0m}
wherein, T01,T02…T0mThe transaction is defined as an infrequent transaction, namely all the characteristics contained in the transaction are all infrequent items, and the FP-Tree algorithm step shows that the frequent item set does not contain the infrequent items, namely the infrequent transaction T does not need to be scanned01,T02…T0m；
Therefore, the original search space of the transaction set B is optimized by adopting an extension transformation method, and the subset B formed by infrequent transactions is removed0It can be expressed as:
T-B＝B-B0
wherein, T-And the reduction transformation in the extension transformation is represented, and the traversal range of the original transaction set is reduced in the process of domain-of-discourse transformation, so that the construction efficiency of the FP-Tree is improved.
Other steps and parameters are the same as those in the first or second embodiment.
The fourth concrete implementation mode: the difference between this embodiment and the first to third embodiments is that the fourth step ranks the strong association rules based on the popularity evaluation method, and processes the fault according to the ranked strong association rules and the priority order, and the specific process includes:
mining frequent items of fault pieces through an extension transformation FP-Tree algorithm, thereby mining strong association rules among fault transactions; the strong association rule is a rule with higher conditional probability, the representation inference result has higher confidence coefficient, but the generation of engine faults often has strong association with a plurality of fault characteristics, a plurality of rules can be inferred from the same constraint antecedent, for example, a certain engine type and fault part position faults may be related to a plurality of fault part batches, when aiming at the plurality of strong association rules, the traditional method adopts indexes such as confidence coefficient, support degree and the like to carry out priority ranking, but facing the operation and maintenance process of the aero-engine, only depending on the fault frequency characteristic to carry out maintenance guidance is improper, and factors such as maintainability, flight safety, economic applicability and the like of fault parts are needed, therefore, based on the strong association rule mined by extension transformation FP-Tree, further analysis is needed, and a plurality of indexes are adopted to carry out comprehensive ranking on maintenance priority (priority degree of the strong association rule), screening out strong association rules meeting the maintenance requirements;
the method for evaluating the extendibility degree expands evaluation indexes on the basis of support degree and confidence degree analysis by constructing extendibility distances and correlation functions, provides an effective idea for multi-dimensional evaluation of reasoning results, adopts the extendibility degree evaluation method aiming at different evaluation indexes, establishes standard relevance degrees, represents the risk degree of engine faults in different dimensions through the goodness value, and further sorts the fault priority;
the method for evaluating the extendibility degree selects diversified measurement indexes, pays attention to the feasibility and stability of the indexes, and the evaluation process comprises the steps of selection of the measurement indexes, weight coefficient distribution, first evaluation, calculation of the relevance degree and the goodness degree and the like;
step four, constructing an engine fault risk evaluation index set:
the failure affairs of the aero-engine relate to a plurality of links such as airplane management, engine management, outfield maintenance, return to factory overhaul and the like, an evaluation index closely related to the engine failure is reasonably selected by combining an engine configuration and maintenance manual, and an engine failure risk evaluation index set is established and can be expressed as follows:
MI＝{MI1,MI2,…,MIi,…,MIn}
the engine fault risk evaluation index set MI comprises n-dimensional sub-indexes MIiUsing a feature element MIi＝{Ci,ViCharacterizing; the method specifically comprises a frequency characteristic, flight safety, economic applicability and maintainability aiming at the failure risk assessment of the aero-engine, wherein the frequency characteristic can be divided into support degree and confidence degree, the economic applicability can be divided into part cost and maintenance man-hour, and the maintainability is divided into the back constraint and the serial conversion number;
fourthly, carrying out weight distribution on the engine fault risk evaluation indexes, and setting the evaluation index alpha which must be meti；
Step four and step three, based on the evaluation index alpha which must be satisfiediReject below αiThe association rule of (1);
step four, removing the product with the removal rate lower than alpha according to the step four and the step threeiConstructing a corresponding extension correlation function according to the correlation rule, obtaining the extension correlation degree of the multiple indexes based on the extension correlation function, and carrying out normalized processing on the extension correlation degree of the multiple indexes;
and step four, obtaining fault risk items based on the multi-index extension association degrees after the normalization processing, sorting according to the goodness of the fault risk items, and indicating that the fault needs to be processed preferentially by the item with the maximum goodness.
Other steps and parameters are the same as those in one of the first to third embodiments.
The fifth concrete implementation mode: the present embodiment is different from the first to the fourth embodiments in that the fourth step is to perform weight distribution on the engine failure risk evaluation indexes and to set the evaluation index α that must be satisfiediIn particularComprises the following steps:
multiple evaluation index MIiEvaluation of failure risk from multiple dimensions, but different indices MIiDifferences exist in an evaluation system, for example, for a certain fault of an aircraft engine, the importance degree of flight safety is often higher than economic applicability, so that the weights of evaluation indexes have differences, and the weight distribution of multiple evaluation indexes can be represented as:
wherein, alpha is a weight coefficient set and corresponds to the weight of each sub-index of the evaluation index set MI; k represents a weight ordinal number;Λrepresents an evaluation index that must be satisfied; alpha is alphaiIs an evaluation index which must be satisfied, except for alphaiThe sum of the weight coefficients of the other evaluation indexes is 1.
Other steps and parameters are the same as in one of the first to fourth embodiments.
The sixth specific implementation mode: this embodiment is different from the first to fifth embodiments in that the fourth and third steps are based on the evaluation index α that must be satisfiediReject below αiThe specific process of the association rule is as follows:
from the analysis of statistics, the confidence coefficient is the calculation of the conditional probability of the inference result, and the association rule with the confidence coefficient lower than the threshold value indicates that the front and back items of the fault feature have no association and should be removed, so that the fault risk must meet the evaluation index alpha for the first evaluationiThe first evaluation index calculation method can be expressed as follows:
wherein,
as association rules
The strong correlation between the inference result and the fault phenomenon is an index which needs to be met.
Other steps and parameters are the same as those in one of the first to fifth embodiments.
The seventh embodiment: the difference between this embodiment and one of the first to sixth embodiments is that the elimination in the fourth and fourth steps is less than alpha according to the elimination in the fourth and third stepsiThe association rule of (2) constructing a corresponding extension association function, obtaining a multi-index extension association degree based on the extension association function, and carrying out normalized processing on the multi-index extension association degree, wherein the specific process comprises the following steps:
set fault risk item set Z ═ Z1,Z2…ZmAll satisfy the first evaluation, the multi-index extension association degree can be expressed as follows:
Ki＝(Ki(Z1),Ki(Z2)…Ki(Zm)),i＝1,2,…n
wherein m represents the total number of failure risk items;
carrying out normalization processing on the multi-index extension correlation degree:
wherein k isijExpressing the normalized extension correlation degree;
the canonical relevance of each evaluation index of the fault risk item can be expressed as:
ki＝(ki1,ki2…kim),i＝1,2…n
the other steps and parameters are the same as in one of the first to sixth embodiments.
The specific implementation mode is eight: the difference between this embodiment and one of the first to seventh embodiments is that, in the fourth and fifth step, the fault risk items are obtained based on the multiple-index extension association degrees after the normalization processing, and the ranking is performed according to the goodness of the fault risk items, and the specific process includes:
for theSet of fault risk terms Z ═ { Z ═ Z1,Z2…ZmUsing an evaluation index MI ═ MI } ═ MI1,MI2…MInPerforming multidimensional evaluation on each fault risk item, wherein a standard association degree matrix can be expressed as:
designing a fault risk item Z according to the fault operation and maintenance requirements of the aircraft enginejThe goodness calculation formula of (2) is as follows:
wherein, P1(Zj) Expressing goodness-weighting of each evaluation index, P2(Zj) Expressing the goodness of each evaluation index as the minimum value, P3(Zj) The goodness indicating each evaluation index takes the maximum value.
For the fault risk item set Z ═ { Z1,Z2…ZmGoodness of P (Z)j) And comparing and sorting according to the goodness, wherein the item with the maximum goodness indicates that the fault needs to be processed preferentially, and the item with the maximum goodness is represented as follows:
the extension degree evaluation method oriented to the engine fault risk evaluation comprehensively analyzes dimensions such as fault frequency characteristics, flight safety, economic applicability, maintainability and the like, widens evaluation indexes of association rule mining, sequences association rules through extension degrees, represents processing priorities of different fault matters, and is used for assisting maintenance decisions of the aircraft engine.
Other steps and parameters are the same as those in one of the first to seventh embodiments.
The specific implementation method nine: the embodiment of the invention relates to an aircraft engine fault diagnosis system based on extension association rule mining, which comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to realize the steps of the method according to any one of claims 1 to 9.
Claims (10)
1. An aeroengine fault diagnosis method based on extension association rule mining is characterized by comprising the following steps:
the method comprises the following steps of firstly, acquiring historical data in the operation process of the aircraft engine, wherein the historical data comprises the following steps: life calculation, flight record, fault condition and supply chain information; establishing an aeroengine fault affair set B based on historical data, namely B ═ B1,B2,...,BnIn which B1,B2,...,BnRepresenting different fault transaction types, and expanding the different fault transaction types into different fault characteristics, namely Bi＝{Oi,{(C1,V1),(C2,V2),...,(Cn,Vn) } in which O isiRepresenting the name of the ith fault; ciIs characteristic of the i-th fault condition, ViIs represented by CiN, n represents the number of fault features;
calculating the correlation between every two fault characteristics by using a Pearson correlation coefficient, setting a correlation threshold value, eliminating one fault characteristic in a fault characteristic pair higher than the correlation threshold value, and taking the residual fault characteristics as a key fault;
extracting frequent items from key fault features by using a frequent item set mining algorithm, and extracting strong association rules from the frequent items, wherein the specific process comprises the following steps:
step three, setting a minimum support threshold Sup _ min, and combining fault transactions which are greater than or equal to the minimum support threshold into a frequent 1-item set;
step three, sorting the fault characteristics in the frequent 1-item set according to the descending order of the support degree to form an ordered set Bs；
Thirdly, removing the infrequent affair set B in the failure affair set of the aero-engine by using an extension transformation algorithm0To obtain a set B-B0；
Step three and four, according to the ordered set BsSet of sequential pairs of B-B0Each transaction in the list is ordered to generate an ordered frequent item set [ P | P]Wherein P is the set [ P | P]P is the rest item;
step three, taking the T as a root node, embedding nodes ([ P | P ], T) by adopting a recursion method, and generating FP-Tree by the nodes;
step three, generating a conditional mode base based on the nodes of the FP-Tree, and mining all frequent item sets by the conditional mode base;
calculating the confidence coefficients of all frequent items in the frequent item set, setting a confidence coefficient threshold value, taking the frequent items with the confidence coefficients higher than the confidence coefficient threshold value, and extracting a strong association rule according to the frequent items with the confidence coefficients higher than the confidence coefficient threshold value;
and fourthly, sequencing the strong association rules based on an extension evaluation method, and diagnosing the faults according to the sequence of the strong association rules.
2. The method for diagnosing the faults of the aero-engine based on the mining of the extension association rule as claimed in claim 1, wherein in the second step, correlation between every two fault features is calculated by using Pearson correlation coefficients, and the formula is as follows:
wherein,
represents a feature CXAnd feature CYThe correlation of (c);
represents a feature CXStandard deviation of (d);
represents a feature CYStandard deviation of (d); cov denotes covariance; x, Y is 1 to n, X.noteq.Y.
3. The method for diagnosing the faults of the aero-engine based on the mining of the extension association rules as claimed in claim 2, wherein the third step is to remove the infrequent affair set B in the fault affair set of the aero-engine by using the extension transformation method0To obtain a set B-B0The specific process comprises the following steps:
assume that there is a set of infrequent transactions B in the set of faulty transactions B0Then infrequent transactions set B0Expressed as:
B0＝{T01,T02…T0m}
wherein, T01,T02…T0mIs an infrequent transaction;
rejecting subsets B formed by infrequent transactions0Expressed as:
T-B＝B-B0
wherein T-represents a subtractive transform in the extension transform.
4. The aero-engine fault diagnosis method based on the extension association rule mining as claimed in claim 3, wherein the fourth step is to rank the strong association rules based on an extension degree evaluation method, and the specific process includes:
step four, constructing an engine fault risk evaluation index set, and recording as MI ═ MI1,MI2,…,MIi,…,MInIn which MI1,MI2,…,MIi,…,MInSub-indexes with n dimensions in the engine fault risk evaluation index set, wherein each sub-index corresponds to a fault affair, and MIi＝{Ci,Vi}；
Fourthly, carrying out weight distribution on the engine fault risk evaluation indexes, and setting the evaluation index alpha which must be meti；
Step four and step three, based on the evaluation index alpha which must be satisfiediReject below αiThe association rule of (1) is reserved, and is greater than or equal to alphaiThe association rule of (1);
step four, aiming at alpha being more than or equal toiConstructing a corresponding extension correlation function according to the correlation rule, obtaining the extension correlation degree of the multiple indexes based on the extension correlation function, and carrying out normalized processing on the extension correlation degree of the multiple indexes;
and step four, obtaining fault risk items based on the multi-index extension relevance degrees after the normalization processing, and sorting according to the goodness of the fault risk items, wherein the item with the largest goodness is the fault to be processed preferentially.
5. The method for diagnosing the faults of the aero-engine mined based on the extension association rule as claimed in claim 4, wherein the four steps are carried out for distributing the weight of the risk evaluation indexes of the faults of the aero-engine, and the evaluation indexes alpha which must be met are setiThe method specifically comprises the following steps:
wherein alpha is a weight coefficient set and corresponds to the weight of each sub-index in the evaluation index set MI; k represents a weight ordinal number; alpha is alphaiExcept for alpha for the evaluation index which must be satisfiediThe sum of the weight coefficients of the other evaluation indexes is 1.
6. The method for diagnosing the faults of the aero-engine mined based on the extension association rule according to claim 5, wherein the fourth step and the third step are based on the evaluation index alpha which must be metiReject below αiThe association rule of (2) is specifically:
wherein,
representing association rules
The relevance of (c).
7. The method for diagnosing the faults of the aero-engine based on the extended association rule mining as claimed in claim 6, wherein the fourth step is performed for alpha or moreiThe association rule of (2) constructing a corresponding extension association function, obtaining a multi-index extension association degree based on the extension association function, and carrying out normalized processing on the multi-index extension association degree, wherein the specific process comprises the following steps:
suppose the fault risk item set Z ═ { Z ═ Z1,Z2…ZmAll satisfy the evaluation index alphaiThen, the multi-index extension association degree is expressed as:
Ki＝(Ki(Z1),Ki(Z2)…Ki(Zm)),i＝1,2,…n
where m represents the total number of failure risk items.
8. The method for diagnosing the faults of the aero-engine based on the mining of the extension association rule according to claim 1 or 7, wherein the normalizing the multi-index extension association degree specifically comprises:
the standard relevance of each evaluation index of the fault risk item is as follows: k is a radical ofi＝(ki1,ki2…kim),i＝1,2…n。
9. The method for diagnosing the faults of the aero-engine mined based on the extension association rule according to claim 8, wherein the fourth step and the fifth step obtain fault risk items based on the multi-index extension association degrees after the normalization processing, the fault risk items are sorted according to the goodness of the fault risk items, and the item with the largest goodness, namely the fault, needs to be processed preferentially, and the specific process includes:
wherein, C1(Zj) Representing goodness weighting of each evaluation index, C2(Zj) The goodness indicating each evaluation index takes the minimum value, C3(Zj) Expressing the maximum goodness of each evaluation index; k (Z)j) Representing a canonical correlation matrix;
setting the fault risk item set Z as Z1,Z2…ZmSorting the goodness, the item with the largest goodness being the fault to be processed preferentially:
10. an aircraft engine fault diagnosis system based on extended association rule mining, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1 to 9 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111063095.8A CN113761032A (en) | 2021-09-10 | 2021-09-10 | Aeroengine fault diagnosis method and system based on extension association rule mining |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111063095.8A CN113761032A (en) | 2021-09-10 | 2021-09-10 | Aeroengine fault diagnosis method and system based on extension association rule mining |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113761032A true CN113761032A (en) | 2021-12-07 |
Family
ID=78794779
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111063095.8A Pending CN113761032A (en) | 2021-09-10 | 2021-09-10 | Aeroengine fault diagnosis method and system based on extension association rule mining |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113761032A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114637793A (en) * | 2022-05-19 | 2022-06-17 | 四川观想科技股份有限公司 | Equipment fault frequent region positioning method based on big data analysis |
| CN114659785A (en) * | 2021-12-27 | 2022-06-24 | 三一重能股份有限公司 | Fault detection method and device for transmission chain of wind driven generator |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112528519A (en) * | 2020-05-14 | 2021-03-19 | 北京赛宝工业技术研究院有限公司 | Method, system, readable medium and electronic device for engine quality early warning service |
| US11055405B1 (en) * | 2019-04-30 | 2021-07-06 | Splunk Inc. | Anomaly event detection using frequent patterns |
| CN113125937A (en) * | 2021-03-09 | 2021-07-16 | 浙江零跑科技有限公司 | Rapid early warning method for operation fault of vehicle-mounted printed circuit board assembly |
| CN113147476A (en) * | 2020-12-31 | 2021-07-23 | 中国电力科学研究院有限公司 | Fault analysis and early warning method and system for electric vehicle charging equipment |
-
2021
- 2021-09-10 CN CN202111063095.8A patent/CN113761032A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11055405B1 (en) * | 2019-04-30 | 2021-07-06 | Splunk Inc. | Anomaly event detection using frequent patterns |
| CN112528519A (en) * | 2020-05-14 | 2021-03-19 | 北京赛宝工业技术研究院有限公司 | Method, system, readable medium and electronic device for engine quality early warning service |
| CN113147476A (en) * | 2020-12-31 | 2021-07-23 | 中国电力科学研究院有限公司 | Fault analysis and early warning method and system for electric vehicle charging equipment |
| CN113125937A (en) * | 2021-03-09 | 2021-07-16 | 浙江零跑科技有限公司 | Rapid early warning method for operation fault of vehicle-mounted printed circuit board assembly |
Non-Patent Citations (2)
| Title |
|---|
| NINGZHE XING 等: "An Early Warning Analysis Technique of Electric Equipment Temperature Based on Association Rule Mining", 2018 IEEE INTERNATIONAL CONFERENCE ON ENERGY INTERNET (ICEI), 5 July 2018 (2018-07-05), pages 258 - 263 * |
| 黄明华: "可拓检测中可测信息的有效性及其挖掘研究", 中国优秀硕士学位论文全文数据库, no. 10, 15 October 2011 (2011-10-15), pages 1 - 68 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114659785A (en) * | 2021-12-27 | 2022-06-24 | 三一重能股份有限公司 | Fault detection method and device for transmission chain of wind driven generator |
| CN114637793A (en) * | 2022-05-19 | 2022-06-17 | 四川观想科技股份有限公司 | Equipment fault frequent region positioning method based on big data analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| DE60315506T2 (en) | IDENTIFICATION OF CRITICAL FEATURES IN A REGIONAL SCALE ROOM | |
| Croft et al. | Retrieving documents by plausible inference: an experimental study | |
| US7590642B2 (en) | Enhanced K-means clustering | |
| US7174343B2 (en) | In-database clustering | |
| US7080063B2 (en) | Probabilistic model generation | |
| US20030212702A1 (en) | Orthogonal partitioning clustering | |
| US5815713A (en) | Method and system for automated transformation of declarative language process specifications | |
| Ciaccia et al. | Processing complex similarity queries with distance-based access methods | |
| Ayouni et al. | Extracting compact and information lossless sets of fuzzy association rules | |
| CN113434418A (en) | Knowledge-driven software defect detection and analysis method and system | |
| CN113761032A (en) | Aeroengine fault diagnosis method and system based on extension association rule mining | |
| JP2001117947A (en) | Method for sorting high dimensional data, sort tree formed by the method and computer system | |
| US7174336B2 (en) | Rule generation model building | |
| Lin et al. | Mining multiple fuzzy frequent patterns with compressed list structures | |
| Maddouri et al. | Incremental rule production: Towards a uniform approach for knowledge organisation | |
| CN113918727A (en) | Construction project knowledge transfer method based on knowledge graph and transfer learning | |
| CN113742396A (en) | Mining method and device for object learning behavior pattern | |
| Ravi et al. | Clustering of symbolic objects using gravitational approach | |
| US5758032A (en) | Method and system for automated transformation of declarative language process specification | |
| Malik et al. | A comprehensive approach towards data preprocessing techniques & association rules | |
| Adhikari et al. | Study of select items in different data sources by grouping | |
| Müller et al. | HSM: Heterogeneous subspace mining in high dimensional data | |
| Yahia et al. | K-nearest neighbor and C4. 5 algorithms as data mining methods: advantages and difficulties | |
| Khoshgoftaar et al. | Data mining of software development databases | |
| Sassi et al. | About database summarization |
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 |