CN103544389B - Autocrane method for diagnosing faults based on fault tree and fuzzy neural network - Google Patents
Autocrane method for diagnosing faults based on fault tree and fuzzy neural network Download PDFInfo
- Publication number
- CN103544389B CN103544389B CN201310492933.2A CN201310492933A CN103544389B CN 103544389 B CN103544389 B CN 103544389B CN 201310492933 A CN201310492933 A CN 201310492933A CN 103544389 B CN103544389 B CN 103544389B
- Authority
- CN
- China
- Prior art keywords
- pressure
- neural network
- value
- autocrane
- fault tree
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a kind of autocrane method for diagnosing faults based on fault tree and fuzzy neural network, step includes：1）Autocrane top event fault tree is established using deduction；2）The input of fuzzy neural network, output node number are determined according to fault tree branch case and Heuristics, establish structure of fuzzy neural network model；3）According to the knowledge contained in each branch of fault tree, training sample is extracted, and neural network is trained, establish ANN Reasoning and calculate required network weight and threshold matrix；4）Using the data on existing autocrane status monitoring platform, the fuzzy membership functions needed for fuzzy preceding operation is established with 3 σ Criterion Methods in statistical parameter method；5) measured data is input in fuzzy neural network and calculated, export fault mode.The method of the present invention avoids the blindness and triviality of detection process, improves accuracy rate of diagnosis.
Description
Technical field
The invention belongs to autocrane fault detection technique fields, are related to a kind of based on fault tree and fuzzy neural network
Autocrane method for diagnosing faults.
Background technology
Autocrane is that a kind of hoisting machinery is partially installed on Universal automobile chassis or special-purpose automobile chassis, has and carries
The wheeled crane of weight running car performance, its mobility is good, and convenient transfer, the speed of service is fast, be widely used in industrial and mining enterprises,
Construction site, station, harbour etc. carry out the various hoisting operations such as cargo handling, transfer, equipment installation and working at height.It is to subtracting
Light work intensity saves manpower, reducings the construction costs, and improves construction quality, and Accelerating The Construction speed realizes engineering construction mechanization
It plays a very important role.But the situation very severe of autocrane safety work for a long time, autocrane product by
Complexity is formed in its system, working environment is severe, usually requires that high load capacity, long-play, in addition the phase of maintaining system
To backwardness, therefore system often will appear various failures, and fatal crass's major accident happens occasionally, and has seriously affected construction project
Progress, benefit and people's property safety.
From both at home and abroad to the present situation of autocrane fault diagnosis from the point of view of, generally still take traditional fault diagnosis side
Method, i.e. maintenance personal are under the guidance of practical experience, according to systematic schematic diagram and actuation cycle table, by comparing, regional analysis,
The methods of comprehensive analysis, determines suspicious hydraulic components, then suspicious hydraulic components is carried out replacing test to judge failure original
Cause eliminates failure finally by the mode for component of exchanging.This method requirement maintenance personal grasps deeper relevant speciality base
This theory and operation principle have stronger discriminatory analysis ability, can ensure the validity and accuracy of diagnosis.Diagnosis process
It is very cumbersome, to pass through a large amount of inspection, verify work, the blindness during system failure detection is inevitable, dismounts work
It measures also larger, and can only be qualitative analysis, and to the fault distinguishing in terms of modern electronic technology and hydraulic control technology more
For difficulty, thus the reason of being diagnosed to be, is often not accurate enough.Therefore, this mode take, be laborious, is inefficient, economic benefit it is bad.
Invention content
The present invention provides a kind of autocrane method for diagnosing faults based on fault tree and fuzzy neural network, solve
In existing autocrane fault diagnosis, there is time-consuming, laborious, inefficient, economic benefit.
The technical scheme is that the autocrane method for diagnosing faults based on fault tree and fuzzy neural network,
Implement according to following steps：
1）Autocrane top event fault tree is established using deduction；
2）The network inputs of fuzzy neural network, network output node are determined according to fault tree branch case and Heuristics
Number establishes structure of fuzzy neural network model；
3）According to the knowledge contained in each branch of fault tree, training sample is extracted, and neural network is trained, built
Vertical ANN Reasoning calculates required network weight and threshold matrix；
4）It is true with 3 σ Criterion Methods in statistical parameter method using the historical data on autocrane status monitoring platform
Fuzzy membership functions needed for vertical fuzzy preceding operation；
5) measured data of autocrane status monitoring platform is input to through the established fuzzy god of above step
Through being calculated in network, diagnostic result is obtained, exports fault mode.
The beneficial effects of the present invention are：Known using the data and fault tree of autocrane standing state monitoring platform
Know, establish autocrane fuzzy neural network model and carry out fault diagnosis and fault prediction, can both imitate the logical thinking of human brain,
The function of human brain neuron can be imitated again；Not only the problem of can describing with fuzzy concept, but also with powerful learning ability
With the direct processing capacity of data；Not only there is stronger structure knowledge ability to express, natural language processing ability, but also with very strong
Fault-tolerant ability；Ensure the validity and accuracy of diagnosis；Maintenance personal without grasp deeper relevant speciality basic theories and
Operation principle without by a large amount of dismounting, inspection, verification work, avoids the blindness and triviality of detection process, solution
The problem of diagnosis of having determined is time-consuming, laborious, inefficient, economic benefit is bad.
Description of the drawings
Fig. 1 is the Troubleshooting Flowchart of the method for the present invention；
Fig. 2 is the hoisting system schematic diagram of the method for the present invention embodiment 1；
Fig. 3 is the fault tree models of the method for the present invention embodiment 1；
Fig. 4 is the Diagnosis of fuzzy neural network model of the method for the present invention embodiment 1；
Fig. 5 is the membership function parameter curve that the method for the present invention embodiment 1 rises hoist motor mouth pressure.
In figure, 1, fuel tank；2nd, oil filter；3rd, variable pump；4th, swivel joint；5th, overflow valve；6th, pressure reducing valve；7th, ratio depressurizes
Valve A；8th, proportional pressure-reducing valve B；9th, pressure reducing valve；10th, two-way electromagnetic valve；11st, reversal valve；12nd, balanced valve；13rd, hoist motor, 14,
Hydraulic braking cylinder；15th, shaft coupling；16th, retarder；17th, reel.
Specific embodiment
The present invention is described in detail with reference to the accompanying drawings and detailed description.
With reference to Fig. 1, the present invention is based on fault tree and the autocrane method for diagnosing faults of fuzzy neural network, failures
Diagnostic process is that, including building fuzzy neural network model, the measured data of autocrane actual measurement platform acquisition inputs the mould
It pastes in neural network model, finally exports fault mode.
Autocrane actual measurement platform measured data acquisition by temperature sensor, pressure sensor, vibrating sensor,
Flow sensor, obliquity sensor and the electric control handle of corresponding button and switch are realized；Temperature sensor setting therein
In hydraulic oil container, pressure sensor is arranged on the inlet and outlet oil circuit of the key components and parts such as pump, valve, motor, vibrating sensor
It is arranged on the case body of the key components and parts such as pump, valve, motor, flow sensor is arranged in the key such as pump, valve, motor zero
On the inlet and outlet oil circuit of component, obliquity sensor is arranged on the arm support of autocrane.
Button and switch mainly include elevator decline permission button, elevator rising allows button, turns round permission button, certainly
Dynamic idling pressure switch, main operating mode button, secondary operating mode button and parameter set button etc., are arranged on and drive indoor electric control handle
On operation panel.
The example, in hardware of the structure of fuzzy neural network model is to build fuzzy neural network controller, fuzzy neural network
The input terminal of controller is connect respectively with above-mentioned data acquisition equipment by signal wire；The output of fuzzy neural network controller
End is connected respectively by display, buzzer by signal wire, realizes the output of fault mode.
The autocrane method for diagnosing faults based on fault tree and fuzzy neural network of the present invention, according to following steps
Implement：
1）Autocrane top event fault tree is established using deduction
It is to find out the various possible factors combination for directly resulting in top event generation first, such as hardware fault, environmental factor
With human error etc.；Next finds out the immediate cause of each factor in the first step, follows the method and deduces downwards step by step, traces back to always
Cause the entire reason of system jam, that is, until analyzing the bottom event for needing not continue to analysis reason；Then at different levels
The corresponding symbol of event and it is suitable for the logic gate of logical relation between them and is connected with top event, obtains one with useful
Part is root, and intermediate event is section, and bottom event is leaf with several grades of inversion fault tree.
2）The network inputs of fuzzy neural network, network output node are determined according to fault tree branch case and Heuristics
Number establishes structure of fuzzy neural network model；
3）According to the knowledge contained in each branch of fault tree, training sample is extracted, and neural network is trained, built
Vertical ANN Reasoning calculates required network weight and threshold value knowledge（Or it is network weight and threshold matrix）；
4）Using the historical data on existing autocrane status monitoring platform, with 3 σ criterion in statistical parameter method
Method establishes the fuzzy membership functions needed for fuzzy preceding operation；
3 σ criterion：Failure bound（Diagnostic parameters value is higher or relatively low）ForParameter normal limits are
In formula：Parameter YiFor i-th of sample；R is sample size.
5) measured data of autocrane status monitoring platform is input to through the established fuzzy god of above step
Through being calculated in network, diagnostic result is obtained, exports fault mode.
Embodiment
The failure of certain autocrane " promoting attonity " is diagnosed
1）Certain autocrane top event fault tree is established using deduction
Fig. 2 show certain autocrane hoisting circuit fundamental diagram, and structure mainly includes, variable pump 3 and fuel tank 1
Connection, is additionally provided with oil filter 2, variable pump 3 is connected with swivel joint 4, and swivel joint 4 is distinguished between variable pump 3 and fuel tank 1
It is connected with overflow valve 5, pressure reducing valve 6, pressure reducing valve 9, reversal valve 11, proportional pressure-reducing valve is parallel between pressure reducing valve 6 and reversal valve 11
A7 and proportional pressure-reducing valve B8；Reversal valve 11 is connected with hoist motor 13, is additionally provided between reversal valve 11 and hoist motor 13
Balanced valve 12；Pressure reducing valve 9 is connected by two-way electromagnetic valve 10 with hydraulic braking cylinder 14；Hoist motor 13 passes sequentially through shaft coupling
15th, retarder 16 is sequentially connected with reel 17, and reel 17 hangs weight by pulley.
As shown in figure 3, autocrane top event fault tree is established for " promoting attonity " top event using deduction
Fault tree models, and to each bottom event of the top event to encoding, as shown in table 1.
1 bottom event coding schedule of table
2）The network inputs of fuzzy neural network, network output node are determined according to fault tree branch case and Heuristics
Number establishes structure of fuzzy neural network model.
In 19 bottom events of table 1, each independent bottom event is exactly a minimal cut set.Fault tree according to fig. 2 is known
Know it is found that the generation of fault tree top event can be led to by sharing 2 branches.Branch CL-1 represents that motor rises mouth insufficient pressure and directly makes
Into the event chain for promoting attonity.Branch CL-2 describes motor and rises the excessive failure exception reason of mouth pressure.Therefore, Ke Yijian
It founds 2 fuzzy neural networks and parallel diagnosis is carried out to branch CL-1 and branch CL-2 respectively, then according in each branch of fault tree
The knowledge contained determines the structure of each automatic network.
As space is limited, here by taking branch CL-2 as an example.It is hoist motor failure that branch CL-2, which includes 6 minimal cut sets,（Bottom
Event 14）, winding speed reducing failure（Bottom event 15）, brake it is stuck（Bottom event 16）, 9 failure of pressure reducing valve（Bottom event 17）、
Two-way electromagnetic valve failure（Bottom event 18）, checking cylinder piston wear leakage（Bottom event 19）.Motor is equipment in normal conditions
Be not in what is built the pressure, and the generation of any one minimal cut set of above 6 minimal cut sets can all cause motor to build the pressure, motor rises mouth
Pressure PmIt is excessive.Minimal cut set X14Or X15Generation, will not be to checking cylinder control mouth pressure PdIt has an impact, and X16Generation can make
PdValue is higher by normal range value and has larger pressure oscillation, X17、X18Or X19Generation can then make PdValue is less than normal range value,
But X19Generation can make PdKeep larger pressure oscillation, and X17Or X18Generation can make PdThen keep smaller pressure oscillation.
PdPressure oscillation situation its mean square deviation D may be usedPdTo characterize.System is only needed by PmAnd PdState-detection, obtain 3
A input variable：Pm、Pd、DPd, following 5 kinds of fault modes can be distinguished：Normal work, { X14Or X15}、X16、{X17Or X18}、
X19.Thus it can determine that the corresponding structure of fuzzy neural network of branch CL-2 has the structure of 3 input, 5 output, hidden layer neuron
Number of nodes is rule of thumb taken as 6, i.e., 3 × 6 × 5, as shown in Figure 4.Wherein, the neuron in hidden layer uses logarithm Sigmoid types
Function, the neuron of output layer use purely linear purelin transforming function transformation functions.
3）According to the knowledge contained in each branch of fault tree, training sample is extracted, and neural network is trained, built
Vertical ANN Reasoning calculates required network weight and threshold value knowledge（Weights and threshold matrix）；
According to step 2）Analysis, extract the training sample data of fuzzy neural network, as shown in table 2.In table,Value
0.1st, 0.5,0.9 represent that motor rises mouth pressure P respectivelymIt is too low, normal, excessively high；Value 0.1,0.5,0.9 represents Pd systems respectively
Dynamic cylinder control mouth hypotony, normal, excessively high, DPdValue 0.1,0.5,0.9 represent respectively checking cylinder control mouth pressure mean square deviation it is too low,
Normally, it is excessively high.
2 train samples of table
4）It is true with 3 σ Criterion Methods in statistical parameter method using the historical data on autocrane status monitoring platform
Fuzzy membership functions needed for vertical fuzzy preceding operation；
First, according to the sample set of input variable each input variable is determined in the distribution situation of input variable interval
Membership function, state variable domain it is low conversion using lower semi-trapezoid distribution membership function, expression formula such as formula（3）, and really
It is 0.1 to determine weight coefficient；Normally convert the membership function for all using trapezoidal profile, expression formula such as formula in state variable domain（4）,
And determine that weight coefficient is 0.5；The high conversion in state variable domain is using the membership function for rising half trapezoidal profile, expression formula such as formula
（5）, and determining that weight coefficient is 0.9, the setting of wherein weights is to handle the detection data being between two states, is utilized
The average weighted method of each membership function is by the data conversion between two states into fuzzy data.
Secondly, the value of each parameter of membership function is determined.Diagnostic parameters y Normal Distributions rule when equipment works normally
Rule, Diagnostic parameters collects a certain amount of sample data during to normal work, statistical disposition is carried out to these data, if test specimens
Sheet and mean value；Distance is more than 2 times or 3 times of variances, then it is assumed that the test sample is uncertain, therefore can be judged to exception.Then
It can obtain alarm threshold：Failure bound（Parameter value is higher or relatively low）For；Parameter normal limits are.This is 3 σ
Criterion.More meet the membership function of objective reality to obtain, the fuzzy net that step is established is made to enter practical application,
This determines the value of each parameter of membership function with 3 σ Criterion Methods in statistical parameter method.To input variable（Pm、Pd、DPd）Sample
3 σ Criterion Methods of data application obtain the value of membership function parameter a, b, c, d of each input variable, are then updated to formula（3）~formula
（5）In, the membership function of each input variable is obtained, that is, establishes the membership function of fuzzy neural network.Fig. 5 is using 3 σ parameters
The neural network input variable hoist motor that statistic law obtains rises the membership function parameter value curve of mouth pressure Pm.
Similarly, can obtain checking cylinder control mouth pressure Pd membership function parameter values is：A, b, c, d | brake-cylinder pressure (bar) }
={ 36.1001,36.3344,36.5688,36.6031 }.In order to make checking cylinder control mouth pressure mean square deviation DPdMembership function parameter
The calculating of value can also apply 3 σ parameter statistics, choose after m continuous data asks for mean square deviation and carry out slip processing；M it is more big more
It can reflect true situation, however m chose conference and influences the real-time of inline diagnosis, was 50 in this value, finally using 3 σ
Parameter statistic obtains DPdMembership function parameter value be：A, b, c, d | brake-cylinder pressure mean square deviation }=0.0294,0.0733,
0.1173,0.1613 }.
5) measured data that monitoring platform collects is input to through the established fuzzy neural network of above step
In calculated, obtain diagnostic result.
To verify the validity of fuzzy neural network fault reasoning model established above, it is applied to crane
On-line fault diagnosis.
First, 2 training sample of table said extracted arrived, using Levenberg-Marquardt algorithms to fuzzy neural
Network model is trained, and calculates rapid convergence by 4 suboptimization, object function reaches 1.93659e-006, obtains good
Effect.Then, after input variable data being carried out fuzzy quantization pretreatment, the neural network being sent into after training is calculated,
Obtain the output Y of each neuron of output layerj（J=1,2..., 5）, it is last according to such as lower threshold value principle failure judgement state and reason
Event：Work as YjWhen >=0.8, fault mode F is representedjOccur；As 0.4≤Yj<When 0.8, fault mode F is representedjIt may occur.
It is known（298.6843 0.773,0.0327）It is that one group of motor rises mouth pressure, checking cylinder control mouth pressure and checking cylinder
Control mouth pressure mean square deviation actual condition data.By this group of floor data after blurring, each input variable is to the person in servitude of respective domain
Category degree is（0.9000,0.1000,0.1608）, obtaining output by network calculations is：（-0.0321-
0.10470.03810.99320.1031）, can determine whether that the corresponding fault mode of the failure is F by above-mentioned discrimination principle4, i.e., it is minimum
Cut set X179 failure of pressure reducing valve or minimal cut set X18 two-way electromagnetic valve failures.Actual state is two-way electromagnetic valve disconnecting.Diagnosis knot
Fruit is consistent with actual state.In addition, fuzzy neural network is to zero load and 40 tons（Steel wire rope multiplying power is 8）It is defeated under fault-free operating mode
Enter variable data to be calculated, output result is respectively（0.9996-0.0012-0.00020.0001-0.0015）、
（0.94180.1473-0.07440.0250-0.0419）.
It follows that the fuzzy neural network in the method for the present invention is to being both made that correct judgement：System pair
The fault mode answered is F1, i.e. system worked well, fault-free.More than diagnostic result shows the crane diagnosis side of the present invention
Method practicability and effectiveness.
Claims (1)
1. a kind of autocrane method for diagnosing faults based on fault tree and fuzzy neural network, it is characterized in that, to certain automobile
The failure of crane " promoting attonity " is diagnosed, and is implemented according to following steps：
1) certain autocrane top event fault tree is established using deduction
Certain autocrane hoisting circuit, structure mainly include, and variable pump (3) is connected with fuel tank (1), variable pump (3) and oil
Oil filter (2) is additionally provided between case (1), variable pump (3) is connected with swivel joint (4), swivel joint (4) respectively with overflow
Valve (5), the first pressure reducing valve (6), the second pressure reducing valve (9), reversal valve (11) connection, the first pressure reducing valve (6) and reversal valve (11) it
Between be parallel with proportional pressure-reducing valve (A7) and proportional pressure-reducing valve (B8)；Reversal valve (11) is connected with hoist motor (13), reversal valve
(11) balanced valve (12) is additionally provided between hoist motor (13)；Second pressure reducing valve (9) by two-way electromagnetic valve (10) with
Hydraulic braking cylinder (14) connects；Hoist motor (13) passes sequentially through shaft coupling (15), retarder (16) and reel (17) transmission and connects
It connects, reel (17) hangs weight by pulley；
Fault tree models of the autocrane top event fault tree for " promoting attonity " top event are established using deduction, and
To each bottom event of the top event to encoding, as shown in table 1,
1 bottom event coding schedule of table
2) network inputs of fuzzy neural network, network output node number are determined according to fault tree branch case and Heuristics
Mesh establishes structure of fuzzy neural network model；
In 19 bottom events of table 1, each independent bottom event is exactly a minimal cut set, and failure can be led to by sharing 2 branches
The generation of treetop event, branch CL-1 represent that motor rises mouth insufficient pressure and directly contributes the event chain for promoting attonity, branch CL-
2, which describe motor, rises the excessive failure exception reason of mouth pressure；Therefore, 2 fuzzy neural networks are established respectively to branch CL-1
Parallel diagnosis is carried out with branch CL-2, the structure of each automatic network is then determined according to the knowledge contained in each branch of fault tree；
Branch CL-2 include 6 minimal cut sets for hoist motor stuck (bottom event 14), winding speed reducing failure (bottom event 15),
Brake stuck (bottom event 16), the second pressure reducing valve (9) failure (bottom event 17), two-way electromagnetic valve failure (bottom event 18), system
Dynamic the cylinder piston wear and leakage (bottom event 19)；Motor is not in build the pressure to equipment in normal conditions, and above 6 minimal cut sets
The generation of any one minimal cut set can all motor be caused to build the pressure, motor rises mouth pressure PmIt is excessive；Minimal cut set X14Or X15Generation,
It will not be to checking cylinder control mouth pressure PdIt has an impact, and X16Generation can make PdValue is higher by normal range value and has larger pressure
Fluctuation, X17、X18Or X19Generation can then make PdValue is less than normal range value, X19Generation can make PdKeep larger pressure wave
It is dynamic, and X17Or X18Generation can make PdThen keep smaller pressure oscillation；PdPressure oscillation situation use its mean square deviation DPdCome
Characterization；System is only needed by PmAnd PdState-detection, obtain 3 input variables：Pm、Pd、DPd, following 5 kinds of events can be distinguished
Barrier pattern：Normal work, { X14Or X15}、X16、{X17Or X18}、X19；Thus the corresponding fuzznets of branch CL-2 be can determine
Network structure has the structure of 3 input, 5 output, and hidden layer neuron number of nodes is rule of thumb taken as 6, i.e., 3 × 6 × 5；Wherein, it is hidden
For neuron in layer using logarithm Sigmoid type functions, the neuron of output layer uses purely linear purelin transforming function transformation functions；
3) according to the knowledge contained in each branch of fault tree, training sample is extracted, and neural network is trained, establishes god
Network weight and threshold matrix needed for being calculated through network reasoning；
According to the analysis of step 2), the training sample data of fuzzy neural network are extracted, as shown in table 2, in table,Value 0.1,
0.5th, 0.9 represent that motor rises mouth pressure P respectivelymIt is too low, normal, excessively high；Value 0.1,0.5,0.9 represents P respectivelydChecking cylinder
Control mouth hypotony, normal, excessively high, DPdValue 0.1,0.5,0.9 represents that checking cylinder control mouth pressure mean square deviation is too low, just respectively
Often, it is excessively high；
2 train samples of table
4) using the historical data on autocrane status monitoring platform, mould is established with 3 σ Criterion Methods in statistical parameter method
Fuzzy membership functions needed for paste pretreatment；
First, the person in servitude of each input variable is determined in the distribution situation of input variable interval according to the sample set of input variable
Membership fuction, the low conversion in state variable domain determine power using the membership function of lower semi-trapezoid distribution, expression formula such as formula (3)
Value coefficient is 0.1；State variable domain normally conversion all use trapezoidal profile membership function, expression formula such as formula (4), and really
It is 0.5 to determine weight coefficient；State variable domain is high to convert using the membership function for rising half trapezoidal profile, expression formula such as formula (5),
And determining that weight coefficient is 0.9, the setting of wherein weights is to handle the detection data being between two states, utilize each person in servitude
The average weighted method of membership fuction is by the data conversion between two states into fuzzy data；
Secondly, the value of each parameter of membership function is determined, Diagnostic parameters y Normal Distribution rules when equipment works normally are right
Diagnostic parameters collects a certain amount of sample data during normal work, and statistical disposition is carried out to these data, if test sample with
Mean valueDistance is more than 2 times or 3 times of variances, then it is assumed that the test sample is uncertain, therefore can be judged to exception；Then it obtains
Alarm threshold：Failure bound parameter value is higher or relatively low to beParameter normal limits areMore meet objective reality to obtain
The membership function on border makes the fuzzy net that step is established enter practical application, at this with 3 σ in statistical parameter method
Criterion Method determines the value of each parameter of membership function；To input variable (Pm、Pd、DPd) sample data using 3 σ Criterion Methods, obtain
The value of membership function parameter a, b, c, d of each input variable are then updated in formula (3)~formula (5), obtain each input variable
Membership function establishes the membership function of fuzzy neural network；
Similarly, checking cylinder control mouth pressure P is obtaineddMembership function parameter value is：A, b, c, d | brake-cylinder pressure (bar) }=
{ 36.1001,36.3344,36.5688,36.6031 }；It chooses after m continuous data asks for mean square deviation and carries out slip processing；M is got over
It is big more to reflect true situation, however m chose conference and influences the real-time of inline diagnosis, is 50 in this value, finally should
D is obtained with 3 σ parameter statisticsPdMembership function parameter value be：A, b, c, d | brake-cylinder pressure mean square deviation }=0.0294,
0.0733,0.1173,0.1613 }；
5) by the measured data that monitoring platform collects be input to by the established fuzzy neural network of above step into
Row calculates, and obtains diagnostic result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310492933.2A CN103544389B (en) | 2013-10-18 | 2013-10-18 | Autocrane method for diagnosing faults based on fault tree and fuzzy neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310492933.2A CN103544389B (en) | 2013-10-18 | 2013-10-18 | Autocrane method for diagnosing faults based on fault tree and fuzzy neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103544389A CN103544389A (en) | 2014-01-29 |
| CN103544389B true CN103544389B (en) | 2018-07-10 |
Family
ID=49967834
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310492933.2A Active CN103544389B (en) | 2013-10-18 | 2013-10-18 | Autocrane method for diagnosing faults based on fault tree and fuzzy neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103544389B (en) |
Families Citing this family (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103984229B (en) * | 2014-05-04 | 2017-06-06 | 永春桂洋常泰农业专业合作社 | A kind of neural network control method of lifting mechanism of tower crane governing system |
| CN104454785B (en) * | 2014-09-25 | 2016-09-07 | 中南大学 | A kind of engineering machinery hydraulic cylinder fault diagnosis system and the fault sample signal acquisition method being suitable for therewith |
| CN104680023B (en) * | 2015-03-12 | 2017-08-29 | 重庆科技学院 | Oil pumper parameter optimization method based on multiobjective decision-making |
| CN105046326B (en) * | 2015-03-13 | 2019-04-19 | 重庆科技学院 | Pumping unit parameter optimization method based on function figure pivot analysis |
| CN105045941B (en) * | 2015-03-13 | 2018-11-02 | 重庆科技学院 | Pumping unit parameter optimization method based on Unscented kalman filtering |
| CN106959681B (en) * | 2016-01-12 | 2023-06-06 | 武汉科技大学 | FHN-based urban rail train door fault diagnosis method |
| CN107037802B (en) * | 2016-10-28 | 2018-01-26 | 华中科技大学 | A kind of method for detecting abnormality of Process Control System protecting information safety |
| CN107463146A (en) * | 2017-08-18 | 2017-12-12 | 大连美恒时代科技有限公司 | A kind of method that crane failure is diagnosed by MHCIMS |
| CN107563069A (en) * | 2017-09-06 | 2018-01-09 | 国电联合动力技术有限公司 | A kind of wind power generating set intelligent fault diagnosis method |
| CN108506112B (en) * | 2018-03-05 | 2019-11-22 | 中国人民解放军国防科技大学 | A kind of pintle formula ejector filler Variable Thrust Engine fault detection method |
| CN108639062B (en) * | 2018-05-08 | 2019-11-19 | 徐州徐工环境技术有限公司 | A kind of Vehicular intelligent control system |
| EP3570130B1 (en) * | 2018-05-15 | 2020-12-16 | Siemens Industry Software NV | Ring-closures in fault trees |
| CN109284818A (en) * | 2018-09-07 | 2019-01-29 | 北方爆破科技有限公司 | A kind of blasting vibration control prediction technique based on accident tree and genetic algorithm |
| CN109670595A (en) * | 2019-03-04 | 2019-04-23 | 丽水学院 | A kind of equipment fault diagnosis method |
| CN110005659B (en) * | 2019-04-08 | 2020-08-11 | 深圳市华星光电技术有限公司 | Real-time monitoring method for air cylinder |
| CN110097219B (en) * | 2019-04-19 | 2020-04-24 | 深圳市德塔防爆电动汽车有限公司 | Electric vehicle operation and maintenance optimization method based on safety tree model |
| CN111046940A (en) * | 2019-12-06 | 2020-04-21 | 中车长春轨道客车股份有限公司 | Vehicle door fault diagnosis method based on decision tree and neural network |
| CN111372049B (en) * | 2020-03-11 | 2021-10-08 | 上海工程技术大学 | Rail transit hazard source identification and intelligent early warning device and method |
| US20210284179A1 (en) * | 2020-03-11 | 2021-09-16 | Ford Global Technologies, Llc | Vehicle health calibration |
| CN112462716A (en) * | 2020-03-30 | 2021-03-09 | 林细兵 | Equipment state identification method and system based on industrial Internet |
| CN112884210B (en) * | 2021-01-31 | 2022-03-15 | 中国人民解放军63963部队 | Vehicle health management system based on fuzzy clustering |
| CN112906764B (en) * | 2021-02-01 | 2023-09-26 | 中国人民解放军海军工程大学 | Communication safety equipment intelligent diagnosis method and system based on improved BP neural network |
| CN113705615B (en) * | 2021-07-29 | 2024-05-28 | 南京邮电大学 | Multi-stage equipment fault diagnosis method and system for electric automobile charging process based on neural network |
| CN115468718B (en) * | 2022-08-24 | 2024-06-14 | 大连海事大学 | Diagnosis method for leakage fault of ship wind wing rotary hydraulic system |
| CN115390550A (en) * | 2022-10-27 | 2022-11-25 | 潍柴雷沃智慧农业科技股份有限公司 | Fault diagnosis method and system of combine harvester, controller and combine harvester |
| CN115796589B (en) * | 2022-12-05 | 2023-09-29 | 三亚学院 | Sensitivity analysis method for risk hidden danger of three-electric system of pure electric vehicle |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1300694C (en) * | 2003-06-08 | 2007-02-14 | 华为技术有限公司 | Fault tree analysis based system fault positioning method and device |
| CN101093559B (en) * | 2007-06-12 | 2010-06-23 | 北京科技大学 | Method for constructing expert system based on knowledge discovery |
| CN101839805A (en) * | 2010-03-19 | 2010-09-22 | 同济大学 | Method for quality inspection of active fault and diagnosis of intelligent fault of engine |
| CN101846992B (en) * | 2010-05-07 | 2011-12-07 | 上海理工大学 | Fault tree construction method based on fault case of numerical control machine |
| CN102968113A (en) * | 2012-11-16 | 2013-03-13 | 国电南瑞科技股份有限公司 | Failure analysis and exhibition method of power generator excitation system |
| CN103020436B (en) * | 2012-11-30 | 2015-08-12 | 工业和信息化部电子第五研究所 | Component failure zero analytical approach and system |
-
2013
- 2013-10-18 CN CN201310492933.2A patent/CN103544389B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN103544389A (en) | 2014-01-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103544389B (en) | Autocrane method for diagnosing faults based on fault tree and fuzzy neural network | |
| CN203497948U (en) | Fuzzy neural network fault diagnosis device of auto crane | |
| CN111507489B (en) | Cloud-edge-coordinated amusement equipment fault prediction and health management system and method | |
| CN103810328B (en) | Transformer maintenance decision method based on hybrid model | |
| CN104992270B (en) | Power transmission and transformation equipment state overhauling aid decision-making system and method | |
| CN107991870A (en) | A kind of fault pre-alarming and life-span prediction method of Escalator equipment | |
| CN103400310B (en) | Method for evaluating power distribution network electrical equipment state based on historical data trend prediction | |
| CN107330529A (en) | Heavy-duty mill roll loads the failure risk appraisal procedure of intelligent robot | |
| CN103678952A (en) | Elevator risk evaluation method | |
| CN108069308A (en) | A kind of electric staircase failure diagnosis method based on sequential probability | |
| CN104573850A (en) | Method for evaluating state of thermal power plant equipment | |
| CN115034578A (en) | Intelligent management construction method and system for hydraulic metal structure equipment based on digital twinning | |
| CN103559648A (en) | Grid equipment state inspection and evaluation training system | |
| CN103020429A (en) | Comprehensive decision-making and evaluating method for health condition of tied-arch bridge | |
| CN110209999A (en) | A kind of mobile unit failure trend prediction method | |
| CN107392324A (en) | The specialized maintenance total management system of transformer | |
| CN109087016A (en) | Harbour gantry crane power-save operation quantization assessment method and system | |
| CN109444575A (en) | Motor driving system runs health state evaluation system and method | |
| CN110794806A (en) | Failure probability calculation method for fault tree bottom event based on fuzzy theory | |
| CN110807580A (en) | Method for analyzing key safety risk of super high-rise construction machinery based on complex network | |
| CN106569052A (en) | Real-time health status-considered method for evaluating reliability of power transformer | |
| CN114757365A (en) | High-speed railway roadbed settlement prediction and early warning method based on deep learning | |
| CN109408998A (en) | Estimating method for fatigue life is carried out based on sample incremental quick obtaining stress spectra | |
| CN109886439A (en) | Electronic-control vehicle remote diagnosis system and its diagnostic method | |
| CN104085789A (en) | Intelligent monitoring method of running state of crane |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |