CN101950382A - Method for optimal maintenance decision-making of hydraulic equipment with risk control - Google Patents

Method for optimal maintenance decision-making of hydraulic equipment with risk control Download PDF

Info

Publication number
CN101950382A
CN101950382A CN2010102724461A CN201010272446A CN101950382A CN 101950382 A CN101950382 A CN 101950382A CN 2010102724461 A CN2010102724461 A CN 2010102724461A CN 201010272446 A CN201010272446 A CN 201010272446A CN 101950382 A CN101950382 A CN 101950382A
Authority
CN
China
Prior art keywords
risk
item sets
value
equipment
hoe
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN2010102724461A
Other languages
Chinese (zh)
Other versions
CN101950382B (en
Inventor
刘晶
蔡大勇
季海鹏
朱清香
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.)
Tianjin Development Zone JINGNUO Ocean Data Technology Co. Ltd.
Original Assignee
Yanshan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yanshan University filed Critical Yanshan University
Priority to CN2010102724461A priority Critical patent/CN101950382B/en
Publication of CN101950382A publication Critical patent/CN101950382A/en
Application granted granted Critical
Publication of CN101950382B publication Critical patent/CN101950382B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention belongs to the field of maintenance decision-making of hydraulic equipment, and relates to a method for the optimal maintenance decision-making of hydraulic equipment with risk control. The method mainly comprises three steps: 1) judging whether a system is in a status of defect by using a variable-weight association rule algorithm, if so, calculating the probability values of occurrences of latent faults of the system; 2) calculating the comprehensive evaluation value for the consequence of each latent fault by using a BP neural network; and 3) multiplying the probability values obtained in step 1 by the comprehensive evaluation values obtained in step 2 so as to obtain the VaRs (values-at-risk) of the latent faults, judging whether the VaRs are more than a specified threshold, if so, ranking the VaRs in descending order so as to determine the maintenance sequence; otherwise, returning to the step of monitoring. The method can judge whether a device is in a status of defect, judge the type of the latent fault and calculate the probability values of occurrences of latent faults only through a calculation; and compared with traditional risk maintenance methods, the method of the invention improves the accuracy of fault diagnosis, speeds up the diagnosis speed, and provides a better reference for online decision-making.

Description

A kind of optimum repair determining method of hydraulically operated equipment (HOE) that has risk control
Technical field
The invention belongs to hydraulically operated equipment (HOE) maintenance decision field, relate to a kind of optimum repair determining method of hydraulically operated equipment (HOE) that has risk control.
Background technology
Hydraulically operated equipment (HOE) has complexity, precision, price height, high-power characteristics, and the working condition of hydraulically operated equipment (HOE) has determined the quality of production efficiency and smelting iron and steel simultaneously, and its security and reliability requirement are than higher.Because each Hydraulic Elements is worked in airtight oil circuit in the hydraulic system, the flow state of pipeline inner fluid and the situation of inner body can't Direct observation, and therefore, the fault diagnosis of hydraulic system is more more difficult than the fault diagnosis of common mechanical, electrical equipment.To hydraulically operated equipment (HOE) system real time and on line monitoring, set up effectively, fault diagnosis and early warning system seem very important accurately.
Existing method has all obtained certain effect in the application of reality, but exist some limitation, mainly as follows: 1, various information detection meanss and method for early warning all fail to regard diagnosis object as an organic whole, and that fails effectively to consider may to exist between each parts of equipment connects each other and influence.2, be difficult to the complex situations handling various faults and deposit.In the equipment failure evolution process of reality, contact is closely arranged between each parts of system, various faults often take place simultaneously, so art methods also is difficult to obtain predicted value comparatively accurately.
Risk is for the quantification perception that may have an accident future, is a kind of for comprehensive assessment uncertain and loss property.In simple terms, risk is exactly the product of contingency occurrence probability and damage sequence.And risk assessment is on the basis of identification risk, to risk measure, analyses such as comparison, judgement and ordering, thereby for formulating the foundation that the precautionary measures and management decision provide science.Risk maintenance (Risk BasedMaintenance is called for short RBM) is based on venture analysis and evaluation and the method for working out maintenance policy.Risk maintenance also is to serve as to pass judgment on the maintenance policy management mode on basis with the risk that equipment or parts are handled.The development of maintenance of equipment pattern and technical system is divided into four-stage, that is: correction maintenance, scheduled maintenance, state maintenance and risk maintenance, this shows, risk maintenance as follow-on be method for maintaining (the Reliability Centered Maintenance at center with the reliability, be called for short RCM), be the developing direction of modern comfort maintenance management.The RBM method has been successfully applied in the large enterprises such as refinery, chemical plant, petrochemical plant, adopts the RBM technology generally to can be enterprise and reduces overhaul of the equipments amount and maintenance cost 15% to 40%.Risk maintenance improves on-road efficiency down, optimization life-cycle period expense not jeopardizing security and influence the preceding of environment.By setting up the risk maintenance system, risk precedence that can each equipment of identification reduces period expense equipment life, reduce equipment breakdown and fault, reach the congruence of maintenance and reduction risk, both considered the security of equipment, considered the repair and maintenance cost of equipment again.
The main flow process of risk maintenance comprises following four steps:
1) system divides and function modeling are determined research object;
2) failure risk analysis and assessment comprise the analysis and the calculating of value-at-risk after probability of malfunction, the fault;
3), determine corresponding maintenance policy based on decision in the face of risk;
4) enforcement of maintenance policy.
The risk method for maintaining mainly contains: Failure Mode Effective Analysis method (FMEA), fault tree analysis (FTA).
Failure Mode Effective Analysis method (Failure Mode and Effects Analysis is called for short FMEA) is a kind of important method of reliability design, is the combination of FMA (failure mode analysis (FMA)) and FEA (failure effect analysis (FEA)).FMEA estimates, analyzes various possible risks, so that eliminate these risks on the basis of existing technology or these risks are reduced to acceptable level.
FMEA is the activity of one group of seriation, and its process comprises: find out fault mode potential in product/process; According to corresponding appraisement system the incipient fault pattern of finding out is carried out the risk quantification assessment; List fault cause/mechanism, seek prevention or innovative approach.FMEA has obtained effect preferably in the application of reality, but FMEA makes analysis at a particular elements, rather than performs an analysis at a process or device systems, therefore has locality, is not suitable for being applied in the hydraulically operated equipment (HOE).
Fault tree analysis (Failute Tree Analysis is called for short FTA), a kind of from the system to parts, arrive part again, by the method for " decline shape " analysis.FTA is launched into tree-shaped branch figure from system gradually by of being drawn out by logical symbol, the probability that comes analysis of failure incident (claiming the top incident again) to take place.Fault tree analysis has obtained in atomic pile, main equipment and the giant brain system using widely in design, the maintenance of aerospace.
The structure fault tree is a step the most key in the fault tree analysis.Usually to and use the maintenance personal to coact by designer, reliability Work personnel, by careful comprehensive and analysis, find out the system failure and the logical relation that causes the factors of this fault of system, and with this relation with specific graphical symbol, be that incident symbol and logical symbol show, becoming with the top incident is " root " long downwards one tree-fault tree.
Fault tree analysis both can be carried out qualitative analysis, can carry out quantitative test and system evaluation again.By qualitative analysis, determine the size of each elementary event to accident impact, thereby can determine to each elementary event carry out security control the priority that should take measures, provide basic foundation for formulating scientific and rational security control measure.By quantitative test, according to the probability that each elementary event takes place, calculate to push up and go up the probability that incident (accident) takes place, for the best safety controlled target that realizes system provides a concrete notion of measuring, help the quantification treatment of all other indexs.But when utilizing FTA that system is carried out quantitative test, must determine the probability that all each elementary events take place in advance, all will recomputate at every turn that operand is big, speed is slow, thereby is not suitable for the inline diagnosis of hydraulically operated equipment (HOE).
Correlation rule (Association Rule) is the mutual relationship that is hidden between the data in order to excavate, and finds out the rule that all can connect one group of incident or data item and another group incident or data item.
Excavate the basic ideas of correlation rule: a given affairs collection, the task of excavating correlation rule generate support (support) and degree of confidence (confidence) exactly respectively greater than the correlation rule of given minimum support of user (minsupp) and min confidence (minconf).The rule that satisfies minimum support, min confidence and degree of correlation requirement is called strong rule.Seeking out all effectively strong rules is exactly the task that the correlation rule data mining will be finished.
If data things collection D, I={i 1, i 2..., i mBe the set of item, element wherein is called (item).Note D is the set of item for the set of transaction T, the T that concludes the business here, and
Figure BSA00000257401000041
Corresponding each transaction has unique sign.If x is the set of an I discipline, if
Figure BSA00000257401000042
Claim transaction T to comprise x so.
Correlation rule be shape as
Figure BSA00000257401000043
Implications, here
Figure BSA00000257401000044
Figure BSA00000257401000045
And x ∩ y=Ф.
Definition one:
Figure BSA00000257401000046
Support in the things database D is that things is concentrated and to be comprised the things number of x and y and the ratio of all things numbers, is designated as support (x ∪ y), that is:
support ( x ∪ y ) = | { T : x ∪ y ⊆ T , T ∈ D } | / | D |
Definition two: The confidence level of concentrating in things is meant the ratio of number of transactions that comprises x and y and the things number that comprises x, is designated as That is:
confidence ( x ⇒ y ) = | { T : x ∪ y ⊆ T , T ∈ D } | / | { T : x ⊆ T , T ∈ D } |
Definition three: if X ∪ Y is a Frequent Item Sets, then degree of confidence is not less than the minimum letter threshold value minconf that puts.
The core concept of Apriori algorithm constantly increases by the Item Sets element number progressively finds Frequent Item Sets.Promptly utilize the alternative manner of priori by successively searching for of Frequent Item Sets character, be about to the k-Item Sets and be used for exploring (k+1)-Item Sets, all Frequent Item Sets of coming the limit data set.At first produce frequent 1-Item Sets L 1, be frequent 2-Item Sets L then 2, algorithm stops up to no longer expanding the element number of Frequent Item Sets.In the k time circulation, process produces the set C of candidate k-Item Sets earlier k, generate support and the frequent k-Item Sets L of test generation by scan database then k
The Apriori algorithm mainly was divided into for two steps, i.e. link and beta pruning.
Connect the step: for finding L K+1, need be with L kIn two item collection connect to obtain a L K+1Candidate collection C K+1If l 1And l 2Be L kIn two item collection, use mark l I[j]Expression l iIn j project; For convenience of description, suppose generally in the transaction data base that every in each transaction record all sorts by dictionary.The symbolically attended operation.If l 1And l 2In preceding k-1 item be identical, i.e. (l 1[1]=l 2[1]) ... (l 1[k-2]=l2 [k-2]) (l1 [k-1]=l2 [k-1])), L then kMiddle l 1And l 2Content just can connect together.Condition (l 1[k]＜l 2[k]) can guarantee that the item that does not produce repetition collects.L then 1And l 2The result who connects is { l 1[1]l 1[2]L 1[k-1]l 1[k]l 2[k].
The beta pruning step (deletion): candidate k-Item Sets C kBe frequent k-Item Sets L kSuperset; Be that wherein each element differs that to establish a capital be Frequent Item Sets, but all frequent k-Item Sets are included in wherein.That is:
Figure BSA00000257401000051
Scan database can be determined C kIn the support frequency of each candidates collection, thereby determine L kIn the frequent k-Item Sets of each element, promptly all frequency are not less than minimum support
The correlation rule that uses in the equipment fault diagnosis at present, be to have on the identical importance basis at each assembly of equipment to propose, and actual situation is, the status of each assembly in entire equipment of equipment is different, different assemblies has the branch of different importance and tangible primary and secondary in equipment, so this prerequisite hypothesis can cause the inaccurate of fault diagnosis result.
Weighted association rules has been introduced the notion of weights, gives different weights with the project in the database, and this has just reflected the difference of projects significance levels.So the good method that the proposition of weighted association rules overcomes the above problems beyond doubt is applied in the Research of Equipment Fault Diagnose, gives different weights with each assembly of equipment, the problem that just can avoid the assembly of different significance levels to put on an equal footing.
Transaction data base D is investigated in the definition of weighted association rules, and the set of its project is I={i 1, i 2..., i n, each transaction all is the subclass of I, and composes with a transaction identifiers TID.
Define 1. correlation rule shapes as
Figure BSA00000257401000052
Wherein
Figure BSA00000257401000053
Figure BSA00000257401000054
And
Define 2. correlation rule shapes as
Figure BSA00000257401000056
Support be the probability that X ∪ Y comprises at transaction data base.
Define 3. correlation rule shapes as Degree of belief in certain transaction, comprising the probability that also comprises Y under the prerequisite of X simultaneously.
Changing a kind of more popular saying is exactly, correlation rule shape as
Figure BSA00000257401000058
Support be the ratio that comprises number of deals with the total number of deals of X ∪ Y in the database; Correlation rule shape as
Figure BSA00000257401000059
Degree of belief be the ratio that comprises number of deals with the number of deals that comprises X of X ∪ Y in the database.
Given project set I={i 1, i 2..., i n, for characterizing the importance of project, we are each project i j, compose with weight w j, its 0≤wj≤1, j={1,2 ..., n}.
Define 4. correlation rule shapes as
Figure BSA00000257401000061
Weighting support (weighted support) be
[ Σ i j ∈ X ∪ Y w j ] ( support ( X ∪ Y ) )
Definition 5. certain k-Item Sets are called as Frequent Item Sets, support threshold values wminsup if its weighting support is not less than lowest weighted, promptly
[ Σ i j ∈ X ∪ Y w j ] ( support ( X ∪ Y ) ) ≥ w min sup
Define 6. correlation rules
Figure BSA00000257401000064
Be interesting, if X ∪ Y is a Frequent Item Sets, and its degree of belief is not less than minimum trust threshold values minconf.
A given transaction data base, its transaction sum is made as T, and to arbitrary k-Item Sets X, its number of support (support count) is for comprising the number of the transaction of X in the transaction data base, be designated as SC (X) if. certain k-Item Sets X is frequent, and its number of support SC (X) should satisfy following formula so:
SC ( X ) ≥ w min sup × T Σ i j ∈ X w j
Make that I is the set of all items, suppose that Y is a q-Item Sets, q＜k. is in residual term day set (I-Y), and the project of (k-q) individual weights maximum is i before the note R1, i R2..., i Rk-q, the maximum possible value that comprises arbitrary k-Item Sets of Item Sets Y so is:
W ( Y , k ) = Σ i j ∈ Y w j + Σ j = 1 k - q w rj
Wherein, the 1st is the weights sum of projects among the q-Item Sets Y with formula, and the 2nd is remaining preceding (k-q) individual maximum weights sum with formula.
In conjunction with above-mentioned two formulas, we can know by inference: if comprise the k-Item Sets of Y is frequent, and its minimum number of support should be so
Figure BSA00000257401000067
We claim this B (Y k) supports expectation for the k-of Y. consider that (Y k) answers round numbers to B, for the k-Item Sets that guarantees to comprise Y might be frequent, we round up, rather than round downwards. otherwise, can be because SC (Y, value k) crosses low and not enough so that the k-Item Sets becomes Frequent Item Sets.
Neural network has very strong nonlinear fitting ability, can shine upon the nonlinear relationship of any complexity, and learning rules is simple, are convenient to computer realization.Have very strong robustness, memory capability, non-linear mapping capability and powerful self-learning capability, therefore very big application market is arranged.
BP (Back Propagation) neural network, i.e. the learning process of error anti-pass error backpropagation algorithm is made up of the forward-propagating of information and two processes of backpropagation of error.Each neuron of input layer is responsible for receiving the input information that comes from the outside, and passes to each neuron of middle layer; The middle layer is the internal information processing layer, is responsible for information conversion, and according to the demand of information change ability, the middle layer can be designed as single hidden layer or many hidden layers structure; Last hidden layer is delivered to each neuronic information of output layer, after further handling, finishes the once forward-propagating processing procedure of study, by output layer to extraneous output information result.When reality output is not inconsistent with desired output, enter the back-propagation phase of error.Error is by output layer, by each layer of mode correction weights of error gradient decline, to the anti-pass successively of hidden layer, input layer.Information forward-propagating that goes round and begins again and error back propagation process, it is the process that each layer weights are constantly adjusted, also be the process of neural network learning training, the error that this process is performed until network output reduces to the acceptable degree, till the perhaps predefined study number of times.
The input node is the quantified property value of each prediction index of fault consequence value in the BP neural network, and output node is the integrated forecasting value of fault consequence value.By some known fault consequence index attribute value data, neural network is carried out learning training, make the relation between its each data of acquisition failure effect value, in the time need carrying out the consequence value prediction to new sample mode, artificial neural network is by reproducing experience, knowledge and the intuitive thought of training gained, with the quantized value vector input neural network of new sample evaluation index property value, the just exportable failure effect integrated forecasting of network value, thus realization is to the comprehensive consequence value prediction of incipient fault.
Summary of the invention
Though diagnosis is excavated in the equipment failure that is applied to of weighted association rules, solved the different and problem that causes in each assembly of equipment status in equipment.But, weighted association rules also is that each assembly of supposition equipment moves under perfect condition, the weights of its assembly only are endowed weights once, the weights of using in whole weighted association rules algorithm all are initial weights, be invariable, what that is to say employing is weights constant principles throughout one's life.And in the reality, this situation is non-existent, because use along with hydraulically operated equipment (HOE), the seriousness difference of component wear degree, the importance of each assembly in whole hydraulically operated equipment (HOE) must change, the weights of promptly giving apparatus assembly at first will inevitably change, and are very inaccurate if only be used as the lifelong weights of apparatus assembly with the weights of giving at first.
The drawback of in the hydraulically operated equipment (HOE) fault diagnosis, using at weighted association rules, the present invention proposes a kind of change power association rule algorithm DVWAR (Discovery of VariableWeighted Association Rules) that is adapted to the hydraulically operated equipment (HOE) fault diagnosis, this algorithm is to be based upon on the basis of weighted association rules, is intended to solve the problem that each assembly weights of equipment is changed owing to wearing and tearing.And with the DVWAR algorithm application in optimum repair determining method (the optimal maintenance method ofHydraulic Equipment base on risk control of the hydraulically operated equipment (HOE) that has risk control, be called for short OMMHERC) in, this method mainly was divided into for three steps: the first step adopts change power association rule algorithm DVWAR to judge whether system is in defect state, calculates the probable value that its incipient fault takes place if be in defect state; Second step utilized the BP neural network to try to achieve the consequence comprehensive evaluation value of every kind of incipient fault; The probability of malfunction value that the 3rd step was tried to achieve the first step and second step failure effect comprehensive evaluation value of trying to achieve multiplies each other and obtains the incipient fault value-at-risk, judge whether this value surpasses the threshold value of regulation in advance, if surpass the threshold value of regulation in advance, then according to trying to achieve the descending arrangement of failure risk value, to determine service sequence; Continue monitoring otherwise return.
Suppose that equipment is in the running always, perhaps, can be similar to less than Maintenance Demand Time and to think that it is in the running always because the intermediate stop time is shorter.Concrete steps are as follows:
(1) from monitored hydraulically operated equipment (HOE), obtains the status monitoring data;
(2) its data are analyzed and handled, extract eigenwert;
(3) pattern that eigenwert and change are weighed in the related early warning library is complementary, and judges promptly whether hydraulically operated equipment (HOE) the sign of degenerating occurs;
(4) if coupling is unsuccessful, illustrate that then the sign of degenerating does not appear in hydraulically operated equipment (HOE), return and continue the monitoring equipment data;
(5) if the match is successful, illustrate that then the degeneration sign appears in hydraulically operated equipment (HOE), does following processing;
(6) this incipient fault is weighed the probable value that the support of mating in the related early warning library is corresponding incipient fault in change;
(7) with hydraulically operated equipment (HOE) self risk, personal risk, environmental risk, social risk and system risk as input, try to achieve the consequence value of corresponding failure by neural net model establishing;
(8) probable value with incipient fault multiplies each other with corresponding consequence value, obtains the value-at-risk of this incipient fault;
(9) judge whether this value-at-risk surpasses the threshold value of regulation in advance;
(10) do not continue the monitoring hydraulically operated equipment (HOE) if surpass then to return;
(11) if surpassing threshold value then sorts the failure risk value from high to low, to determine optimum service sequence;
(12) carry out shutdown maintenance according to optimizing decision, and return and continue the monitoring equipment data.
Wherein (1)-(5) belong to risk and differentiate the stage, promptly discern potential risk; (6)-(11) belong to the risk assessment stage, promptly quantitatively or qualitatively analyze; (12) belong to the risk control stage, promptly formulate the measure that averts risks accordingly.
The present invention compared with prior art has following beneficial effect:
1, the inventive method is regarded diagnosis object as an organic whole, and that effectively considers may to exist between each parts of hydraulically operated equipment (HOE) connects each other and influence.
2, the inventive method has solved the pre existing survey technology various faults and the complex situations of depositing has been handled coarse problem.
3, the inventive method adopts the Real-time and Dynamic decision-making, adjusts judgment device state of living in real time according to the information in the hydraulically operated equipment (HOE) operational process, and it is more tallied with the actual situation.
4, the weights for each assembly of weighted association rules supposition equipment are invariable in whole life cycle; Each assembly of this and equipment is worn factor in actual motion the fact that weights can change that affects contradicts; A kind of change power association rule algorithm DVWAR (Discovery of Variable Weighted AssociationRules) that is applicable to Fault Diagnosis of Hydraulic Systems has been proposed; And on this basis the variation weight table that produces in the algorithm has been carried out the discretization processing; To improve Fault Diagnosis of Hydraulic Systems accuracy rate and efficiency of algorithm
5, at the singularity of hydraulically operated equipment (HOE), proposition has optimum repair determining method (the optimal maintenance method of Hydraulic Equipment base onrisk control of hydraulically operated equipment (HOE) of risk control, be called for short OMMHERC), whether this method only need be calculated once just can judgment device be in defect state, incipient fault type and incipient fault probability of happening value, compare with traditional risk method for maintaining, improved the fault diagnosis accuracy, accelerated diagnosis speed simultaneously, for on-line decision provides better reference.
Description of drawings
Fig. 1 is a process flow diagram of the present invention
The mathematical function curve of Fig. 2 curve B
The mathematical function curve of Fig. 3 curve A
The weights change curve of Fig. 4 assembly A
The weights change curve of Fig. 5 assembly B
Fig. 6 assembly C weights change curve
The weights change curve of Fig. 7 assembly D
The weights change curve of Fig. 8 assembly E
Fig. 9 OMMHERC/FMEA/FTA time ratio
Embodiment
Below in conjunction with certain iron company's hydraulically operated equipment (HOE) technical scheme of the present invention is further described.
Method flow diagram as shown in Figure 1, concrete steps are as follows:
Certain large hydraulic system of iron company mainly is made up of A, B, three equipment of C, a large amount of service data (comprising normal condition data and fault state data) before having stored in the database of this system comprises four sampled points on three equipment: temperature, pressure, vibrations, rotating speed.
Implementation step is as follows:
The first step makes up to become and weighs related early warning library, and is specific as follows;
From database, obtain the sample data of temperature, pressure, vibrations, rotating speed, clean and remove inconsistent data; Excessive for the attribute that prevents to have higher value with respect to the attribute weight of smaller value, data are carried out normalized; With the data bi-directional scaling of above-mentioned normalized, they are dropped on [0,1], apparatus for establishing monitor data collection again.
Become the tables of data that the power association rule algorithm is handled, represent with boolean's form.And the data of enterprise are the continually varying numerical parameters, therefore the continuity historical data of the monitoring of tools data centralization that previous step is obtained is carried out cluster analysis and discretize, obtain the generalization result of the affiliated cluster scope of Various types of data, thus the monitoring of tools data set after generally being changed.And utilize the monitoring of tools data after generalization are excavated, make up to become and weigh related early warning library; If the probable value P of temperature T, pressure P, vibrations S, rotating speed R and incipient fault 1, P 2, P 3Correlation rule be: X-＞Y[minsup, minconf], wherein X={T, P, S, R}, Y={P 1, P 2, P 3, minsup is a minimum support, minconf is a degree of confidence.
Typical fault rate curve according to equipment, it is tub curve, show equipment age at failure in early days, similar to curve A, and it is similar at chance failure period to curve B, so we are that example is illustrated to follow the highest curve A of rate and B, and the hypothesis equipment failure is a model to observe these two kinds of curve failure rates.At first, curve A and B are redefined with the form of mathematical function, as shown in Figures 2 and 3.
The mathematical function curve of curve B as shown in Figure 2, transverse axis x express time, longitudinal axis y represents the failure rate of assembly, and equipment is represented failure rate with an invariable value in whole life cycle, and establishing this curve is y 0, since 0 moment y 0Just keep a certain steady state value always.
The mathematical function curve of curve A as shown in Figure 3, transverse axis x express time, longitudinal axis y represents the failure rate of assembly, along with the increase or the minimizing of component faults rate, corresponding assembly weights also increase or reduce, this be because of, the significance level of the height decision assembly of failure rate, promptly when this assembly changed the degree of wear in time and increases the weight of, the weights of this assembly also became greatly, and it is big that the probability that breaks down also becomes.From figure, as can be seen, whole A curve can be divided into two sections, time t mPoint is a turning point, at time 0～t mBetween this section curve, establish it for y 1, another section curve is made as y 2According to function definition, can make y 1=ax+b, y 2=cx+d.
Become in the power correlation rule model, basic definition and weighted association rules are basic identical, and difference is the weights of time and variation, in becoming the power association rule algorithm, two variablees have been added, one is the variation of time, and another is the variation of weights, and weights over time specifically.
With the curve A is example, at first, sets the variable t of a whole process dBe the current time, and each assembly weights of equipment are initialized as W 0, time t mBe certain fixed value.Then, according to the variation of curve A, if current time t dAlso do not arrive t mThe time, weights are with curve y 1And change, arrive t mAfterwards, weights are with curve y 2And change.The characteristics of bathtub curve A are that in the variation of turning point front and back, before the turning point, weights will reduce rapidly; And crossed after the turning point, weights increase very slowly.
At the equipment failure curve A, become the power association rule algorithm and be described below.
Algorithm.Discovery?of?Variable?Weighted?Association?Rules——DVWAR.
Input：(1)A?transaction?databases?D，in?which?each?item?i j?has?itsweight?w j；
(2)Two?threshold?values?wminsup?and?minconf.
Output：Changed?Weighted?Association?Rules.
Begin
(1) size＝Scan(D)；
(2)
Figure BSA00000257401000131
(3) for(i＝1；i≤size；i++){
(4)
Figure BSA00000257401000132
(5) }
(6) W＝W 0
(7) Call?WA
(8) If?t d≤t m
(9) then?if?W＝W 0
(10) then?Rules-Set＝Rules-Gen(L)；
(11) else?W＝Wy 1
(12) call?WA
(13) el?se?if?W＝W 0
(14) then?Rules-Set＝Rules-Gen(L)；
(15) else?W＝Wy 2
(16) call?WA
(17) WA：for?each?transaction?do
(18) (SC，C 1)＝Count(D，W)；
(19) for(k＝2；k≤size；k++){
(20) C k＝Join(C k-1)；
(21) C k＝Prune(C k)；
(22) (C k，L k)＝Check(C k，D)；
(23) L＝L∪L k；
(24) }
(25) Rules-Set＝Rules-Gen(L)；
(26) End.
Explanation to the symbol implication that occurs in the above-mentioned algorithm:
1.D: transaction data base
2.W: the set of project weights
3.W 0: the set of initial project weights
4.WA: the association rule algorithm behind the weights is given in calculating.
5.Wy 1: y 1The set of section project weights; Wy 2: y 2The set of section project weights
6.t d: the current time
7.t m: the turnover time
8.L k: frequent k-Item Sets
9.C k: the set of the possible frequent k-subset of items of frequent j-Item Sets
10.SC (X): the number of support of Item Sets X
11.wminsup: lowest weighted is supported threshold values
12.minconf: minimum trust threshold values
The execution in step that becomes power association rule algorithm DVWAR is as follows:
1. begin to carry out, at first transaction data base D is scanned, by subroutine Scan (D), find the wherein maximal possible length of Frequent Item Sets, and return this numerical value.
2. initial weight is given equipment each components values, initial weight is determined according to historical Monitoring Data.
3. subroutine WA is the program that often need call in the whole change power association rule algorithm operational process, and its purpose is more new database of the inspection that continues.
4. in carrying out the WA process, (k-that calculates each 1-Item Sets supports expectation to subroutine Count for D, the W) number of support of accumulative total 1-Item Sets.Collect its number of support then and be not less than the 1-Item Sets that k-supports expectation, form C 1
5. subroutine Join (C K-1) according to C K-1Generate C kLink, generate on-link mode (OLM) with the Apriori algorithm.
6. subroutine Prune (C k) pruning of project implementation collection: C kThe subclass of middle candidates collection is not at C K-1In; Estimate the upper bound of the number of support (SC (X)) of candidate k-Item Sets X, it is C K-1Minimum number of support in middle k different (k-1)-subset of items.K-according to all items collection that has calculated supports expectation, if the upper bound that number of support SC (X) is estimated shows that Item Sets X can not become the subclass of any Frequent Item Sets in follow-up traversal, this Item Sets X just can be pruned away so.
7. subroutine Check (C k, D) check traversal transaction data base D, upgrade C kIn the support counting of all candidates collection.By the method for similar shearing procedure, delete the candidates collection that those do not satisfy all possibility Frequent Item Sets support expectations.Remaining candidates collection all is kept at C kIn.Then, reexamine the weighting support of projects collection, select frequent k-Item Sets L k
8. subroutine Rules-Gen (L) generates the correlation rule that meets minimum trust threshold according to the Frequent Item Sets among the L.
9. with regard to equipment failure curve law B, there is a time turning point t mSo, done branch's situation discussion in the program, work as t d≤ t mThe time, and process is checked when contrast finds that initial weight has changed, according to y 1The section curve changes in time gives weights again; Work as t d＞t mThe time, and process is checked when contrast finds that initial weight has changed, according to y 2The section curve changes in time gives weights again.
Second step, monitor the hydraulically operated equipment (HOE) data, and its related data is analyzed and handled, extract eigenwert;
The 3rd step was complementary eigenwert with the pattern of weighing in the related early warning library based on change, judge promptly whether hydraulically operated equipment (HOE) the sign of degenerating occurs;
The 4th step if coupling is unsuccessful, illustrated then that the sign of degenerating did not appear in hydraulically operated equipment (HOE), returned and continued the monitoring equipment data;
In the 5th step, if the match is successful, then the degeneration sign appears in devices illustrated, does following processing;
The 6th step, utilize association rule algorithm to calculate the probable value of corresponding incipient fault, promptly the incipient fault support of mating in becoming the related early warning library of power is the probable value of corresponding incipient fault;
The 7th step, as input, try to achieve the consequence value of corresponding failure with equipment self risk, personal risk, environmental risk, social risk and system risk by neural net model establishing, it is specific as follows:
● the selection standard fault sample;
● each standard fault sample is learnt;
● sample to be tested is input to the BP neural network model that has trained, tries to achieve the consequence comprehensive evaluation value of corresponding failure.
The 8th step, the probable value of incipient fault and corresponding consequence value are multiplied each other, obtain the value-at-risk of this incipient fault;
In the 9th step, judge whether this value-at-risk surpasses the threshold value of regulation in advance;
In the tenth step, then do not return the continuation monitoring equipment if surpass;
The 11 step surpassed threshold value and then the failure risk value is sorted from high to low, to determine optimum service sequence;
In the 12 step, carry out shutdown maintenance according to optimal ordering, and return and continue monitoring hydraulically operated equipment (HOE) data.
Embodiment 1: utilize to become the probable value that power association rule algorithm DVWAR calculates corresponding incipient fault
Present embodiment is analyzed with two kinds of methods respectively, and a kind of is the weighted association rules algorithm, and another kind is to become the power association rule algorithm, and what use in the example is same group of data, by calculating final result and comparing.
Example: hydraulically operated equipment (HOE) M is made up of 5 kinds of assemblies altogether, and its contingent fault has 7 kinds.Table 1 is the initial weight of each parts, and table 2 is each Mishap Database, the assembly that expression can Gong be traced when certain fault takes place.If minimum support threshold value wminsup is 1.
Each Mishap Database of initial weight table 2 equipment M of each parts of table 1 equipment M
Figure BSA00000257401000171
1. use the weighted association rules algorithm to excavate
Because the weighted association rules algorithm is in the life cycle of entire equipment, the weights of each assembly only are endowed once, so each assembly initial weight of equipment M is invariable, the weights that are assembly A are 0.2, the weights of assembly B are 0.1, the weights of assembly C are 0.3, and the weights of assembly D are 0.4, and the weights of assembly E are 0.8.According to the method for digging of weighted association rules, excavate according to the step of its algorithm, the association rule model that finally obtains is BDE.
2. use change power association rule algorithm to excavate
If the weights of assembly A change as shown in Figure 4.The initial weight of assembly A is 0.2, and its weights change follows curve y A=0.2.The weights of assembly B change as shown in Figure 5, and the initial weight of assembly B is 0.1, and its weights change follows curve y B=0.1.
Assembly C weights change as shown in Figure 6.The initial weight of assembly C is 0.3, and its weights change follows curve y c=0.05x+0.3.
Assembly D weights change as shown in Figure 7.The initial weight of assembly D is 0.4, turning point time t m=1, current working time t d≤ t mThe time, the ratio of its shared weights is with curve y D1=0.3 x-0.1 and change, work as t d＞t mThe time, the ratio of its shared weights is with curve y D2=0.02x+0.18 and changing.Assembly E weights change as shown in Figure 8.The initial weight of assembly E is 0.8, turning point time t M '=1, current working time t d≤ t M 'The time, the ratio of its shared weights is with curve y E1=0.6 x-0.2 and change, work as t d＞t M 'The time, the ratio of its shared weights is with curve y E2=0.01x+0.39 and changing.
Each assembly weights during each assembly weight table 5x=10 during each assembly weight table 4x=1 during 3x=0.5 of table
By becoming in the power association rule algorithm, the variation of each assembly weights of equipment, can reflect each assembly difference constantly, the difference of significance level in the same equipment.Be not difficult to find that the weights of assembly A and B do not change by the contrast of chart.Assembly C is along with the growth of time, and it is big that weights become gradually.The weights of assembly D and E are at turnover time point t mDiminish gradually, at t before mIncrease gradually afterwards.
According to the weights shown in table 3, table 4, the table 5, excavate with becoming the power association rule algorithm, the association rule model in three moment that obtain is respectively BDE, DE, CDE.
Weigh result's contrast that two kinds of methods of correlation rule obtain from weighted association rules and change, can see that the result that two kinds of methods draw is discrepant, based on each the assembly weights of equipment that become the power association rule algorithm is constantly to change, its status in entire equipment also is constantly to change, and remain unchanged based on the weights of each assembly of equipment of weighted association rules algorithm, the result who so finally causes is exactly the deviation of the maintenance decision done. alwaysIt is the key factor that pay attention to that the weights that example proof assembly causes because of wearing and tearing change, and becomes the power association rule algorithm and can diagnose equipment failure more accurately in actual applications.
It should be noted that, growth along with the time, the weights of each assembly of equipment M all can change, and the weights of each assembly should be to change according to different curves, and the concrete curve that changes should be determined according to the Monitoring Data and the historical monitoring of physical device.Because in whole change power association rule algorithm, real-time variation can take place the weights of each assembly, therefore can form N opens weight table.This becomes the marrow part of power correlation rule just.
Embodiment 2: utilize neural net model establishing to try to achieve the consequence value of corresponding failure
For the application of BP neural network in the failure effect value prediction is described, choose five of hydraulically operated equipment (HOE) self value-at-risk, personal value-at-risk, environmental risk value, social risk value and system risk values respectively as input item, output item is an incipient fault consequence comprehensive evaluation value.Because data are bigger, table 6 is listed the learning sample of partial fault consequence comprehensive evaluation value prediction.
The learning sample of table 6 partial fault consequence comprehensive evaluation value prediction
In the design of BP neural network, if network hidden layer node number is selected very little, the Nonlinear Mapping function and the fault-tolerance of network are relatively poor, and the node number selects too much can make again learning time to increase, and influences learning efficiency.More about the definite experimental formula of hidden layer node number at present, this patent adopts H=log commonly used 2N determines hidden layer node number, i.e. H=log 25=2.(wherein H represents the hidden layer node number, and N representative input is counted, and O represents the output node number).According to following steps, set up failure effect comprehensive evaluation value forecast model.
The first step is at first to weight coefficient w IjPut initial value, promptly to the weight coefficient w of each layer IjPut a less non-zero random number, but w wherein I, n+1=-θ.
In second step, import a sample X=(x 1, x 2, x 3, x 4, x 5, 1) and corresponding desired output Y=(y 1, y 2..., y 5).
In the 3rd step, the output that calculate each layer is for k layer i neuronic output
Figure BSA00000257401000192
, have:
Figure BSA00000257401000193
Figure BSA00000257401000194
w I, n+1=-θ
The 4th goes on foot, and asks the study error of each layer
Figure BSA00000257401000195
For output layer k=m is arranged, have Each layer for other has
Figure BSA00000257401000197
The 5th step, modified weight coefficient w IjWith threshold values θ.Formula is as follows:
w ij ( t + 1 ) = w ij ( t ) - η d i k x j k - 1 , Δ w ij ( t ) = - η d i k x j k - 1 + αΔ w ij ( t - 1 ) = w ij ( t ) - w ij ( t - 1 )
In the 6th step, after having obtained each weight coefficient of each layer, can whether meet the demands by given index of quality differentiation.If meet the demands, then algorithm finishes; If backlog demand was then returned for the 3rd step and is carried out.
This learning process all will be carried out for arbitrary given sample and desired output, till satisfying all input and output requirements.
According to the failure effect comprehensive evaluation value forecast model of above-mentioned structure, the node number that utilizes MATLAB to set up input layer, hidden layer and an output layer is respectively 5,2, and 1 BP neural network model is used to carry out network training and check, and default error is 0.01.This example is utilized 20 groups of data of collecting in advance, through after the standardization, 16 groups of data wherein is used for the training of model as learning sample, has reached the error requirements of setting through 800 multisteps training back error.After network training is finished, to remain 4 groups of data as the checking sample, result and expection evaluation result with model output compare, and reality is exported and expected output basically identical as a result, and this indicates that the failure effect comprehensive evaluation value forecast model based on the BP neural network successfully builds up.
Embodiment 3:OMMHERC method and classic method contrast
From the collection of certain steel plant's hydraulically operated equipment (HOE) 236 failure loggings, set up the excavation storehouse that comprises 200 records and comprised the test library of 36 records, on the test machine of CPU3.06GHz, internal memory 2G, windowsXP operating system, test.Utilize VC++6.0 to write equipment failure Risk Calculation program, the optimum repair determining method OMMHERC of hydraulically operated equipment (HOE) that will have risk control is applied to this data, and its accuracy rate of diagnosis is as shown in table 7.Also listed in the table 7 diagnostic result of same sample based on FMEA algorithm and FTA algorithm, visible OMMHERC method is compared the tangible accurate rate of diagnosis that improved with classic method.
Table 7 OMMHERC method example
Figure BSA00000257401000203
Annotate: a/b/c, a is for adopting this patent OMMHERC method diagnostic result, and b is for adopting the FMEA algorithm, and c is for adopting the diagnostic result of FTA algorithm
In order to verify the digging efficiency of OMMHERC method, the excavation storehouse of 200 records record duplicated expand to 600,800 and 1000 records, as seen the efficient of 3 kinds of algorithms excavated in large scale database as shown in Figure 9, and OMMHERC method diagnosis efficiency is the highest.
Show by above-mentioned test:
● the optimum repair determining method OMMHERC of hydraulically operated equipment (HOE) slave unit self historical data that has risk control is set out, and real-time monitor data is analyzed and pattern match, can improve accurate rate of diagnosis greatly.
● the optimum repair determining method OMMHERC of hydraulically operated equipment (HOE) that has risk control only need calculate probability of happening value and the consequence the value whether system of once can trying to achieve has incipient fault, incipient fault, improved the efficient of diagnosis greatly, illustrated that this algorithm can be applied to inline diagnosis.

Claims (2)

1. the optimum repair determining method of hydraulically operated equipment (HOE) that has risk control is characterized in that comprising the steps:
The first step: utilize the hydraulically operated equipment (HOE) historical data to make up and become the related early warning library of power, wherein said device history data comprises non-malfunction and fault state data;
Second step: monitor the hydraulically operated equipment (HOE) data, and its related data is analyzed and handled, extract eigenwert;
The 3rd the step: with eigenwert be complementary based on the pattern in the related early warning library, promptly whether judgment device the sign of degenerating occurs;
The 4th step: if coupling is unsuccessful, illustrate that then the sign of degenerating does not appear in hydraulically operated equipment (HOE), return and continue the monitoring equipment data;
The 5th step:, illustrate that then the degeneration sign appears in hydraulically operated equipment (HOE), does following processing if the match is successful;
The 6th step: utilize to become the probable value that the power association rule algorithm calculates corresponding incipient fault, promptly the incipient fault support of mating in becoming the related early warning library of power is the probable value of corresponding incipient fault;
The 7th step: as input, try to achieve the consequence value of corresponding failure by neural net model establishing with hydraulically operated equipment (HOE) self risk, personal risk, environmental risk, social risk and system risk;
The 8th step: the probable value and the corresponding consequence value of incipient fault are multiplied each other, obtain the value-at-risk of this incipient fault;
The 9th step: judge whether this value-at-risk surpasses the threshold value of regulation in advance;
The tenth step: do not continue the monitoring hydraulically operated equipment (HOE) if surpass then to return;
The 11 step: surpass threshold value and then the failure risk value is sorted from high to low, to determine optimum service sequence;
The 12 step: carry out shutdown maintenance according to optimal ordering, and return and continue monitoring hydraulically operated equipment (HOE) data.
2. the method for claim 1 is characterized in that: set up and become the related early warning library of power and comprise:
1) begins to carry out, at first transaction data base D is scanned, find the wherein maximal possible length of Frequent Item Sets, and return this numerical value;
2) initial weight is given equipment each components values, initial weight is determined according to historical Monitoring Data;
3) number of support of accumulative total 1-Item Sets, the k-that calculates each 1-Item Sets supports expectation, collects its number of support then and is not less than the 1-Item Sets that k-supports expectation, generates candidate 1-Item Sets C 1
4) according to candidate k-1-Item Sets C K-1Generate candidate k-Item Sets C kLink;
5) pruning of project implementation collection;
6) traversal transaction data base D upgrades candidate k-Item Sets C kIn the support counting of all candidates collection, the candidates collection that all may Frequent Item Sets supports expectation is not satisfied in deletion, remaining candidates collection all is kept at candidate k-Item Sets C kIn, then, reexamine the weighting support of projects collection, select frequent k-Item Sets L k
7) generate the correlation rule that meets minimum trust threshold according to Frequent Item Sets.
CN2010102724461A 2010-09-01 2010-09-01 Method for optimal maintenance decision-making of hydraulic equipment with risk control Active CN101950382B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2010102724461A CN101950382B (en) 2010-09-01 2010-09-01 Method for optimal maintenance decision-making of hydraulic equipment with risk control

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2010102724461A CN101950382B (en) 2010-09-01 2010-09-01 Method for optimal maintenance decision-making of hydraulic equipment with risk control

Publications (2)

Publication Number Publication Date
CN101950382A true CN101950382A (en) 2011-01-19
CN101950382B CN101950382B (en) 2013-03-06

Family

ID=43453873

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2010102724461A Active CN101950382B (en) 2010-09-01 2010-09-01 Method for optimal maintenance decision-making of hydraulic equipment with risk control

Country Status (1)

Country Link
CN (1) CN101950382B (en)

Cited By (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102623910A (en) * 2012-04-27 2012-08-01 重庆大学 Reliability-based maintenance decision method for switch equipment
CN102809965A (en) * 2012-07-30 2012-12-05 燕山大学 Fault early warning method for hydraulic equipment based on fault frequent pattern
CN102981073A (en) * 2012-11-20 2013-03-20 中国兵器工业计算机应用技术研究所 Portable fault detector for electronic information system
CN104408525A (en) * 2014-11-11 2015-03-11 西北工业大学 Quantitative evaluation and control method of job shop scheduling risks
CN104820716A (en) * 2015-05-21 2015-08-05 中国人民解放军海军工程大学 Equipment reliability evaluation method based on data mining
CN106598030A (en) * 2016-12-22 2017-04-26 西安理工大学 Axle temperature correlation analysis method based on data
CN108062088A (en) * 2017-12-15 2018-05-22 广州广电运通金融电子股份有限公司 Station terminal failure risk method for early warning, device, terminal and storage medium
CN108267962A (en) * 2016-12-30 2018-07-10 中国移动通信有限公司研究院 A kind of control method and device
CN108288019A (en) * 2017-12-06 2018-07-17 中国铁道科学研究院电子计算技术研究所 A kind of urban track traffic preventative maintenance object identifying method and device
CN108460523A (en) * 2018-02-12 2018-08-28 阿里巴巴集团控股有限公司 A kind of air control rule generating method and device
CN108757643A (en) * 2018-07-04 2018-11-06 重庆大学 A kind of hydraulic system initial failure active removing method
CN109086804A (en) * 2018-07-12 2018-12-25 中石化石油机械股份有限公司 A kind of hydraulic device fault forecast method merged based on multi source status monitoring information and reliability characteristic
CN109376877A (en) * 2018-10-11 2019-02-22 华自科技股份有限公司 Equipment O&M method for early warning, device, computer equipment and storage medium
CN109544166A (en) * 2018-11-05 2019-03-29 阿里巴巴集团控股有限公司 A kind of Risk Identification Method and device
CN109579220A (en) * 2018-10-15 2019-04-05 平安科技（深圳）有限公司 Air-conditioning system fault detection method, device, electronic equipment
CN109697210A (en) * 2018-12-26 2019-04-30 天津瑞源电气有限公司 A kind of Wind turbines relevant fault inline diagnosis method
CN109780812A (en) * 2019-01-04 2019-05-21 青岛海尔股份有限公司 The treating method and apparatus of the running state data of refrigerator
CN110852854A (en) * 2018-08-01 2020-02-28 阿里巴巴集团控股有限公司 Generation method of quantitative yield model and evaluation method of risk control strategy
CN111579978A (en) * 2020-05-18 2020-08-25 刘柱华 System and method for realizing relay fault identification based on artificial intelligence technology
CN111626239A (en) * 2020-05-29 2020-09-04 山东山科智能科技有限公司 Criminal risk tendency prediction method and system based on behavior characteristics
CN112541597A (en) * 2019-09-04 2021-03-23 上海杰之能软件科技有限公司 Multi-equipment maintenance method and device, storage medium and terminal
CN113688224A (en) * 2021-10-26 2021-11-23 成都飞机工业（集团）有限责任公司 Self-adaptive processing method for complex equipment delivery problem based on grey correlation
CN113723725A (en) * 2020-05-25 2021-11-30 中国石油化工股份有限公司 Risk early warning method and device for operation process of chemical device and terminal equipment
CN118523491A (en) * 2024-05-13 2024-08-20 青岛裕华电子科技有限公司 Power equipment data safety monitoring system and method based on Internet of things

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101777248A (en) * 2010-01-15 2010-07-14 李钢 Method for implementing system maintenance in security network
US20100199010A1 (en) * 2006-08-23 2010-08-05 Ori Goren Device having priority upgrade mechanism capabilities and a method for updating priorities

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100199010A1 (en) * 2006-08-23 2010-08-05 Ori Goren Device having priority upgrade mechanism capabilities and a method for updating priorities
CN101777248A (en) * 2010-01-15 2010-07-14 李钢 Method for implementing system maintenance in security network

Cited By (38)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102623910B (en) * 2012-04-27 2014-11-26 重庆大学 Reliability-based maintenance decision method for switch equipment
CN102623910A (en) * 2012-04-27 2012-08-01 重庆大学 Reliability-based maintenance decision method for switch equipment
CN102809965A (en) * 2012-07-30 2012-12-05 燕山大学 Fault early warning method for hydraulic equipment based on fault frequent pattern
CN102981073B (en) * 2012-11-20 2015-07-15 中国兵器工业计算机应用技术研究所 Portable fault detector for electronic information system
CN102981073A (en) * 2012-11-20 2013-03-20 中国兵器工业计算机应用技术研究所 Portable fault detector for electronic information system
CN104408525B (en) * 2014-11-11 2018-01-09 西北工业大学 The quantitative evaluation and control method of solving job shop scheduling problem risk
CN104408525A (en) * 2014-11-11 2015-03-11 西北工业大学 Quantitative evaluation and control method of job shop scheduling risks
CN104820716A (en) * 2015-05-21 2015-08-05 中国人民解放军海军工程大学 Equipment reliability evaluation method based on data mining
CN104820716B (en) * 2015-05-21 2017-11-28 中国人民解放军海军工程大学 Equipment Reliability appraisal procedure based on data mining
CN106598030A (en) * 2016-12-22 2017-04-26 西安理工大学 Axle temperature correlation analysis method based on data
CN106598030B (en) * 2016-12-22 2019-09-27 西安理工大学 A kind of axis temperature association analysis method based on data
CN108267962A (en) * 2016-12-30 2018-07-10 中国移动通信有限公司研究院 A kind of control method and device
CN108288019A (en) * 2017-12-06 2018-07-17 中国铁道科学研究院电子计算技术研究所 A kind of urban track traffic preventative maintenance object identifying method and device
CN108288019B (en) * 2017-12-06 2020-06-16 中国铁道科学研究院集团有限公司电子计算技术研究所 Method and device for identifying preventive maintenance object of urban rail transit
CN108062088A (en) * 2017-12-15 2018-05-22 广州广电运通金融电子股份有限公司 Station terminal failure risk method for early warning, device, terminal and storage medium
CN108062088B (en) * 2017-12-15 2019-11-19 广州广电运通金融电子股份有限公司 Station terminal failure risk method for early warning, device, terminal and storage medium
CN108460523A (en) * 2018-02-12 2018-08-28 阿里巴巴集团控股有限公司 A kind of air control rule generating method and device
CN108757643A (en) * 2018-07-04 2018-11-06 重庆大学 A kind of hydraulic system initial failure active removing method
CN109086804A (en) * 2018-07-12 2018-12-25 中石化石油机械股份有限公司 A kind of hydraulic device fault forecast method merged based on multi source status monitoring information and reliability characteristic
CN109086804B (en) * 2018-07-12 2022-04-22 中石化石油机械股份有限公司 Hydraulic equipment early failure prediction method based on fusion of multi-source state monitoring information and reliability characteristics
CN110852854B (en) * 2018-08-01 2023-10-31 创新先进技术有限公司 Method for generating quantitative gain model and method for evaluating risk control strategy
CN110852854A (en) * 2018-08-01 2020-02-28 阿里巴巴集团控股有限公司 Generation method of quantitative yield model and evaluation method of risk control strategy
CN109376877A (en) * 2018-10-11 2019-02-22 华自科技股份有限公司 Equipment O&M method for early warning, device, computer equipment and storage medium
CN109579220A (en) * 2018-10-15 2019-04-05 平安科技（深圳）有限公司 Air-conditioning system fault detection method, device, electronic equipment
CN109579220B (en) * 2018-10-15 2022-04-12 平安科技（深圳）有限公司 Air conditioning system fault detection method and device and electronic equipment
CN109544166B (en) * 2018-11-05 2023-05-30 创新先进技术有限公司 Risk identification method and risk identification device
CN109544166A (en) * 2018-11-05 2019-03-29 阿里巴巴集团控股有限公司 A kind of Risk Identification Method and device
CN109697210A (en) * 2018-12-26 2019-04-30 天津瑞源电气有限公司 A kind of Wind turbines relevant fault inline diagnosis method
CN109697210B (en) * 2018-12-26 2023-09-05 天津瑞源电气有限公司 Online diagnosis method for wind turbine generator set associated faults
CN109780812B (en) * 2019-01-04 2022-10-28 海尔智家股份有限公司 Method and device for processing running state data of refrigerator
CN109780812A (en) * 2019-01-04 2019-05-21 青岛海尔股份有限公司 The treating method and apparatus of the running state data of refrigerator
CN112541597A (en) * 2019-09-04 2021-03-23 上海杰之能软件科技有限公司 Multi-equipment maintenance method and device, storage medium and terminal
CN111579978A (en) * 2020-05-18 2020-08-25 刘柱华 System and method for realizing relay fault identification based on artificial intelligence technology
CN111579978B (en) * 2020-05-18 2024-01-02 珠海施诺电力科技有限公司 System and method for realizing relay fault identification based on artificial intelligence technology
CN113723725A (en) * 2020-05-25 2021-11-30 中国石油化工股份有限公司 Risk early warning method and device for operation process of chemical device and terminal equipment
CN111626239A (en) * 2020-05-29 2020-09-04 山东山科智能科技有限公司 Criminal risk tendency prediction method and system based on behavior characteristics
CN113688224A (en) * 2021-10-26 2021-11-23 成都飞机工业（集团）有限责任公司 Self-adaptive processing method for complex equipment delivery problem based on grey correlation
CN118523491A (en) * 2024-05-13 2024-08-20 青岛裕华电子科技有限公司 Power equipment data safety monitoring system and method based on Internet of things

Also Published As

Publication number Publication date
CN101950382B (en) 2013-03-06

Similar Documents

Publication Publication Date Title
CN101950382B (en) Method for optimal maintenance decision-making of hydraulic equipment with risk control
Boral et al. An integrated approach for fuzzy failure modes and effects analysis using fuzzy AHP and fuzzy MAIRCA
He et al. Joint decision-making of parallel machine scheduling restricted in job-machine release time and preventive maintenance with remaining useful life constraints
CN106444703B (en) Dynamic equipment running status fuzzy evaluation and prediction technique based on fault mode probability of happening
Keats Statistical process control in automated manufacturing
Wachla et al. A method of leakage location in water distribution networks using artificial neuro-fuzzy system
Di Maio et al. Ensemble-approaches for clustering health status of oil sand pumps
CN101957889B (en) Selective wear-based equipment optimal maintenance time prediction method
CN108873859B (en) Bridge type grab ship unloader fault prediction model method based on improved association rule
CN105096053A (en) Health management decision-making method suitable for complex process system
CN111126853A (en) Fuzzy FMEA-based hydraulic engineering risk early warning analysis method and system
CN112083244A (en) Integrated avionics equipment fault intelligent diagnosis system
CN105975797A (en) Product early-fault root cause recognition method based on fuzzy data processing
Manlin et al. A novel combination belief rule base model for mechanical equipment fault diagnosis
Borissova et al. An integrated framework of designing a decision support system for engineering predictive maintenance
Yu et al. An active construction dynamic schedule management model: using the fuzzy earned value management and BP neural network
CN117893020A (en) Hydraulic engineering quality safety risk assessment method
Marinelli et al. Earthmoving trucks condition level prediction using neural networks
CN113989550A (en) Electric vehicle charging pile operation state prediction method based on CNN and LSTM hybrid network
Wang et al. Reliability analysis of complex electromechanical systems: State of the art, challenges, and prospects
CN103902845A (en) Method for evaluating risk of simulation system based on fuzzy FMEA
CN105868115A (en) Building method and system for software test model of software intensive system
Rane et al. Fuzzy AHP and Fuzzy TOPSIS as an effective and powerful Multi-Criteria Decision-Making (MCDM) method for subjective judgements in selection process
Fan Data mining model for predicting the quality level and classification of construction projects
CN117312972A (en) Method for identifying health state of scraper conveyor speed reducer

Legal Events

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

Owner name: TIANJIN DEVELOPMENT ZONE JINGNUO HANHAI DATA TECHN

Free format text: FORMER OWNER: YANSHAN UNIVERSITY

Effective date: 20150424

C41 Transfer of patent application or patent right or utility model
COR Change of bibliographic data

Free format text: CORRECT: ADDRESS; FROM: 066004 QINHUANGDAO, HEBEI PROVINCE TO: 300457 HANGU, TIANJIN

TR01 Transfer of patent right

Effective date of registration: 20150424

Address after: Tianjin economic and Technological Development Zone 300457 Fifth Avenue Taihua Road No. 12 TEDA SME Development Center

Patentee after: Tianjin Development Zone JINGNUO Ocean Data Technology Co. Ltd.

Address before: Hebei Street West Harbor area, 066004 Hebei city of Qinhuangdao province No. 438

Patentee before: Yanshan University