CN104008423B - Method and system for predicting life of gear case according to wind conditions - Google Patents
Method and system for predicting life of gear case according to wind conditions Download PDFInfo
- Publication number
- CN104008423B CN104008423B CN201310060297.6A CN201310060297A CN104008423B CN 104008423 B CN104008423 B CN 104008423B CN 201310060297 A CN201310060297 A CN 201310060297A CN 104008423 B CN104008423 B CN 104008423B
- Authority
- CN
- China
- Prior art keywords
- life
- wind speed
- scheduled time
- training sample
- wind
- 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 Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
- General Details Of Gearings (AREA)
- Wind Motors (AREA)
Abstract
The invention discloses a method and system for predicting the life of a gear case according to wind conditions. The method comprises: according to a predetermined time interval, acquiring wind field wind speed data, and performing continuous acquisition for first predetermined time; counting the wind speed data, and calculating a wind speed change rate; performing normalization processing on obtained first accumulation frequency, second accumulation frequency, first accumulation time and second accumulation to obtain a first training sample element, a second training sample element, a third training sample element and a fourth training sample element as an input layer of a nerve network; continuously acquiring the load spectrum of the gear box within the first predetermined time; calculating to obtain gear case life lost within every second predetermined time; taking a result obtained after the normalization processing is performed on the gear case life as the output of the nerve network, and constructing the nerve network; arranging a training error; and when a training deviation is smaller than the training error, determining that the nerve network is successfully trained. By using the method and system for predicting the life of the gear case according to the wind conditions, provided by the invention, the fatigue life of the gear case can be scientifically and accurately evaluated.
Description
Technical field
The present invention relates to technical field of wind power generator, more particularly to a kind of method for predicting gearbox life according to wind regime
And system.
Background technology
Wind energy is a kind of regenerative resource of cleaning.Used as the principal mode of Wind Power Utilization, wind-power electricity generation is current technology
It is most ripe, one of renewable energy power generation mode of on the largest scaleization exploit condition and commercialized development prospect.
In wind power generating set manufacture, geared teeth roller box is a critical component, once catastrophic failure, often makes
User is caught unprepared, and causes great economic loss, even results in catastrophic effect.
According to statistics, in various mechanical breakdowns, gear failure accounts for the 60% of sum, and the destruction 50%-90% of machine components
It is because fatigue rupture is caused.
Due to wind turbine gearbox it is expensive, and installation, debug time length, complex management, therefore, how scientifically and accurately
The fatigue life of assessment gear-box, the in good time maintenance for gear-box provides correct theoretical direction, so as to avoid owing to safeguard and mistake
Safeguard that phenomenon occurs, it is final to realize improving operating efficiency, it is ensured that safety in production, wind-resources availability of blower fan etc. is improved, all
Tool is of great significance.
Existing conventional gearbox life Forecasting Methodology has S-N nominal stress methods, and e-N local strain methods, LEFM crackles expand
Exhibition Life method etc..Because the life-span of gear-box is both controlled by internal structure, and by external operating environment and the shadow of working condition
Ring, specifically there is the nonlinear dissipation of height, existing conventional gearbox life Forecasting Methodology is difficult to predict gear exactly
The life-span of case.
The content of the invention
The technical problem to be solved in the present invention is to provide a kind of method and system that gearbox life is predicted according to wind regime, uses
In the fatigue life for scientifically and accurately assessing gear-box, the in good time maintenance for gear-box provides correct theoretical direction.
First aspect according to embodiments of the present invention, there is provided a kind of method that gearbox life is predicted according to wind regime, institute
The method of stating includes：
To schedule the wind speed of interval collection wind field obtains air speed data, and the scheduled time of continuous acquisition first；Institute
Stated for first scheduled time including several second scheduled times；
The air speed data in second scheduled time each described is counted respectively, calculates [vin, vout] interval in
The wind speed rate of change Δ v of the adjacent wind speed of any two；Wherein, vin, voutRespectively Wind turbines switch in and out wind speed；
Count wind speed rate of change Δ v in each described second scheduled time and meet Δ v1＜ Δ v ＜ Δ v2The first of condition is tired out
Plus frequency nI, 1；
Count wind speed rate of change Δ v in each described second scheduled time and meet Δ v2Cumulative time of the second of ＜ Δ v conditions
Number nI, 2；
Count wind speed in each described second scheduled time and be in [vin, vrated] interval in the first accumulation interval tI, 1,
Wherein vratedFor rated wind speed；
Count wind speed in each described second scheduled time and be in (vrated,vout] interval in the second accumulation interval tI, 2；
With the first accumulative frequency nI, 1, the second accumulative frequency nI, 2, the first accumulation interval tI, 1, the second accumulation interval tI, 2Carry out
Normalized obtains the first training sample element xI, 1, the second training sample element xI, 2, the 3rd training sample element xI, 3,
Four training sample element xI, 4, as the input layer parameter of neutral net；
The loading spectrum of the gear-box of first scheduled time described in continuous acquisition；
It is calculated the gear life y of each second scheduled time lossi’；
By the gear life y of the lossi' the 5th training sample element y is obtained after normalizedi, as neutral net
Output parameter, build the neutral net；
Training error ε is set；
Train the neutral net, train deviation be less than the training error when, determine the neural metwork training into
Work(；
Using former years wind regime data, the service life of gear-box according to the neural network prediction.
Preferably, it is described with the first accumulative frequency nI, 1, the second accumulative frequency ni,2, the first accumulation interval tI, 1, second add up
Time ti,2The employing formula being normalized：
Obtain the first training sample element xi,1, the second training sample element xi,2, the 3rd training sample element xi,3、
4th training sample element xi,4。
Preferably, the gear life y ' of the lossiNormalized adopt formula：
Obtain the 5th training sample element yi；Wherein, Y is that to calculate each using nominal stress method described second pre-
Fix time the gear life y of lossi' cumulative sum.
Preferably, methods described further includes, the initial weight that the neutral net is set be in [0,1] interval with
Machine number.
Preferably, the hidden layer node quantity of the neutral net is 5 or 6.
Preferably, the predetermined time interval is 1 second, and/or, first scheduled time is 12 months, and/or, institute
Stated for second scheduled time for one month.
Second aspect according to embodiments of the present invention, discloses a kind of system for predicting gearbox life according to wind regime, institute
The system of stating includes：
Wind speed collecting unit, the wind speed for being spaced collection wind field to schedule obtains air speed data, and continuously adopts
Collected for first scheduled time；First scheduled time includes several second scheduled times；
First computing unit, for counting to the air speed data in second scheduled time each described respectively, calculates
Go out [vin, vout] interval in the adjacent wind speed of any two wind speed rate of change Δ v；Wherein, vin, voutRespectively Wind turbines
Switch in and out wind speed；
Statistic unit, for counting wind speed rate of change Δ v in each described second scheduled time Δ v is met1＜ Δ v ＜ Δs
v2First accumulative frequency n of conditionI, 1；
And, meet Δ v for counting wind speed rate of change Δ v in each described second scheduled time2The of ＜ Δ v conditions
Two accumulative frequency ni,2；
And, it is in [v for counting wind speed in each described second scheduled timein, vrated] interval in first add up
Time tI, 1, wherein vratedFor rated wind speed；
And, it is in (v for counting wind speed in each described second scheduled timerated,vout] interval in second add up
Time ti,2；
First normalized unit, for the first accumulative frequency ni,1, the second accumulative frequency ni,2, the first accumulation interval
tI, 1, the second accumulation interval tI, 2It is normalized and obtains the first training sample element xI, 1, the second training sample element xI, 2、
3rd training sample element xI, 3, the 4th training sample element xi,4, as the input layer parameter of neutral net；
Loading spectrum collecting unit, for the loading spectrum of the gear-box of first scheduled time described in continuous acquisition；
Second computing unit, for being calculated each described second scheduled time（Monthly）The gear life y of lossi’；
Second normalized unit, for by the gear life y of the lossi' the 5th instruction is obtained after normalized
Practice sample elements yi, as the output parameter of neutral net, build the neutral net；
Training error setting unit, for arranging training error ε；
Neural metwork training unit, for training the neutral net, when training deviation to be less than the training error, really
The fixed neural metwork training success；
Life forecast unit, for using former years wind regime data, the gear-box according to the neural network prediction
Service life.
Preferably, the first normalized unit is according to formula：
Obtain the first training sample element xI, 1, the second training sample element xI, 2, the 3rd training sample element xI, 3、
4th training sample element xI, 4。
Preferably, the second normalized unit is according to formula：
Obtain the 5th training sample element yi；Wherein, Y is that to calculate each using nominal stress method described second pre-
Fix time the gear life y of lossi' cumulative sum.
Preferably, the system further includes initial weight setting unit, for arranging the initial of the neutral net
Weights are the random number in [0,1] interval.
Preferably, the hidden layer node quantity of the neutral net is 5 or 6.
Preferably, the predetermined time interval is 1 second, and/or, first scheduled time is 12 months, and/or, institute
Stated for second scheduled time for one month.
Compared with prior art, the present invention has advantages below：
The method that gearbox life is predicted according to wind regime that the present invention is provided, due to the use longevity according to wind speed and gear-box
Life establish neutral net, it is possible to use the multimodal functional realiey of neutral net with highly precise approach Any Nonlinear Function,
Realize the science Accurate Prediction of gearbox life.
Description of the drawings
Fig. 1 is to predict the method flow diagram of gearbox life according to wind regime described in the embodiment of the present invention；
Fig. 2 is the neutral net schematic diagram that the embodiment of the present invention builds；
Fig. 3 is to predict the system construction drawing of gearbox life according to wind regime described in the embodiment of the present invention.
Specific embodiment
It is understandable to enable the above objects, features and advantages of the present invention to become apparent from, below in conjunction with the accompanying drawings to the present invention
Specific embodiment be described in detail.
The present invention provides a kind of method and system that gearbox life is predicted according to wind regime, for scientifically and accurately assessing tooth
The fatigue life of roller box, the in good time maintenance for gear-box provides correct theoretical direction.
It is that the method flow diagram of gearbox life is predicted according to wind regime described in the embodiment of the present invention referring to Fig. 1 and Fig. 2, Fig. 1；
Fig. 2 is the neutral net schematic diagram that the embodiment of the present invention builds.
The method for predicting gearbox life according to wind regime described in the embodiment of the present invention, methods described includes：
S100, the to schedule wind speed of interval collection wind field obtain air speed data, and the pre- timing of continuous acquisition first
Between；First scheduled time includes several second scheduled times.
Predetermined time interval is specifically as follows 1 second, precision can need to be set as needed.
First scheduled time specifically can be with 12 months；Second scheduled time was specifically as follows one month.First scheduled time,
Second scheduled time, equally precision can need to be set as needed.
S200, the air speed data in second scheduled time each described counted respectively, calculate [vin, vout] area
Between in the adjacent wind speed of any two wind speed rate of change Δ v；Wherein, vin, voutRespectively Wind turbines switch in and out wind
Speed.
Specifically, the air speed data of every month counted respectively.
When predetermined time interval is specially 1 second, the wind speed rate of change Δ v is exactly the speed difference between adjacent two seconds.
S300, count wind speed rate of change Δ v in each described second scheduled time and meet Δ v1＜ Δ v ＜ Δ v2Condition
First accumulative frequency nI, 1。
Δv1With Δ v2Equally precision can need to be set as needed, in a specific embodiment of the invention,
Δv10.5m/s can be taken;Δv24m/s can be taken.
S400, count wind speed rate of change Δ v in each described second scheduled time and meet Δ v2The second of ＜ Δ v conditions is tired out
Plus frequency ni,2。
S500, count in each described second scheduled time wind speed and be in [vin, vrated] interval in the first accumulation interval
tI, 1, wherein VratedFor rated wind speed.
S600, count in each described second scheduled time wind speed and be in (vrated,vout] interval in the second accumulation interval
tI, 2。
S700, with the first accumulative frequency nI, 1, the second accumulative frequency nI, 2, the first accumulation interval tI, 1, the second accumulation interval
ti,2It is normalized and obtains the first training sample element xi,1, the second training sample element xI, 2, the 3rd training sample element
xI, 3, the 4th training sample element xI, 4, as the input layer parameter of neutral net.
It is described with the first accumulative frequency nI, 1, the second accumulative frequency ni,1, the first accumulation interval ti,1, the second accumulation interval ti,2
Being normalized can adopt formula：
Obtain the first training sample element xI, 1, the second training sample element xI, 2, the 3rd training sample element xI, 3、
4th training sample element xI, 4。
S800, with the loading spectrum of the gear-box of first scheduled time described in continuous acquisition.
S900, the gear life y for being calculated each second scheduled time lossi’。
The gear life y of each second scheduled time loss specifically can be calculated using nominal stress methodi’。
S1000, by the gear life y of the lossi' the 5th training sample element y is obtained after normalizedi, as god
The output parameter of Jing networks, builds the neutral net.
Because neutral net has the advantages that its is unique, i.e., with fault-tolerant, association, supposition, memory, self adaptation, self study
With process complicated multimodal function, and may certify that three-layer neural network can approach any non-linear letter with arbitrary accuracy
Number, this is that gearbox life prediction opens up a new way.
Because three-layer neural network just can be realized with arbitrary accuracy Approximation of Arbitrary Nonlinear Function, therefore the present invention is implemented
Example methods described can arrange structure three-layer neural network.
The gear life y of the lossi' normalized adopt formula：
Obtain the 5th training sample element yi；Wherein, Y is that to calculate each using nominal stress method described second pre-
Fix time the gear life y of lossi' cumulative sum.
S1100, setting training error ε.
Wherein, the concrete value of training error ε can be determined according to required gearbox life precision of prediction.
S1200, the training neutral net, when training deviation to be less than the training error, determine the neutral net
Train successfully.
Wherein, deviation is trained to be difference of the neutral net in training between reality output and desired output, this value is less
Then represent neural metwork training more successful.
S1300, using former years wind regime data, the service life of gear-box according to the neural network prediction.
The wind regime data in statistics former years, as the input data of neutral net, draw the prediction loss of following every month
Life-span, the life-span of the actual loss that adds up, if being more than the design of gear-box in cumulative sum of following j-th month actual loss life-span
During the life-span, then gearbox failure is can be determined that.The concrete predicted time of gearbox failure is obtained, so as to obtain the pre- of gear-box
Survey the life-span.
The method that gearbox life is predicted according to wind regime that the present invention is provided, due to the use longevity according to wind speed and gear-box
Life establish neutral net, it is possible to use the multimodal functional realiey of neutral net with highly precise approach Any Nonlinear Function,
Realize the science Accurate Prediction of gearbox life.
Embodiment of the present invention methods described may further include, and the initial weight for arranging the neutral net is [0,1]
Random number in interval.Wherein, weights are the parameters for representing the bonding strength in neutral net between two neurons, nerve net
Network just can determine that after training weights.Initial weight general random before training gives, it is preferred that can arrange the nerve net
The initial weight of network is the random number in [0,1] interval.But in the training process, weights can be changed at any time as needed,
So that the continuous close desired output of the output of neutral net.Once completing, then weights determine for training, and neutral net can be used to predict
Gearbox life.
In embodiments of the present invention, the input layer of neutral net is responsible for receiving from extraneous input information, and passes to
Hidden layer；Hidden layer is the internal information process layer of neutral net, is responsible for information conversion；Outwardly output information is processed output layer
As a result.The nodes of hidden layer are estimated according to the complexity of target problem, in the training process of neutral net, can be with
The nodes of hidden layer are suitably adjusted according to training result.
According to neural network theory, the nodes of hidden layer should be greater than the node number of input layer.Input layer shown in Fig. 2
Node number is 4, therefore, the hidden layer node quantity of the neutral net can be 5 or 6.Fig. 2 show the feelings for selecting 5
Condition.
Referring to Fig. 3, the figure is to predict the system construction drawing of gearbox life according to wind regime described in the embodiment of the present invention.
The system that gearbox life is predicted according to wind regime described in the embodiment of the present invention, including：
Wind speed collecting unit 11, the wind speed for being spaced collection wind field to schedule obtains air speed data, and continuously
Gathered for first scheduled time；First scheduled time includes several second scheduled times.
Predetermined time interval is specifically as follows 1 second, precision can need to be set as needed.
First scheduled time specifically can be with 12 months；Second scheduled time was specifically as follows one month.First scheduled time,
Second scheduled time, equally precision can need to be set as needed.
First computing unit 12, for counting to the air speed data in second scheduled time each described respectively, counts
Calculate [vin, vout] interval in the adjacent wind speed of any two wind speed rate of change Δ v；Wherein, vin, voutRespectively wind turbine
That what is organized switches in and out wind speed；
When second scheduled time is one month, specifically, the air speed data of every month is counted respectively.
When predetermined time interval is specially 1 second, the wind speed rate of change Δ v is exactly the speed difference between adjacent two seconds.
Statistic unit 13, for counting wind speed rate of change Δ v in each described second scheduled time Δ v is met1＜ Δ v ＜
Δv2First accumulative frequency n of conditionI, 1；
And, meet Δ v for counting wind speed rate of change Δ v in each described second scheduled time2The of ＜ Δ v conditions
Two accumulative frequency ni,2；
And, it is in [v for counting wind speed in each described second scheduled timein, vrated] interval in first add up
Time ti,1, wherein VratedFor rated wind speed；
And, it is in (v for counting wind speed in each described second scheduled timerated,vout] interval in second add up
Time ti,2；
Δv1With Δ v2Equally precision can need to be set as needed.
First normalized unit 14, for the first accumulative frequency nI, 1, the second accumulative frequency nI, 2, first it is cumulative when
Between tI, 1, the second accumulation interval ti,2It is normalized and obtains the first training sample element xI, 1, the second training sample element
xi,2, the 3rd training sample element xI, 3, the 4th training sample element xI, 4, as the input layer parameter of neutral net；
First normalized unit 14 can adopt formula：
Obtain the first training sample element xI, 1, the second training sample element xI, 2, the 3rd training sample element xI, 3、
4th training sample element xI, 4。
Loading spectrum collecting unit 15, for the loading spectrum of the gear-box of first scheduled time described in continuous acquisition；
Second computing unit 16, for being calculated the gear life y of each second scheduled time lossi’；
Second computing unit 16 specifically can be calculated each described second scheduled time loss using nominal stress method
Gear life yi’。
Second normalized unit 17, for by the gear life y of the lossi' the 5th is obtained after normalized
Training sample element yi, as the output parameter of neutral net, build the neutral net.
Because three-layer neural network just can be realized with arbitrary accuracy Approximation of Arbitrary Nonlinear Function, therefore the present invention is implemented
Example methods described can arrange structure three-layer neural network.
Second normalized unit 17 can adopt formula：
Obtain the 5th training sample element yi；Wherein, Y is that to calculate each using nominal stress method described second pre-
Fix time the gear life y of lossi' cumulative sum.
Training error setting unit 18, for arranging training error ε.Wherein, the concrete value of training error ε can basis
Required gearbox life precision of prediction is determined.
Neural metwork training unit 19, for training the neutral net, when training deviation to be less than the training error,
Determine the neural metwork training success.Wherein, train deviation be neutral net in training reality output and desired output it
Between difference, this value is more little, represents neural metwork training more successful.
Life forecast unit 20, for using former years wind regime data, the gear according to the neural network prediction
The service life of case.
Life forecast unit 20 counts the wind regime data in former years, as the input data of neutral net, draws future
The life-span of the prediction loss of every month, the life-span of the actual loss that adds up, if adding up it in following j-th month actual loss life-span
During with the projected life for being more than gear-box, then gearbox failure is can be determined that.The concrete predicted time of gearbox failure is obtained,
So as to obtain the bimetry of gear-box.
System of the present invention can further include initial weight setting unit, for arranging the neutral net
Initial weight is the random number in [0,1] interval.
The system that gearbox life is predicted according to wind regime that the present invention is provided, due to the use longevity according to wind speed and gear-box
Life establish neutral net, it is possible to use the multimodal functional realiey of neutral net with highly precise approach Any Nonlinear Function,
Realize the science Accurate Prediction of gearbox life.
The above, is only presently preferred embodiments of the present invention, and any pro forma restriction is not made to the present invention.Though
So the present invention is disclosed above with preferred embodiment, but is not limited to the present invention.It is any to be familiar with those skilled in the art
Member, under without departing from technical solution of the present invention ambit, all using the methods and techniques content of the disclosure above to the present invention
Technical scheme makes many possible variations and modification, or the Equivalent embodiments for being revised as equivalent variations.Therefore, it is every without departing from
The content of technical solution of the present invention, according to the technical spirit of the present invention to any simple modification made for any of the above embodiments, equivalent
Change and modification, still fall within the range of technical solution of the present invention protection.
Claims (12)
1. it is a kind of according to wind regime predict gearbox life method, it is characterised in that methods described includes：
To schedule the wind speed of interval collection wind field obtains air speed data, and the continuous acquisition within first scheduled time；Institute
Stated for first scheduled time including several second scheduled times；
The air speed data in second scheduled time each described is counted respectively, calculates [vin,vout] interval in it is any
The wind speed rate of change Δ v of two adjacent wind speed；Wherein, vin,voutRespectively Wind turbines switch in and out wind speed；
Count in each described second scheduled time, [the vin,vout] interval in the adjacent wind speed of any two wind speed change
Rate Δ v meets Δ v1＜ Δ v ＜ Δ v2First accumulative frequency n of conditioni,1；
Count in each described second scheduled time, [the vin,vout] interval in the adjacent wind speed of any two wind speed change
Rate Δ v meets Δ v2Second accumulative frequency n of ＜ Δ v conditionsi,2；
Count wind speed in each described second scheduled time and be in [vin,vrated] interval in the first accumulation interval ti,1, wherein
vratedFor rated wind speed；
Count wind speed in each described second scheduled time and be in (vrated,vout] interval in the second accumulation interval ti,2；
With the first accumulative frequency ni,1, the second accumulative frequency ni,2, the first accumulation interval ti,1, the second accumulation interval ti,2Carry out normalizing
Change is processed and obtains the first training sample element xi,1, the second training sample element xi,2, the 3rd training sample element xi,3, the 4th instruction
Practice sample elements xi,4, as the input layer parameter of neutral net；
The loading spectrum of the gear-box of first scheduled time described in continuous acquisition；
It is calculated the gear life y ' of each second scheduled time lossi；
By the gear life y ' of the lossiThe 5th training sample element y is obtained after normalizedi, as the defeated of neutral net
Go out parameter, build the neutral net；
Training error ε is set；
The neutral net is trained, when training deviation to be less than the training error, the neural metwork training success is determined；
Using former years wind regime data, the service life of gear-box according to the neural network prediction.
2. the method that gearbox life is predicted according to wind regime according to claim 1, it is characterised in that described tired with first
Plus frequency ni,1, the second accumulative frequency ni,2, the first accumulation interval ti,1, the second accumulation interval ti,2The employing being normalized
Formula：
Obtain the first training sample element xi,1, the second training sample element xi,2, the 3rd training sample element xi,3, the 4th
Training sample element xi,4。
3. it is according to claim 1 according to wind regime predict gearbox life method, it is characterised in that the tooth of the loss
Wheel life-span y 'iNormalized adopt formula：
Obtain the 5th training sample element yi；Wherein, Y is to calculate each described second pre- timing using nominal stress method
Between lose gear life y 'iCumulative sum.
4. it is according to claim 1 according to wind regime predict gearbox life method, it is characterised in that methods described enters one
Step includes that the initial weight for arranging the neutral net is the random number in [0,1] interval.
5. it is according to claim 1 according to wind regime predict gearbox life method, it is characterised in that the neutral net
Hidden layer node quantity be 5 or 6.
6. it is according to claim 1 according to wind regime predict gearbox life method, it is characterised in that the scheduled time
At intervals of 1 second, and/or, first scheduled time is 12 months, and/or, second scheduled time is one month.
7. it is a kind of according to wind regime predict gearbox life system, it is characterised in that the system includes：
Wind speed collecting unit, the wind speed for being spaced collection wind field to schedule obtains air speed data, and predetermined first
Continuous acquisition in time；First scheduled time includes several second scheduled times；
First computing unit, for counting to the air speed data in second scheduled time each described respectively, calculates
[vin,vout] interval in the adjacent wind speed of any two wind speed rate of change Δ v；Wherein, vin,voutRespectively Wind turbines
Switch in and out wind speed；
Statistic unit, for counting in each described second scheduled time, [the vin,vout] interval in any two it is adjacent
The wind speed rate of change Δ v of wind speed meets Δ v1＜ Δ v ＜ Δ v2First accumulative frequency n of conditioni,1；
And, for counting in each described second scheduled time, [the vin,vout] interval in the adjacent wind speed of any two
Wind speed rate of change Δ v meet Δ v2Second accumulative frequency n of ＜ Δ v conditionsi,2；
And, it is in [v for counting wind speed in each described second scheduled timein,vrated] interval in the first accumulation interval
ti,1, wherein vratedFor rated wind speed；
And, it is in (v for counting wind speed in each described second scheduled timerated,vout] interval in the second accumulation interval
ti,2；
First normalized unit, for the first accumulative frequency ni,1, the second accumulative frequency ni,2, the first accumulation interval ti,1、
Second accumulation interval ti,2It is normalized and obtains the first training sample element xi,1, the second training sample element xi,2, the 3rd
Training sample element xi,3, the 4th training sample element xi,4, as the input layer parameter of neutral net；
Loading spectrum collecting unit, for the loading spectrum of the gear-box of first scheduled time described in continuous acquisition；
Second computing unit, for being calculated the gear life y ' of each second scheduled time lossi；
Second normalized unit, for by the gear life y ' of the lossiThe 5th training sample is obtained after normalized
Element yi, as the output parameter of neutral net, build the neutral net；
Training error setting unit, for arranging training error ε；
Neural metwork training unit, for training the neutral net, when training deviation to be less than the training error, determines institute
State neural metwork training success；
Life forecast unit, for using former years wind regime data, gear-box to make according to the neural network prediction
Use the life-span.
8. it is according to claim 7 according to wind regime predict gearbox life system, it is characterised in that first normalizing
Change processing unit according to formula：
Obtain the first training sample element xi,1, the second training sample element xi,2, the 3rd training sample element xi,3, the 4th
Training sample element xi,4。
9. it is according to claim 7 according to wind regime predict gearbox life system, it is characterised in that second normalizing
Change processing unit according to formula：
Obtain the 5th training sample element yi；Wherein, Y is to calculate each described second pre- timing using nominal stress method
Between lose gear life y 'iCumulative sum.
10. it is according to claim 7 according to wind regime predict gearbox life system, it is characterised in that the system is entered
One step includes initial weight setting unit, and the initial weight for arranging the neutral net is the random number in [0,1] interval.
11. systems that gearbox life is predicted according to wind regime according to claim 7, it is characterised in that the nerve net
The hidden layer node quantity of network is 5 or 6.
12. systems that gearbox life is predicted according to wind regime according to claim 7, it is characterised in that the pre- timing
Between at intervals of 1 second, and/or, first scheduled time be 12 months, and/or, second scheduled time be one month.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310060297.6A CN104008423B (en) | 2013-02-26 | 2013-02-26 | Method and system for predicting life of gear case according to wind conditions |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310060297.6A CN104008423B (en) | 2013-02-26 | 2013-02-26 | Method and system for predicting life of gear case according to wind conditions |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104008423A CN104008423A (en) | 2014-08-27 |
| CN104008423B true CN104008423B (en) | 2017-05-03 |
Family
ID=51369071
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310060297.6A Active CN104008423B (en) | 2013-02-26 | 2013-02-26 | Method and system for predicting life of gear case according to wind conditions |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104008423B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105787561B (en) * | 2016-03-22 | 2019-04-30 | 新疆金风科技股份有限公司 | Recognition with Recurrent Neural Network model building method, gearbox fault detection method and device |
| JP6315836B2 (en) * | 2016-07-04 | 2018-04-25 | 株式会社日本製鋼所 | Windmill monitoring device, windmill monitoring method, and windmill monitoring program |
| CN106503794A (en) * | 2016-11-08 | 2017-03-15 | 上海电机学院 | A kind of gear case of blower method for predicting residual useful life |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102095573A (en) * | 2009-12-11 | 2011-06-15 | 上海卫星工程研究所 | State monitoring and diagnosis alarm method for mechanical component of satellite borne rotary equipment |
| CN102156043A (en) * | 2010-12-31 | 2011-08-17 | 北京四方继保自动化股份有限公司 | Online state monitoring and fault diagnosis system of wind generator set |
| CN102486833A (en) * | 2010-12-03 | 2012-06-06 | 财团法人工业技术研究院 | Method for predicting efficiency and detecting fault of device |
| CN202494565U (en) * | 2012-02-16 | 2012-10-17 | 永济电机天作电气有限责任公司 | Wind power gear box test data detection and acquisition system |
-
2013
- 2013-02-26 CN CN201310060297.6A patent/CN104008423B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102095573A (en) * | 2009-12-11 | 2011-06-15 | 上海卫星工程研究所 | State monitoring and diagnosis alarm method for mechanical component of satellite borne rotary equipment |
| CN102486833A (en) * | 2010-12-03 | 2012-06-06 | 财团法人工业技术研究院 | Method for predicting efficiency and detecting fault of device |
| CN102156043A (en) * | 2010-12-31 | 2011-08-17 | 北京四方继保自动化股份有限公司 | Online state monitoring and fault diagnosis system of wind generator set |
| CN202494565U (en) * | 2012-02-16 | 2012-10-17 | 永济电机天作电气有限责任公司 | Wind power gear box test data detection and acquisition system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104008423A (en) | 2014-08-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103259285B (en) | Method for optimizing short running of electric power system comprising large-scale wind power | |
| CN103138256B (en) | A kind of new energy electric power reduction panorama analytic system and method | |
| Wu et al. | A cloud decision framework in pure 2-tuple linguistic setting and its application for low-speed wind farm site selection | |
| CN102486833B (en) | Method for predicting efficiency and detecting fault of device | |
| CN104700321A (en) | Analytical method of state running tendency of transmission and distribution equipment | |
| CN107330183A (en) | A kind of wind power utilization computational methods based on service data | |
| CN103679282B (en) | The Forecasting Methodology of wind power climbing | |
| CN103514366A (en) | Urban air quality concentration monitoring missing data recovering method | |
| CN103106544B (en) | A kind of photovoltaic generation prognoses system based on T-S Fuzzy neutral net | |
| CN107730044A (en) | A kind of hybrid forecasting method of renewable energy power generation and load | |
| CN107153929A (en) | Gearbox of wind turbine fault monitoring method and system based on deep neural network | |
| CN107016235A (en) | The equipment running status health degree appraisal procedure adaptively merged based on multiple features | |
| CN101592538B (en) | Method for computing steady-state output power of wind power station based on actual measured data | |
| CN103987054A (en) | Wireless network sensor network coverage distributed method | |
| CN104156783A (en) | Maximum daily load prediction system and method of electric system considering meteorological accumulative effect | |
| CN105550943A (en) | Method for identifying abnormity of state parameters of wind turbine generator based on fuzzy comprehensive evaluation | |
| CN104063612A (en) | Tunnel engineering risk situation fuzzy evaluation method and system | |
| CN106199174A (en) | Extruder energy consumption predicting abnormality method based on transfer learning | |
| Arias | Assessment of present/future decommissioned wind blade fiber-reinforced composite material in the United States | |
| CN103310388A (en) | Method for calculating composite index of grid operation based on information source entropy | |
| CN109636066A (en) | A kind of wind power output power prediction technique based on fuzzy time series data mining | |
| CN104008423B (en) | Method and system for predicting life of gear case according to wind conditions | |
| CN104935017B (en) | Based on the wind-powered electricity generation and fired power generating unit combined method for improving light Robust Optimization Model | |
| CN103810532A (en) | Method for optimizing running state of urban drainage system | |
| CN105825002A (en) | Method for modeling dynamic equivalence of wind power farm based on dynamic grey-relevancy analysis method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210716 Address after: 102200 Sany Industrial Park, Beiqing Road, Changping District, Beijing Patentee after: Sany Heavy Energy Co.,Ltd. Address before: 410000 third floor, Sany administrative center, Sany industrial city, Sany Road, economic development zone, Changsha City, Hunan Province Patentee before: SANY GROUP Co.,Ltd. Effective date of registration: 20210716 Address after: 410000 third floor, Sany administrative center, Sany industrial city, Sany Road, economic development zone, Changsha City, Hunan Province Patentee after: SANY GROUP Co.,Ltd. Address before: 102206 Sany Industrial Park, 8 Beiqing Road, Huilongguan, Changping District, Beijing Patentee before: BEIJING SANY AUTOMATION TECHNOLOGY Co.,Ltd. |