CN107392304A - A kind of Wind turbines disorder data recognition method and device - Google Patents
A kind of Wind turbines disorder data recognition method and device Download PDFInfo
- Publication number
- CN107392304A CN107392304A CN201710658482.3A CN201710658482A CN107392304A CN 107392304 A CN107392304 A CN 107392304A CN 201710658482 A CN201710658482 A CN 201710658482A CN 107392304 A CN107392304 A CN 107392304A
- Authority
- CN
- China
- Prior art keywords
- data
- wind
- wind turbines
- neural network
- network model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000002159 abnormal effect Effects 0.000 claims abstract description 56
- 238000003062 neural network model Methods 0.000 claims abstract description 52
- 238000012360 testing method Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 38
- 230000006870 function Effects 0.000 claims description 23
- 238000012545 processing Methods 0.000 claims description 16
- 238000013480 data collection Methods 0.000 claims description 15
- 239000012141 concentrate Substances 0.000 claims description 13
- 230000008569 process Effects 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 6
- 238000005070 sampling Methods 0.000 claims description 5
- 238000004458 analytical method Methods 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000005520 cutting process Methods 0.000 claims description 3
- 210000005036 nerve Anatomy 0.000 claims description 2
- 238000001514 detection method Methods 0.000 abstract description 2
- 238000013528 artificial neural network Methods 0.000 description 15
- 230000007935 neutral effect Effects 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 210000002569 neuron Anatomy 0.000 description 6
- 230000005611 electricity Effects 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 230000004913 activation Effects 0.000 description 3
- 238000012886 linear function Methods 0.000 description 3
- 230000001172 regenerating effect Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010248 power generation Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000005094 computer simulation Methods 0.000 description 1
- 230000009514 concussion Effects 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000005290 field theory Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Wind Motors (AREA)
Abstract
The present invention relates to a kind of Wind turbines disorder data recognition method and device, including：Using Wind turbines air speed data as test sample, the BP neural network model built in advance is inputted；The target component exported according to the BP neural network model, judges Wind turbines abnormal data；The BP neural network model is built according to the air speed data of group of motors.Technical scheme provided by the invention, the abnormal data for obtaining Wind turbines can be accurately identified, solve thes problems, such as Wind turbines failure and disorder data recognition difficulty so that the detection of abnormal data has higher accuracy；So as to provide support to calculate Wind turbines and wind power plant theoretical power (horse-power).
Description
Technical field
The invention belongs to technical field of energy storage, more particularly to a kind of Wind turbines disorder data recognition method and device.
Background technology
In recent years, global renewable energy utilization annual growth reach the utilizations of 25%. regenerative resources will be with electric power row
Industry is leading, and the generating ratio of non-waterpower regenerative resource will be enlarged by twice.According to statistics, though the consumption of regenerative resource in 2002
About 1,400,000,000 t oil equivalents, the year two thousand thirty will be more than 2,200,000,000 t oil equivalents.Wind power generation is as technology is most ripe in addition to hydroelectric generation one
Kind renewable energy power generation, wind-driven generator are that wind energy is converted into mechanical work, the rotation of mechanical work rotor driven, and final output is handed over
Flow the power equipment of electricity.The operation principle of wind-driven generator is fairly simple, and wind wheel rotates under the action of the forces of the wind, and it keeps watch dynamic
The mechanical energy of rotor shaft, generator rotary electrification under the drive of rotor shaft can be changed into.
Wind turbines data are of great significance to wind-power electricity generation relation technological researching with wind statistics tool is abandoned.China is new
Energy development speed is fast, and wind-powered electricity generation installed capacity in 2006 to 2015 increases about 10 times so that the current wind-powered electricity generation station number in China
Amount has reached more than 1500 seats, and for Wind turbines more than 10,000, the current collection frequency for generator unit data is at least 15min,
The unit data of magnanimity are thereby produced, when data occur abnormal, when especially individual data occurs abnormal, are lacked effective
Identification and early warning means.
The content of the invention
In order to make up above-mentioned technological gap, the present invention provides a kind of Wind turbines disorder data recognition method and device, knot
Close BP neural network effectively to identify the abnormal data of Wind turbines, Wind turbines and wind-powered electricity generation field theory work(are calculated to be accurate
Rate provides support.
The purpose of the present invention is realized using following technical proposals：
A kind of Wind turbines disorder data recognition method, methods described include：
Using Wind turbines air speed data as test sample, the BP neural network model built in advance is inputted；
The target component exported according to the BP neural network model, judges Wind turbines abnormal data；
The BP neural network model is built according to the air speed data of group of motors.
Preferably, the BP neural network model is built according to the air speed data of group of motors, including：
Running of wind generating set status data is divided according to velocity wind levels, obtains analyze data collection；
The input parameter for selecting the analyze data to concentrate, and the input parameter in the air speed data is normalized
Processing, build training sample；
The target parameter of Wind turbines abnormal data is defined, and it is normalized；
Judgement input parameter using the input parameter as the abnormal data of Wind turbines, using the target parameter as target
Data establish BP neural network model, and the BP neural network model is trained using the training sample, and acquisition is sentenced
The BP neural network model of disconnected Wind turbines abnormal data.
Further, the analyze data collection is determined according to running of wind generating set status data, including：If wind turbine
Group traffic coverage is between incision wind speed and rated wind speed, passes through { SI}={ S | dIDetermine analyze data collection；
If running of wind generating set section is between rated wind speed and excision wind speed, pass through { SC}={ S | dCDetermine analysis
Data set；
In formula, { S } represents default Wind turbines actual operating data sample set, dIt is specified≤dI＜ dExcision；dIRepresent Wind turbines
Traffic coverage is in incision wind speed dIncisionWith rated wind speed dIt is specifiedBetween velocity wind levels, dCRepresent that running of wind generating set section is in
Rated wind speed dIt is specifiedWith cutting off wind speed dExcisionBetween velocity wind levels；dIAnd dCMeet dIncision＜ dI＜ dIt is specified。
Further, the input parameter of analyze data concentration is determined by following formula：[xi,t-3,xi,t-2,xi,t-1,xi,t]T
In formula, xi,t-3,xi,t-2,xi,t-1,xi,tRespectively in sampling t input parameter xiClose on data, xi(xi∈
Rd), i=1 ..., n；R is the air speed data of analyze data collection, and i numbers for blower fan, and n is Wind turbines number of units, and d is data value
Traffic coverage.
Further, the input parameter in air speed data is normalized, and builds training sample, such as following formula
It is shown：
S1'={ (si,t-3)}；si,t-3∈{S1}
S2'={ (si,t-2)}；si,t-2∈{S2}
S3'={ (si,t-1)}；si,t-1∈{S3}
S4'={ (si,t)}；si,t∈{S4}
Wherein, S1,S2,S3,S4Respectively described grouped data concentrates the data sample for lacking several, doomed dead, wrong numbers and tuple
Collection；S1',S2',S3',S4' it is respectively that the grouped data concentrates input parameter xi,t-3,xi,t-2,xi,t-1,xi,tAbnormal data
In lack several, doomed dead, wrong numbers and tuple normalization after data set；si,t-3,si,t-2,si,t-1,si,tRespectively described classification number
According to the data after the normalization of the scarce number, doomed dead, wrong number and tuple of the air speed data for concentrating input parameter.
Further, the abnormal data for defining Wind turbines is target component, and it is normalized, and is wrapped
Include：
The target component for defining scarce several, doomed dead, wrong numbers and tuple in the abnormal data of Wind turbines is respectively 1,2,3 and 4；
Place is normalized to lack the target component of several, doomed dead, wrong numbers and tuple in the abnormal data of 4 pairs of Wind turbines
Reason, the target parameter normalized value for obtaining the abnormal data of Wind turbines are 1.
Preferably, the target component exported according to the BP neural network model, the abnormal number of Wind turbines is judged
According to, including：
As the target component S of BP neural network model outputI, tFor 0 when, represent running of wind generating set it is normal, work as institute
State the target component S of BP neural network model outputI, tFor 1 when, then it represents that abnormal number in the process of running be present in Wind turbines
According to.
Preferably, the hidden layer of the BP neural network model chooses n neuron, input layer and hidden layer transmission function
Tansig functions are chosen, logsig functions are chosen between hidden layer and output layer, performance function chooses mse, model training method
Using trainlm, weights and threshold learning method selection learngdm.
A kind of Wind turbines disorder data recognition device, described device include：
Input module, for using Wind turbines air speed data as test sample, inputting the BP neural network mould built in advance
Type；
Identification module, for the target component exported according to the BP neural network model, judge Wind turbines exception number
According to；The BP neural network model is built according to the air speed data of group of motors.
Further, the input module includes：
Acquiring unit, for being divided according to velocity wind levels to running of wind generating set status data, obtain analyze data
Collection；
First processing units, for the input parameter for selecting the analyze data to concentrate, and in the air speed data
Input parameter is normalized, and builds training sample；
Second processing unit, it is normalized for defining the target parameter of Wind turbines abnormal data, and to it；
Training unit, for the judgement input parameter using the input parameter as the abnormal data of Wind turbines, with described
Target parameter is that target data establishes BP neural network model, and the BP neural network model is entered using the training sample
Row training, obtains the BP neural network model for judging Wind turbines abnormal data.
Compared with immediate prior art, beneficial effects of the present invention are：
The present invention program proposes a kind of Wind turbines disorder data recognition method and device, using Wind turbines air speed data as
Test specimens
This, inputs the BP neural network model built in advance；The target component exported according to the BP neural network model,
Sentence
The abnormal data of disconnected Wind turbines, it can effectively solve the problem that Wind turbines failure and disorder data recognition be difficult asks
Topic, makes
The detection for obtaining abnormal data has higher accuracy；There is provided for abnormal data recovery, dealing of abnormal data important
Information；
So as to provide support to calculate Wind turbines and wind power plant theoretical power (horse-power).
Brief description of the drawings
Fig. 1 is a kind of Wind turbines disorder data recognition schematic device provided in the embodiment of the present invention；
Fig. 2 is the BP neural network structural representation provided in the embodiment of the present invention；
Fig. 3 is the complete training flow chart of the BP neural network provided in the embodiment of the present invention；
Fig. 4 is the Wind turbines disorder data recognition model schematic provided in the embodiment of the present invention；
Embodiment：
The embodiment of the present invention is elaborated below in conjunction with the accompanying drawings.
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
Part of the embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The all other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
In order to make up when Wind turbines air speed data occurs abnormal, when especially individual data occurs abnormal, shortage has
The identification of effect and early warning means.The present invention provides a kind of Wind turbines disorder data recognition method and device, with reference to BP nerve nets
Network is effectively identified to the abnormal data of Wind turbines, and branch is provided for accurate calculating Wind turbines and wind power plant theoretical power (horse-power)
Support.
Wherein, the general principle of involved BP neural network and algorithm include：
BP neural network (Back Propagation Neural Network) refers to based on error backpropagation algorithm
Multilayer feedforward neural network, using the training method for having tutor.It is D.E.Rumelhart and J.L.McCelland and its ground
Study carefully what group studied and designed in 1986.BP neural network has following features：(1) it can be approached and appointed with arbitrary accuracy
What Nonlinear Mapping, is realized to complex system modeling；(2) can learn with adaptive unknown message, if system is become
Change can change prediction effect by changing the connection value of network；(3) distributed information storage and processing structure, have certain
Fault-tolerance, therefore the system constructed has preferable robustness；(4) model structure of multi input, multi output, it is adapted to place
Manage challenge.
BP networks also have one or more layers implicit node in addition to Inport And Outport Node, with not having any connection in node layer.
Input signal is transmitted through each implicit node from input layer successively, is then passed to output node layer, the output shadow per node layer
Ring the output of next node layer.BP networks total algorithm is ripe, and its information processing capability comes to simple non-linear functions
It is repeatedly compound.BP neural network general structure is as shown in Figure 2.
The mathematical modeling of BP algorithm is to solve for the Optimal solution problem such as minor function：
Wherein, x is training sample, yk(t) it is the reality output of network, yk(t) it is the desired output of network, wijFor input
Node layer i is to hidden layer node j weights, vjkWeights for hidden layer node j to output node layer k, θjFor hidden layer node j
The threshold value at place, γtFor the threshold value at output node t, f (x) is activation primitive.Realize global error function E on curved surface by
Gradient declines, and using gladient rule, seeks negative gradients of the E to the connection weight and threshold value of output layer and hidden layer：
Decline principle by gradient, i.e. the change of connection weight and threshold value is proportional to negative gradient, therefore has：
Wherein, η is learning rate, and 0 ＜ η ＜ 1, i=1,2 ..., m, t=1,2 ..., n, j=1,2 ..., p, bjFor centre
The output of each neuron of layer, sjFor the intermediate result of neural n ary operation.
Network connection weights and threshold value after adjustment are as follows：
Wherein, l represents frequency of training.
The complete training process of BP neural network is as shown in Figure 3：
As long as the number of hidden layer neuron is fully more, then hidden layer neuron activation function is three layers of god of linear function
Any function can be approached through network.BP neural network can be answered by the Compound Mappings of simple nonlinear processing unit
Miscellaneous Nonlinear Processing ability.
The training process of neutral net is actually learning process of the network to training sample inherent law, and network is entered
The purpose of row training is primarily to allow network to have correct mapping ability to the data beyond training sample.Neutral net
Generalization ability refers to the popularization energy of adaptability of the neutral net to the new samples beyond training sample, also referred to as neutral net
Power, it is considered to be the important indicator of neutral net performance is weighed, the neutral net with generalization ability can be applied in practice,
Otherwise do not possess application value.
The generalization ability of neutral net is influenceed by following factor：
1) characteristic of sample
Only when training sample is enough the principal character for characterizing the problem of studied, network passes through rational study mechanism
Generalization ability can be made it have, rational sampling structure is the necessary condition that neutral net has generalization ability.
2) factor of network itself
The learning algorithm of the structure of such as network, initial value and network.The structure of network mainly including network hidden layer number,
The number of hidden node and the activation primitive of hidden node.
When hidden node function bounded, three-layer forward networks, which have to approach with arbitrary accuracy, is defined on appointing on compact subset
The ability of meaning nonlinear function.This explanation uses three layers of BP neural network, and hidden node function is Sigmoid functions, output section
Point function uses linear function, fully achieves the requirement that network approaches." over-fitting " phenomenon is that network hidden node is excessive
Inevitable outcome, influence the generalization ability of network, while think under the requirement for meeting precision, the exponent number of approximating function it is more few more
Good, low order, which is approached, can effectively prevent " over-fitting ", so as to improve the predictive ability of network.
The selection of neutral net initial value also influences the generalization ability of network.General random gives one group of weights, then adopts
With certain learning rules, progressively adjusted in training, finally give one group of preferable weights distribution.Because BP algorithm is to be based on
Gradient descent method, different initial weights may result in different results.If value is improper, concussion may be caused not receive
Hold back, even if convergence also results in training time growth, or be absorbed in Local Extremum, cannot get suitable weights distribution, influence net
The generalization ability of network.
It the described method comprises the following steps：
S1 inputs the BP neural network model built in advance using Wind turbines air speed data as test sample；Such as Fig. 4 institutes
Show.
The target component that S2 exports according to the BP neural network model, judges Wind turbines abnormal data；
The BP neural network model is built according to the air speed data of group of motors, including：
Running of wind generating set status data is divided according to velocity wind levels, obtains analyze data collection；
The input parameter for selecting the analyze data to concentrate, and the input parameter in the air speed data is normalized
Processing, build training sample；
The target parameter of Wind turbines abnormal data is defined, and it is normalized；
To input judgement input parameter of the parameter as the abnormal data of Wind turbines, established by target data of target parameter
BP neural network model, and the BP neural network model is trained using the training sample, acquisition judges wind turbine
The BP neural network model of group abnormal data.
Analyze data collection is determined according to running of wind generating set section, including：If running of wind generating set section is in and cut
Enter between wind speed and rated wind speed, pass through { SI}={ S | dIDetermine analyze data collection；
If running of wind generating set section is between rated wind speed and excision wind speed, pass through { SC}={ S | dCDetermine analysis
Data set；
In formula, { S } represents default Wind turbines actual operating data sample set, dIt is specified≤dI＜ dExcision；dIRepresent Wind turbines
Traffic coverage is in incision wind speed dIncisionWith rated wind speed dIt is specifiedBetween velocity wind levels, dCRepresent that running of wind generating set section is in
Rated wind speed dIt is specifiedWith cutting off wind speed dExcisionBetween velocity wind levels；dIAnd dCMeet dIncision＜ dI＜ dIt is specified。
The input parameter of analyze data concentration is determined by following formula：[xi,t-3,xi,t-2,xi,t-1,xi,t]T
In formula, xi,t-3,xi,t-2,xi,t-1,xi,tRespectively in sampling t input parameter xiClose on data, xi(xi∈
Rd), i=1 ..., n；R is the air speed data of analyze data collection, and i numbers for blower fan, and n is Wind turbines number of units, and d is data value
Traffic coverage.
Input parameter in air speed data is normalized, training sample is built, is shown below：
S1'={ (si,t-3)}；si,t-3∈{S1}
S2'={ (si,t-2)}；si,t-2∈{S2}
S3'={ (si,t-1)}；si,t-1∈{S3}
S4'={ (si,t)}；si,t∈{S4}
Wherein, S1,S2,S3,S4Respectively described grouped data concentrates the data sample for lacking several, doomed dead, wrong numbers and tuple
Collection；S1',S2',S3',S4' it is respectively that the grouped data concentrates input parameter xi,t-3,xi,t-2,xi,t-1,xi,tAbnormal data
In lack the data set after the normalization of several, doomed dead, wrong numbers and tuple, i.e. training sample；si,t-3,si,t-2,si,t-1,si,tRespectively
The grouped data concentrates the data after the normalization of the scarce number, doomed dead, wrong number and tuple of the air speed data of input parameter.Its
In, scarce number refers to that, in given sampling instant, data actual acquisition result is empty situation；
Doomed dead to refer under continuous non-zero m/s gustiness, data parameter is constant to continue to exceed 4 sampled points, or 0m/s wind speed
State, the continuous constant situation more than 96 sampled points of data parameter；
Wrong number refers to that wind speed collection result is less than 0m/s or the situation more than 50m/s；
Tuple refers to that air speed data sampled result is distributed the abnormal feelings that unnatural, neighbouring sample point wind speed rate of change is fixed
Condition.
Such as wrong number, doomed dead for conventional abnormal conditions existing for input data, method can be calculated normally, but lack number, such as
Nan, the iterative calculation for making BP neural network is made a mistake, thus scarce number is subjected to generation not produce the about definite value of normal interference
Replace, such as replaced with 99.
The abnormal data for defining Wind turbines is target component, and it is normalized, including：
The target component for defining scarce several, doomed dead, wrong numbers and tuple in the abnormal data of Wind turbines is respectively 1,2,3 and 4；
Place is normalized to lack the target component of several, doomed dead, wrong numbers and tuple in the abnormal data of 4 pairs of Wind turbines
Reason, the target parameter normalized value for obtaining the abnormal data of Wind turbines are 1.
In S2, according to the target component of BP neural network model output, judge the abnormal data of Wind turbines, wrap
Include：
As the target component S of BP neural network model outputI, tFor 0 when, represent running of wind generating set it is normal, when BP nerve
The target component S of network model outputI, tFor 1 when, then it represents that abnormal data in the process of running be present in Wind turbines.
Because data volume is huge, abnormal data state is judged for complicated Nonlinear Mapping, hidden layer neuron number
It can suitably increase, multilayer hidden layer can be used if necessary.The hidden layer of BP neural network model chooses n neuron, input layer
Tansig functions are chosen with hidden layer transmission function, logsig functions are chosen between hidden layer and output layer, performance function is chosen
Mse, model training method use trainlm, weights and threshold learning method selection learngdm.
The Wind turbines disorder data recognition method proposed using this patent, the sample data for training pattern should foot
Enough abundant, data volume suggestion is more than 20000 groups.The real-time knowledge of more similar abnormal datas of Wind turbines in wind power plant can be achieved
Not, after tested, identification accuracy reaches 99.99%.
Based on same inventive concept, the present invention also provides a kind of Wind turbines disorder data recognition device, as shown in figure 1,
Including：
Input module, for using Wind turbines air speed data as test sample, inputting the BP neural network mould built in advance
Type；
Identification module, for the target component exported according to the BP neural network model, judge Wind turbines exception number
According to；BP neural network model is built according to the air speed data of group of motors.
Wherein, input module includes：
Acquiring unit, for being divided according to velocity wind levels to running of wind generating set status data, obtain analyze data
Collection；
First processing units, for the input parameter for selecting the analyze data to concentrate, and in the air speed data
Input parameter is normalized, and builds training sample；
Second processing unit, it is normalized for defining the target parameter of Wind turbines abnormal data, and to it；
Training unit, for the judgement input parameter using the input parameter as the abnormal data of Wind turbines, with described
Target parameter is that target data establishes BP neural network model, and the BP neural network model is entered using the training sample
Row training, obtains the BP neural network model for judging Wind turbines abnormal data.
It should be understood by those skilled in the art that, embodiments herein can be provided as method, system or computer program
Product.Therefore, the application can use the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware
Apply the form of example.Moreover, the application can use the computer for wherein including computer usable program code in one or more
The computer program production that usable storage medium is implemented on (including but is not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of product.
The application is with reference to the flow according to the method for the embodiment of the present application, equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that can be by every first-class in computer program instructions implementation process figure and/or block diagram
Journey and/or the flow in square frame and flow chart and/or block diagram and/or the combination of square frame.These computer programs can be provided
The processors of all-purpose computer, special-purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce
A raw machine so that produced by the instruction of computer or the computing device of other programmable data processing devices for real
The device for the function of being specified in present one flow of flow chart or one square frame of multiple flows and/or block diagram or multiple square frames.
These computer program instructions, which may be alternatively stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that the instruction being stored in the computer-readable memory, which produces, to be included referring to
Make the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one square frame of block diagram or
The function of being specified in multiple square frames.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that counted
Series of operation steps is performed on calculation machine or other programmable devices to produce computer implemented processing, so as in computer or
The instruction performed on other programmable devices is provided for realizing in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in individual square frame or multiple square frames.
Finally it should be noted that：Above example is only illustrating the technical scheme of the application rather than to its protection domain
Limitation, although the application is described in detail with reference to above-described embodiment, those of ordinary skill in the art should
Understand：Those skilled in the art read the embodiment of application can be still carried out after the application a variety of changes, modification or
Person's equivalent substitution, these changes, modification or equivalent substitution, it applies within pending right at it.
Claims (10)
- A kind of 1. Wind turbines disorder data recognition method, it is characterised in that methods described includes：Using Wind turbines air speed data as test sample, the BP neural network model built in advance is inputted；The target component exported according to the BP neural network model, judge the abnormal data of Wind turbines；The BP neural network model is built according to the air speed data of group of motors.
- 2. the method as described in claim 1, it is characterised in that the BP neural network model is according to the air speed data of group of motors Built, including：Running of wind generating set status data is divided according to velocity wind levels, obtains analyze data collection；The input parameter for selecting the analyze data to concentrate, and place is normalized to the input parameter in the air speed data Reason, build training sample；The target parameter of Wind turbines abnormal data is defined, and it is normalized；Judgement input parameter using the input parameter as the abnormal data of Wind turbines, using the target parameter as target data BP neural network model is established, and the BP neural network model is trained using the training sample, acquisition judges wind The BP neural network model of group of motors abnormal data.
- 3. method as claimed in claim 2, it is characterised in that the analyze data collection is according to running of wind generating set status data It is determined, including：If running of wind generating set section is between incision wind speed and rated wind speed, pass through { SI}={ S | dIReally Setting analysis data set；If running of wind generating set section is between rated wind speed and excision wind speed, pass through { SC}={ S | dCDetermine analyze data Collection；In formula, { S } represents default Wind turbines actual operating data sample set, dIt is specified≤dI＜ dExcision；dIRepresent running of wind generating set Section is in incision wind speed dIncisionWith rated wind speed dIt is specifiedBetween velocity wind levels, dCRepresent that running of wind generating set section is in specified Wind speed dIt is specifiedWith cutting off wind speed dExcisionBetween velocity wind levels；dIAnd dCMeet dIncision＜ dI＜ dIt is specified。
- 4. method as claimed in claim 2, it is characterised in that the input parameter of analyze data concentration is determined by following formula： [xi,t-3,xi,t-2,xi,t-1,xi,t]TIn formula, xi,t-3,xi,t-2,xi,t-1,xi,tRespectively in sampling t input parameter xiClose on data, xi(xi∈Rd), i =1 ..., n；R is the air speed data of analyze data collection, and i numbers for blower fan, and n is Wind turbines number of units, and d is the fortune of data value Row section.
- 5. method as claimed in claim 4, it is characterised in that place is normalized in the input parameter in air speed data Reason, training sample is built, is shown below：S1'={ (si,t-3)}；si,t-3∈{S1}S2'={ (si,t-2)}；si,t-2∈{S2}S3'={ (si,t-1)}；si,t-1∈{S3}S4'={ (si,t)}；si,t∈{S4}Wherein, S1,S2,S3,S4Respectively described grouped data concentrates the set of data samples for lacking several, doomed dead, wrong numbers and tuple；S1', S2',S3',S4' it is respectively that the grouped data concentrates input parameter xi,t-3,xi,t-2,xi,t-1,xi,tAbnormal data in lack number, Data set after the normalization of doomed dead, wrong number and tuple；si,t-3,si,t-2,si,t-1,si,tRespectively described grouped data is concentrated defeated Enter the data after the normalization of the scarce number, doomed dead, wrong number and tuple of the air speed data of parameter.
- 6. method as claimed in claim 2, it is characterised in that the abnormal data for defining Wind turbines is target component, And it is normalized, including：The target component for defining scarce several, doomed dead, wrong numbers and tuple in the abnormal data of Wind turbines is respectively 1,2,3 and 4；It is normalized, is obtained with lacking the target component of several, doomed dead, wrong numbers and tuple in the abnormal data of 4 pairs of Wind turbines The target component normalized value for taking the abnormal data of Wind turbines is 1.
- 7. the method as described in claim 1, it is characterised in that the target exported according to the BP neural network model is joined Number, judge the abnormal data of Wind turbines, including：As the target component S of BP neural network model outputI, tFor 0 when, represent running of wind generating set it is normal, as the BP The target component S of neural network model outputI, tFor 1 when, then it represents that abnormal data in the process of running be present in Wind turbines.
- 8. the method as described in claim 1, it is characterised in that the hidden layer of the BP neural network model chooses n nerve Member, input layer choose tansig functions with hidden layer transmission function, and logsig functions, performance are chosen between hidden layer and output layer Function chooses mse, and model training method uses trainlm, weights and threshold learning method selection learngdm.
- 9. a kind of Wind turbines disorder data recognition device, it is characterised in that described device includes：Input module, for using Wind turbines air speed data as test sample, inputting the BP neural network model built in advance；Identification module, for the target component exported according to the BP neural network model, judge Wind turbines unusual condition；The BP neural network model is built according to the air speed data of group of motors.
- 10. device as claimed in claim 9, it is characterised in that the input module includes：Acquiring unit, for being divided according to velocity wind levels to running of wind generating set status data, obtain analyze data collection；First processing units, for the input parameter for selecting the analyze data to concentrate, and to the input in the air speed data Parameter is normalized, and builds training sample；Second processing unit, it is normalized for defining the target parameter of Wind turbines abnormal data, and to it；Training unit, for the judgement input parameter using the input parameter as the abnormal data of Wind turbines, with the target Parameter is that target data establishes BP neural network model, and the BP neural network model is instructed using the training sample Practice, obtain the BP neural network model for judging Wind turbines abnormal data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710658482.3A CN107392304A (en) | 2017-08-04 | 2017-08-04 | A kind of Wind turbines disorder data recognition method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710658482.3A CN107392304A (en) | 2017-08-04 | 2017-08-04 | A kind of Wind turbines disorder data recognition method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107392304A true CN107392304A (en) | 2017-11-24 |
Family
ID=60344380
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710658482.3A Pending CN107392304A (en) | 2017-08-04 | 2017-08-04 | A kind of Wind turbines disorder data recognition method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107392304A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107977710A (en) * | 2017-12-21 | 2018-05-01 | 南方电网科学研究院有限责任公司 | Multiplexing electric abnormality data detection method and device |
| CN108629095A (en) * | 2018-04-23 | 2018-10-09 | 明阳智慧能源集团股份公司 | A kind of modeling method of gearbox of wind turbine bearing temperature |
| CN109783881A (en) * | 2018-12-21 | 2019-05-21 | 中国电力科学研究院有限公司 | A kind of Wind turbines generated output determines method and device |
| CN110489852A (en) * | 2019-08-14 | 2019-11-22 | 北京天泽智云科技有限公司 | Improve the method and device of the wind power system quality of data |
| CN110634081A (en) * | 2019-08-02 | 2019-12-31 | 国网四川省电力公司映秀湾水力发电总厂 | Method and device for processing abnormal data of hydropower station |
| CN110674864A (en) * | 2019-09-20 | 2020-01-10 | 国网上海市电力公司 | Wind power abnormal data identification method with synchronous phasor measurement device |
| WO2020119118A1 (en) * | 2018-12-13 | 2020-06-18 | 平安医疗健康管理股份有限公司 | Abnormal data processing method, apparatus and device, and computer readable storage medium |
| CN112994101A (en) * | 2021-03-11 | 2021-06-18 | 中国长江三峡集团有限公司 | Neural network-based wind power plant generated power post-evaluation and monitoring method |
| CN113196303A (en) * | 2018-12-11 | 2021-07-30 | 亚马逊技术股份有限公司 | Inappropriate neural network input detection and processing |
| CN113884705A (en) * | 2021-09-28 | 2022-01-04 | 上海电气风电集团股份有限公司 | Monitoring method and system of cluster fan anemometer and computer readable storage medium |
| CN114358012A (en) * | 2021-11-23 | 2022-04-15 | 华能大理风力发电有限公司洱源分公司 | Equipment abnormal semantic recognition method, device, equipment and medium |
| CN115146718A (en) * | 2022-06-27 | 2022-10-04 | 北京华能新锐控制技术有限公司 | Depth representation-based wind turbine generator anomaly detection method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069476A (en) * | 2015-08-10 | 2015-11-18 | 国网宁夏电力公司 | Method for identifying abnormal wind power data based on two-stage integration learning |
| CN105550943A (en) * | 2016-01-18 | 2016-05-04 | 重庆大学 | Method for identifying abnormity of state parameters of wind turbine generator based on fuzzy comprehensive evaluation |
| CN105719002A (en) * | 2016-01-18 | 2016-06-29 | 重庆大学 | Wind turbine generator state parameter abnormity identification method based on combination prediction |
| CN106124982A (en) * | 2016-06-14 | 2016-11-16 | 都城绿色能源有限公司 | Automatic expert's resultant fault diagnostic system of a kind of Wind turbines and diagnostic method |
| US20170091615A1 (en) * | 2015-09-28 | 2017-03-30 | Siemens Aktiengesellschaft | System and method for predicting power plant operational parameters utilizing artificial neural network deep learning methodologies |
| CN106570790A (en) * | 2016-11-10 | 2017-04-19 | 甘肃省电力公司风电技术中心 | Wind farm output power data restoration method considering segmental characteristics of wind speed data |
-
2017
- 2017-08-04 CN CN201710658482.3A patent/CN107392304A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105069476A (en) * | 2015-08-10 | 2015-11-18 | 国网宁夏电力公司 | Method for identifying abnormal wind power data based on two-stage integration learning |
| US20170091615A1 (en) * | 2015-09-28 | 2017-03-30 | Siemens Aktiengesellschaft | System and method for predicting power plant operational parameters utilizing artificial neural network deep learning methodologies |
| CN105550943A (en) * | 2016-01-18 | 2016-05-04 | 重庆大学 | Method for identifying abnormity of state parameters of wind turbine generator based on fuzzy comprehensive evaluation |
| CN105719002A (en) * | 2016-01-18 | 2016-06-29 | 重庆大学 | Wind turbine generator state parameter abnormity identification method based on combination prediction |
| CN106124982A (en) * | 2016-06-14 | 2016-11-16 | 都城绿色能源有限公司 | Automatic expert's resultant fault diagnostic system of a kind of Wind turbines and diagnostic method |
| CN106570790A (en) * | 2016-11-10 | 2017-04-19 | 甘肃省电力公司风电技术中心 | Wind farm output power data restoration method considering segmental characteristics of wind speed data |
Non-Patent Citations (2)
| Title |
|---|
| P. BANGALORE,ET AL.: "An artificial neural network-based condition monitoring method for wind turbines, with application to the monitoring of the gearbox", 《WIND ENERGY》 * |
| 张慧亭等: "大数据分析技术在风电设备异常预测中的应用", 《电脑知识与技术》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107977710A (en) * | 2017-12-21 | 2018-05-01 | 南方电网科学研究院有限责任公司 | Multiplexing electric abnormality data detection method and device |
| CN107977710B (en) * | 2017-12-21 | 2020-12-25 | 南方电网科学研究院有限责任公司 | Electricity consumption abnormal data detection method and device |
| CN108629095A (en) * | 2018-04-23 | 2018-10-09 | 明阳智慧能源集团股份公司 | A kind of modeling method of gearbox of wind turbine bearing temperature |
| CN113196303A (en) * | 2018-12-11 | 2021-07-30 | 亚马逊技术股份有限公司 | Inappropriate neural network input detection and processing |
| WO2020119118A1 (en) * | 2018-12-13 | 2020-06-18 | 平安医疗健康管理股份有限公司 | Abnormal data processing method, apparatus and device, and computer readable storage medium |
| CN109783881A (en) * | 2018-12-21 | 2019-05-21 | 中国电力科学研究院有限公司 | A kind of Wind turbines generated output determines method and device |
| CN110634081A (en) * | 2019-08-02 | 2019-12-31 | 国网四川省电力公司映秀湾水力发电总厂 | Method and device for processing abnormal data of hydropower station |
| CN110489852A (en) * | 2019-08-14 | 2019-11-22 | 北京天泽智云科技有限公司 | Improve the method and device of the wind power system quality of data |
| CN110674864A (en) * | 2019-09-20 | 2020-01-10 | 国网上海市电力公司 | Wind power abnormal data identification method with synchronous phasor measurement device |
| CN110674864B (en) * | 2019-09-20 | 2024-03-15 | 国网上海市电力公司 | Wind power abnormal data identification method comprising synchronous phasor measurement device |
| CN112994101A (en) * | 2021-03-11 | 2021-06-18 | 中国长江三峡集团有限公司 | Neural network-based wind power plant generated power post-evaluation and monitoring method |
| CN113884705A (en) * | 2021-09-28 | 2022-01-04 | 上海电气风电集团股份有限公司 | Monitoring method and system of cluster fan anemometer and computer readable storage medium |
| CN114358012A (en) * | 2021-11-23 | 2022-04-15 | 华能大理风力发电有限公司洱源分公司 | Equipment abnormal semantic recognition method, device, equipment and medium |
| CN115146718A (en) * | 2022-06-27 | 2022-10-04 | 北京华能新锐控制技术有限公司 | Depth representation-based wind turbine generator anomaly detection method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107392304A (en) | A kind of Wind turbines disorder data recognition method and device | |
| CN110929918B (en) | 10kV feeder fault prediction method based on CNN and LightGBM | |
| CN108363382B (en) | A kind of complex equipment fault diagnosis method and system | |
| CN108832619A (en) | Transient stability evaluation in power system method based on convolutional neural networks | |
| CN108197648A (en) | A kind of Fault Diagnosis Method of Hydro-generating Unit and system based on LSTM deep learning models | |
| CN111680875B (en) | Unmanned aerial vehicle state risk fuzzy comprehensive evaluation method based on probability baseline model | |
| CN104777418B (en) | A kind of analog-circuit fault diagnosis method based on depth Boltzmann machine | |
| CN107944648A (en) | A kind of accurate Forecasting Methodology of large ship speed of a ship or plane rate of fuel consumption | |
| CN108053128A (en) | A kind of Power Network Transient Stability fast evaluation method based on ELM and TF | |
| CN109902339A (en) | A kind of Fault Diagnosis of Roller Bearings based on IAGA-SVM | |
| CN104408562A (en) | Photovoltaic system generating efficiency comprehensive evaluation method based on BP (back propagation) neural network | |
| CN105590144A (en) | Wind speed prediction method and apparatus based on NARX neural network | |
| CN113659565B (en) | Online prediction method for frequency situation of new energy power system | |
| CN109029974A (en) | A kind of epicyclic gearbox fault detection method based on one-dimensional convolutional neural networks | |
| CN111190349A (en) | Method, system and medium for monitoring state and diagnosing fault of ship engine room equipment | |
| CN106897945A (en) | The clustering method and equipment of wind power generating set | |
| CN104218571B (en) | A kind of running status appraisal procedure of wind power plant | |
| CN112069723A (en) | Method and system for evaluating transient stability of power system | |
| CN110081966A (en) | The fault diagnosis system and method for fan vibration sensor | |
| CN113612237A (en) | Method for positioning resonance-induced subsynchronous oscillation source in offshore wind farm | |
| CN110991689B (en) | Distributed photovoltaic power generation system short-term prediction method based on LSTM-Morlet model | |
| CN107527093A (en) | A kind of running of wind generating set method for diagnosing status and device | |
| CN112836876A (en) | Power distribution network line load prediction method based on deep learning | |
| CN112163474A (en) | Intelligent gearbox diagnosis method based on model fusion | |
| CN116910513A (en) | Centrifugal pump rotor fault diagnosis method based on CWGAN-GP and double-flow CNN model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171124 |
|
| RJ01 | Rejection of invention patent application after publication |