CN105677791A - Method and system used for analyzing operating data of wind generating set - Google Patents

Method and system used for analyzing operating data of wind generating set Download PDF

Info

Publication number
CN105677791A
CN105677791A CN201511029436.4A CN201511029436A CN105677791A CN 105677791 A CN105677791 A CN 105677791A CN 201511029436 A CN201511029436 A CN 201511029436A CN 105677791 A CN105677791 A CN 105677791A
Authority
CN
China
Prior art keywords
data
described
generating set
wind power
power generating
Prior art date
Application number
CN201511029436.4A
Other languages
Chinese (zh)
Other versions
CN105677791B (en
Inventor
赵云鹏
Original Assignee
新疆金风科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 新疆金风科技股份有限公司 filed Critical 新疆金风科技股份有限公司
Priority to CN201511029436.4A priority Critical patent/CN105677791B/en
Publication of CN105677791A publication Critical patent/CN105677791A/en
Application granted granted Critical
Publication of CN105677791B publication Critical patent/CN105677791B/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/245Query processing
    • G06F16/2458Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2465Query processing support for facilitating data mining operations in structured databases
    • GPHYSICS
    • G06COMPUTING; CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/28Databases characterised by their database models, e.g. relational or object models
    • G06F16/284Relational databases

Abstract

The invention provides a method and system used for analyzing operating data of a wind generating set. The method comprises that operating data, state data and set fault data, of the wind generating set, acquired in a predetermined statistical time period serve as initial data; the data is analyzed, the attribute related to the operating state change of the wind generating set is extracted, samples of the initial data are taken, sample data of a plurality of operating state types is obtained, statistics and analysis are carried out to the sample data of the plurality of operating state types in dependence on the extracted attribute, a plurality of operation characteristic quantities, of the wind generating set, which accord with a predetermined condition of the plurality of operating state types are obtained, each operation characteristic quantity comprises the value of a single attribute or the values of a plurality of attributes, the variation diagram of the operation characteristic quantity in a specified state time period is drawn in dependence on the obtained plurality of operation characteristic quantities, in order to represent the change of the operation characteristic quantity of the wind generating set in a specified state.

Description

For analyzing method and the system of the service data of wind power generating set

Technical field

The present invention relates to the information processing technology of wind-force generating, particularly relate to method and the system of a kind of service data for analyzing wind power generating set.

Background technology

Computing ability in recent years improve constantly and the fast development of distributed computing technology makes it possible to process the data of magnanimity, variform, i.e. big data based on distributed framework.

Big data technique based on distributed framework relates to data mining, process, storage and theme analysis. Wherein, data mining refers to the process being hidden in wherein information from a large amount of data by algorithm search, obtains target information by all multi-methods such as statistics, Data Environments, information retrieval, machine learning, expert systems (relying on thumb rule in the past) and pattern recognitions.

In technical field of wind power generation, in process wind power generating set wherein monitored in the process that wind power generating set is run and in wind energy turbine set, produce a large amount of service datas and state data, fault data etc. Based on big data technique, these data being excavated and analyzed, condition monitoring and policy control optimization for wind power generating set are significant.

Summary of the invention

It is an object of the present invention to provide the technical scheme of a kind of service data for analyzing wind power generating set, to set up the association between service data and running status based on the several data from wind power generating set collection, the optimization of subcontrol strategy.

According to an aspect of the present invention, a kind of method of service data for analyzing wind power generating set is provided, described method comprises: obtain the primary data relevant to the operation of wind power generating set, and extract the attribute relevant to the change of wind power generating set running status from described primary data, and described primary data is sampled, obtains the sampled data of multiple running status classification;The sampled data of described multiple running status classification is carried out statistical study by attribute according to extracting respectively, to obtain multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting described multiple running status classification; The variation diagram of the operation characteristic amount specified in status time period is drawn, to characterize the change of described wind power generating set its operation characteristic amount under prescribed conditions according to the multiple operation characteristic amount obtained.

Can selection of land, described primary data comprises the service data of wind power generating set, state data and the unit fault data that gather in predetermined timing statistics section.

Can selection of land, the described primary data statistic data that to be also included in described timing statistics section relevant to described service data, state data and unit fault data.

Can selection of land, described extract the process changing relevant attribute with wind power generating set running status from described primary data and comprise: according to principal component analytical method and factor-analysis approach, each attribute described primary data is screened and returned choosing; To screening and return the attribute selected to analyze according to genetic algorithm, select the attribute that the change of the running status to wind power generating set is relevant.

Can selection of land, described method also comprises: use the sampled data of described multiple running status classification and described primary data, respectively multiple operation characteristic amounts of the wind power generating set of the described predetermined condition meeting described multiple running status classification are checked according to Wilcoxen signed rank test algorithm, to reject the operation characteristic amount that Wilcoxen statistic does not meet normal distribution.

Can selection of land, described method also comprises: the service data of the wind power generating set of the collection of scheduled duration and state data before acquisition; The value of the relevant attribute of the change of the running status to wind power generating set is calculated according to described service data and state data; Value according to the attribute relevant to the change of the running status of wind power generating set calculated respectively variation diagram with the operation characteristic amount in described appointment status time period determine the running status that described wind power generating set is current.

Can selection of land, described described primary data is sampled, the process of the sampled data obtaining multiple running status classification comprises: according to described multiple running status classification, described primary data is carried out stratified sampling, to obtain the sampled data of described multiple running status classification.

Preferably, by performing the step in described method based on the data platform of the distributed framework of HADOOP, wherein, store described primary data and sampled data with HIVE form, the attribute information of extraction is stored in relational database.

According to a further aspect in the invention, also providing the system of a kind of service data for analyzing wind power generating set, described system comprises: primary data acquisition device, for obtaining the primary data relevant to the operation of wind power generating set; Attributes extraction device, for extracting the attribute relevant to the change of wind power generating set running status from described primary data; Sampling device, for described primary data being sampled, obtains the sampled data of multiple running status classification; Feature apparatus for establishing, for carrying out statistical study according to the attribute extracted respectively to the sampled data of described multiple running status classification, to obtain multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting described multiple running status classification; Model apparatus for establishing, for drawing the variation diagram of the operation characteristic amount specified in status time period, to characterize the change of described wind power generating set its operation characteristic amount under prescribed conditions according to the multiple operation characteristic amount obtained.

Can selection of land, described primary data comprises the service data of wind power generating set, state data and the unit fault data that gather in predetermined timing statistics section.

Can selection of land, the described primary data statistic data that to be also included in described timing statistics section relevant to described service data, state data and unit fault data.

Can selection of land, described attributes extraction device comprises: attribute primary election unit, for each attribute in described primary data being screened according to principal component analytical method and factor-analysis approach and return choosing; The selected unit of attribute, for screening and returning the attribute selected to analyze according to genetic algorithm, selects the attribute that the change of the running status to wind power generating set is relevant.

Can selection of land, described system also comprises: characteristic quantity checking device, for using the sampled data of described multiple running status classification and described primary data, respectively multiple operation characteristic amounts of the wind power generating set of the described predetermined condition meeting described multiple running status classification are checked according to Wilcoxen signed rank test algorithm, to reject the operation characteristic amount that Wilcoxen statistic does not meet normal distribution.

Can selection of land, described system also comprises: state determination device, for the service data of wind power generating set of collection and the state data of scheduled duration before obtaining, calculate the value of the relevant attribute of the change of the running status with wind power generating set according to described service data and state data, and the value changing relevant attribute according to the running status to wind power generating set of calculating respectively variation diagram with the operation characteristic amount in described appointment status time period determine the running status that described wind power generating set is current.

Can selection of land, described sampling device for described primary data being carried out stratified sampling according to described multiple running status classification, to obtain the sampled data of described multiple running status classification.

Can selection of land, described system is by the data platform based on the distributed framework of HADOOP, and wherein, described primary data and sampled data are stored with HIVE form, described attributes extraction device extract attribute information be stored in relational database.

The method of the service data for analyzing wind power generating set provided according to embodiments of the present invention and system are by big data mining technology, from the service data of wind power generating set, state data, unit number of faults according to this and statistical Data Mining go out the variation diagram of the operation characteristic amount corresponding to one or more operating states of the units, such that it is able to these variation diagrams are as condition adjudgement model, the service data current according to wind power generating set judges its running status.

Accompanying drawing explanation

Fig. 1 is the framework schema of the service data for analyzing wind power generating set illustrating the general plotting according to the present invention;

Fig. 2 is the illustrative diagram of the data of the data warehouse storage illustrating the big data platform of wind-powered electricity generation;

Fig. 3 is the schema of the method for the service data for analyzing wind power generating set illustrating according to embodiments of the present invention;

Fig. 4 is the example of the variation diagram illustrating the operation characteristic amount that the method for the service data for analyzing wind power generating set according to embodiments of the present invention is drawn;

Fig. 5 is the logic diagram of the system of the service data for analyzing wind power generating set illustrating according to embodiments of the present invention two.

Embodiment

The basic design of the present invention is, by to all kinds of running of wind generating set data gathered, state data, unit fault data, based on big data technique and the Various types of data mining algorithm such as classification, cluster, create the eigenwert detection algorithm of wind-powered electricity generation service data, acquirement can accurate description set state change operation characteristic amount, set up the correlation model between set state and operation characteristic amount, and then use these features the running status that wind power generating set is current to be judged.

For this reason, the technical scheme of the service data for analyzing wind power generating set of proposition of the present invention can be carried out based on the big data platform of the distributed framework of HADOOP.

Every portion wind power generating set, in the process run, gathers/produces a lot of service data and state data; In addition, the control center of wind energy turbine set is in the process that the wind power generating set in wind energy turbine set carries out monitor and forecast, a large amount of service datas and state data can be got from each wind power generating set, and the unit fault data etc. about wind power generating set can be collected. These data all can be used as the object of the technical scheme process that the present invention proposes. In addition, also producing the statistic data relevant with state data to service data in the process wind power generating set in wind energy turbine set monitored, these statistic datas also can be used as the input data of the technical scheme that the present invention proposes.

Specifically, the service data of wind power generating set can comprise, but it is not limited to, the data of operating parameter and environmental parameter, the generated energy of the pressure of the rotating speed of such as generator, envrionment temperature, driftage clamp, wind speed, wind power generating set, power, voltage, electric current etc.

The overall operation state of the state data instruction unit of wind power generating set, such as, but is not limited to, starts shooting, shuts down, heavily opens, standby state and time of origin thereof etc.

The unit fault data comprise of wind power generating set, but be not limited to, the alarm message etc. that the time of unit generation fault, fault type, failture evacuation time, fault are correlated with.

Aforesaid statistic data comprise the operating parameter collected from unit within for some time and the Various types of data of environmental parameter are added up the Various types of data obtained maximum value, minimum value, average, the statistical value such as the frequency.

Hereinafter with reference to Fig. 1 detail according to the process of the service data for analyzing wind power generating set of the general plotting of the present invention. Fig. 1 is the illustrative diagram of the data warehouse storage data illustrating the big data platform of wind-powered electricity generation.

With reference to Fig. 1, first, gather the Various types of data relevant to the operation of wind power generating set, and these data are stored in the data warehouse of big data platform (10). Various types of data mentioned here be the service data of aforesaid wind power generating set, state data, unit number of faults according to this and all kinds of statistic data.

Specifically, by technology such as ETL process (extract, change, load), obtain from multiple data sources and comprise (i.e. service data and state data), unit fault data, statistic data etc.

Usually, difference is all there is in the diversified data obtained from multiple data source from the aspect such as data layout, existing way, therefore before these data being stored in the data warehouse of big data platform, it is necessary to these data are comprised the process such as data importing, stdn, cleaning.

Specifically, first, data source can be the file etc. of the real-time data base of particular memory form, relevant database or different-format, it is necessary to import these data from data source. Secondly, according to the data management standard of enterprise, name, the form of Various types of data is carried out standardization, transfers and use these data to facilitate. Again, set rational data area and rule according to the feature of Various types of data, reject the not abnormal data in reasonable value range. Such as, data check and cleaning can be carried out according to following rule: whether allow whether null value, type mate, whether form mates, whether data are unique, whether data interpretation exists ambiguity, whether data have consistence, whether meet table associates relation, whether meet predetermined data area, whether meet business logic etc.

After carrying out aforementioned processing, it is possible to be stored in Hadoop data warehouse to the formal distribution formula of Hive data.Like this, reading and writing data speed can either be accelerated, can be again that the distributed computing in later stage is ready.

Fig. 2 is the illustrative diagram of the data of the data warehouse storage illustrating the big data platform of wind-powered electricity generation. As shown in Figure 2, will gather and treated datum number storage according to corresponding data in warehouse classify under. Specifically, the blower fan information table that the data of storage are organized as wind energy turbine set information table, fault data sheet, machine group type information table, statistic data table, real time data table, fan condition data sheet and associate with these tables.

The process of aforementioned data collection and storage can be performed cumulatively, it is also possible to carry out this process once. Hereafter, when needing the analyzing and processing carrying out proposition of the present invention, perform follow-up process.

First, extract the attribute (20) relevant to the change of wind power generating set running status based on the relevant data (namely hereinafter described " primary data ") stored in data warehouse and go forward side by side that rower is accurate samples (30).

On the one hand, described primary data is analyzed, extract the attribute relevant to the change of wind power generating set running status, that is, from the attribute that unit operation is correlated with, select relevant attribute strong to set state change according to predetermined selection algorithm.

First, each attribute in described primary data is screened and returned choosing. Such as, it is possible to use the attribute that analyze is screened and returns choosing by following method:

I. main analysis of components: by multiple variable by linear transformation to select a kind of multiviate statistical analysis method of less number significant variable. Can for original all attributes proposed, it is unnecessary to be left out by the attribute (attribute of close relation) repeated, set up the least possible new attribute so that these new attributes are uncorrelated between two, and these new attributes keep original information as far as possible at the message context of reflection problem.

Ii. factorial analysis: the statistical technique extracting the general character factor from variable group, is included into a factor by the variable (attribute) of homogeneity, can reduce the number of attribute, also can check the hypothesis of relation between attribute. The method of factorial analysis about has kind more than 10, and such as center of gravity method, image analysing computer method, maximum likelihood solution, least square method, A Erfa take out because of method, draw typical case difficult to understand to take out because of method etc.

Then, the Attributions selection algorithm based on genetic algorithm can be used, from the attribute relevant with returning the unit operation of choosing through screening, select some attributes obviously relevant to set state change. In the genetic algorithm used, it is possible to information gain carries out the index of its individual evaluation, Selecting operation, crossing operation and variation computing as measure. Specifically, the information gain of computation attribute is carried out by the difference of information entropy and condition entropy:

InfoGains [k]=entropy [k]-condEntropy [k]

Wherein, InfoGains [k] represents the information gain of kth attribute, and entropy [k] represents the information entropy of kth attribute, and condEntropy [k] represents the condition entropy of kth attribute.

Using the attribute that gone out based on the Attributions selection algorithms selection of genetic algorithm by this as strong relevant attribute.

The process analyzed by aforementioned attributes, the insignificant attribute of analysis is filtered out by some, such as accumulative conduction time, system normal time, blower fan self-starting counting etc., and extract the attribute or combinations of attributes that can be used as and analyze input, such as power, wind speed-rotating speed, temperature-voltage, blade angle-generated energy etc., the mode being combined through linear combination of these attributes forms new attribute.Owing to itself just has certain business association, it is possible to make follow-up analysis more accurate.

On the other hand, described primary data is sampled, obtain the sampled data of multiple running status classification, the mode of stratified sampling is preferably taked to sample: first according to target call, primary data to be divided into some classifications, each layer (class) is being carried out such as stochastic sampling, get up to form sample set by the sample extracted every layer combination, set up for follow-up feature foundation process and model (namely hereinafter described variation diagram). Can obtaining different sample sets according to different set states, as unit fault, unit such as heavily opens at the different states event, extracts different sample sets, uses for subsequent calculations.

Different data storage format can be utilized to come storing sample data and attribute data. Owing to the data volume of sampled data is big, therefore can continue sampled data to be stored in Hive data warehouse; And the combinations of attributes obtained by attributive analysis has, with classification, the form and field determined, can be stored into and directly may have access in the relational database transferred.

After completing attributes extraction and standard sampling, perform the process (40) that the feature shown in Fig. 1 is set up.

Specifically, the sampled data of described multiple running status classification is carried out statistical study by attribute according to extracting respectively, to obtain multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting described multiple running status classification, operation characteristic amount described in each comprises the value of single attribute or the value of multiple attribute.

By various data statistical approach such as such as regression analysis, variance analysis, correlation analysis, set up the operation characteristic amount of the various value comprising attribute or combinations of attributes, by the sampled data of the classification that the sampling of standard before obtains, classification and pattern recognition algorithm is used to judge, the operation characteristic amount meeting judged result is preserved, as the result that feature is set up.

Such as, in the process using correlation analysis, the value that setting relation conefficient is predetermined judges as threshold value, by the value of attribute or combinations of attributes is carried out comparison between two, between the combination finding wind speed rotating speed and the combination of blade angle-generated energy, the relation conefficient of its most of the time has exceeded threshold value 0.8, then illustrate that the value of this attribute or combinations of attributes has certain variation tendency, using the value of this attribute or combinations of attributes as selected operation characteristic amount.

On this basis, it may be preferred that by predetermined check algorithm, these selected operation characteristic amounts are verified, to guarantee reasonableness and the accuracy of these operation characteristic amounts. Such as, sample data set and the overall primary data extracted before wilcoxon method (Wilcoxen signed rank test method) can be used to use are compared checking, as follows with the unique point formula of Double Data collection:

z = W x - μ ± 0.5 σ = w x - n 1 ( n 1 + n 2 + 1 ) 2 ± 0.5 n 1 n 2 ( n 1 + n 2 + 1 ) 12 - n 1 n 2 Σ ( τ j 3 - τ ) 12 ( n 1 + x 2 ) ( x 1 + x 2 - 1 ) ~ N ( 0 , 1 )

Wherein, z represents the wilcoxon statistic of this unique point. If this statistic meets the normal distribution of N (0,1), then assert that this operation characteristic amount is effective. WxFor order and and the distance absolute value sum of all values and median of this operation characteristic amount. μ is the average of this operation characteristic amount. σ is the standard deviation of this operation characteristic amount. n1For the number of the primary data (totally) that this operation characteristic amount uses, n2For the number of the 2nd data set (certain sample data set) that this operation characteristic amount uses. τ is the knot value in operation characteristic amount, τjFor the number of wherein jth knot value.

If by checking, then illustrating that this operation characteristic amount index presents obvious change at this sample data set place, filtering out multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting described multiple running status classification further. Such as, can be normal operation, rebooting status, arbitrary fault state set up the multiple operation characteristic amounts meeting these conditions respectively.

Hereafter, carry out the process (50) that the model shown in Fig. 1 is set up. Specifically, draw the variation diagram of the operation characteristic amount specified in status time period according to the multiple operation characteristic amount obtained, to characterize the change of described wind power generating set its operation characteristic amount under prescribed conditions.

In the process that feature is set up, get multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting multiple running status classification. In the process that model is set up, first the running status classification of concrete analysis and the detection period of this running status is determined, such as the predetermined amount of time before fault generation, then carry out dot matrix variation diagram drafting for the operation characteristic amount that the detection period of this running status is corresponding, thus the change of wind power generating set its operation characteristic amount under this running status can be observed.

Fig. 4 illustrates the variation diagram of the operation characteristic amount detected when unit occurs certain fault state to occur, it represents for some time before certain set generator generation fault, the figure that this eigenwert of wind speed-temperature excavated according to aforementioned processing changes, this shows, this operation characteristic amount creates obvious reaction for the state change of unit. The running status of unit can be predicted further by the variation diagram of the operation characteristic amount corresponding to running status drawn as the service data that state model is current according to unit.

On this basis, it may be preferred that before also calculating the generation of all same class trouble spots by these operation characteristic quantitative statistics data, the possibility of this feature transportation load generation ANOMALOUS VARIATIONS. If this possibility reaches certain ratio value (such as more than 75%), then can for the curvilinear characteristic of this variation diagram, set up the calibration model of the average for this operation characteristic amount, variance, maximum value, normal distribution etc., detect the average of this operation characteristic amount within continuous for some time, arrange whether this operation characteristic amount of threshold decision exceeds normal horizontal scope according to what set in advance, to be revised by this variation diagram further.

According to the process that aforesaid data gathering, attributes extraction, standard sampling, feature foundation and model are set up, from the service data of wind power generating set, state data, unit number of faults according to this and statistical Data Mining go out the variation diagram of the operation characteristic amount corresponding to one or more operating states of the units, such that it is able to these variation diagrams are as condition adjudgement model, the service data current according to wind power generating set judges its running status, and carries out the prediction of follow-up state.

The exemplary embodiment of the present invention is described in detail below in conjunction with accompanying drawing 3~Fig. 5.

Embodiment one

Fig. 3 is the schema of the method for the service data for analyzing wind power generating set illustrating according to embodiments of the present invention.

With reference to Fig. 3, in step S310, obtain the primary data (S310A) relevant to the operation of wind power generating set, and extract the attribute (S310B) relevant to the change of wind power generating set running status from described primary data.

Primary data can comprise, but is not limited to, the service data of the wind power generating set of collection, state data and unit fault data in predetermined timing statistics section. Preferably, primary data is also included in statistic data relevant to described service data, state data and unit fault data in described timing statistics section.

As previously mentioned, can obtain, from HADOOP data warehouse, the input that these data excavate as follow-up data.

Specifically, as previously mentioned, the process of step S310BA comprises: each attribute in described primary data is screened according to principal component analytical method and factor-analysis approach and returns choosing, and to screening and return the attribute selected to analyze according to genetic algorithm, select the attribute that the change of the running status to wind power generating set is relevant.

In step S320, described primary data is sampled, obtain the sampled data of multiple running status classification.

Specifically, according to described multiple running status classification, described primary data is carried out stratified sampling, to obtain the sampled data of described multiple running status classification.

In step S330, according to the attribute extracted in step S310, the sampled data of described multiple running status classification is carried out statistical study respectively, to obtain multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting described multiple running status classification, operation characteristic amount described in each comprises the value of single attribute or the value of multiple attribute.

Specifically, as previously mentioned, by various data statistical approach such as such as regression analysis, variance analysis, correlation analysis, the operation characteristic amount of the various values comprising attribute or combinations of attributes of the predetermined condition meeting multiple running status classification is set up.

In addition, in order to guarantee the dependency of multiple operation characteristic amounts that aforementioned processing obtains further, by the sampled data of the classification that the sampling of standard before obtains, it may also be useful to classification and pattern recognition algorithm judge, retain the operation characteristic amount meeting judged result.

Specifically, use the sampled data of described multiple running status classification and described primary data, respectively multiple operation characteristic amounts of the wind power generating set of the described predetermined condition meeting described multiple running status classification are checked according to Wilcoxen signed rank test algorithm, to reject the operation characteristic amount that Wilcoxen statistic does not meet normal distribution.

In step S340, draw the variation diagram of the operation characteristic amount specified in status time period according to the multiple operation characteristic amount obtained, to characterize the change of described wind power generating set its operation characteristic amount under prescribed conditions.

By the process of abovementioned steps, can from the service data of wind power generating set, state data, unit number of faults according to this and statistical Data Mining go out the variation diagram of the operation characteristic amount corresponding to one or more operating states of the units, such that it is able to these variation diagrams are as condition adjudgement model, the service data current according to wind power generating set judges its running status.

Correspondingly, the method for the service data for analyzing wind power generating set that exemplary embodiment of the present proposes also comprises: the service data of the wind power generating set of the collection of scheduled duration and state data before acquisition; The value of the relevant attribute of the change of the running status to wind power generating set is calculated according to described service data and state data; Value according to the attribute relevant to the change of the running status of wind power generating set calculated respectively variation diagram with the operation characteristic amount in described appointment status time period determine the running status that described wind power generating set is current.

On this basis, also can setting up based on Time-series Techniques, set up unit behavior prediction model further, the unit behavior in for some time be predicted, auxiliary monitoring management personnel formulate or adjustment unit operation reserve.

Embodiment two

Fig. 5 is the logic diagram of the system of the service data for analyzing wind power generating set illustrating according to embodiments of the present invention two.

With reference to Fig. 5, the system of the service data for analyzing wind power generating set that embodiment two provides comprises: primary data acquisition device 510, attributes extraction device 520, sampling device 530, feature apparatus for establishing 540 and model apparatus for establishing 550.

Primary data acquisition device 510 for obtain to wind power generating set run relevant primary data, described primary data comprises the service data of the wind power generating set of collection in predetermined timing statistics section, state data and unit fault data.

Can selection of land, the described primary data statistic data that to be also included in described timing statistics section relevant to described service data, state data and unit fault data.

Attributes extraction device 520, for described primary data being analyzed, extracts the attribute relevant to the change of wind power generating set running status.

Specifically, attributes extraction device 520 comprises: attribute primary election unit, for each attribute in described primary data being screened according to principal component analytical method and factor-analysis approach and return choosing; The selected unit of attribute, for screening and returning the attribute selected to analyze according to genetic algorithm, selects the attribute that the change of the running status to wind power generating set is relevant.

Sampling device 530, for described primary data being sampled, obtains the sampled data of multiple running status classification.

Specifically, sampling device 530 for described primary data being carried out stratified sampling according to described multiple running status classification, to obtain the sampled data of described multiple running status classification.

Feature apparatus for establishing 540 is for carrying out statistical study according to the attribute extracted respectively to the sampled data of described multiple running status classification, to obtain multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting described multiple running status classification, operation characteristic amount described in each comprises the value of single attribute or the value of multiple attribute.

Can selection of land, this system also comprises: characteristic quantity checking device, for using the sampled data of described multiple running status classification and described primary data, respectively multiple operation characteristic amounts of the wind power generating set of the described predetermined condition meeting described multiple running status classification are checked according to Wilcoxen signed rank test algorithm, to reject the operation characteristic amount that Wilcoxen statistic does not meet normal distribution.

Model apparatus for establishing 550 for drawing the variation diagram of the operation characteristic amount specified in status time period according to the multiple operation characteristic amount obtained, to characterize the change of described wind power generating set its operation characteristic amount under prescribed conditions.

Can selection of land, this system also comprises: state determination device 560, for the service data of wind power generating set of collection and the state data of scheduled duration before obtaining, calculate the value of the relevant attribute of the change of the running status with wind power generating set according to described service data and state data, and the value changing relevant attribute according to the running status to wind power generating set of calculating respectively variation diagram with the operation characteristic amount in described appointment status time period determine the running status that described wind power generating set is current.

Can selection of land, described system is by the data platform based on the distributed framework of HADOOP, and wherein, described primary data and sampled data are stored with HIVE form, described attributes extraction device extract attribute information be stored in relational database.

It may be noted that, according to the needs implemented, each components/steps described in the application can be split as more multi-part/step, it is possible to the part operational group of two or more components/steps or components/steps is synthesized new components/steps, to realize the object of the present invention.

The above-mentioned method according to the present invention can at hardware, firmware realizes, or it is implemented as and can be stored in recording medium (such as CDROM, RAM, floppy disk, hard disk or magneto-optic disk) in software or computer code, or be implemented by the original storage of web download in long-range recording medium or non-temporary transient machine computer-readable recording medium and the computer code that will be stored in local recording medium, thus method described here can be stored in use multi-purpose computer, such software processes on the recording medium of application specific processor or able to programme or specialized hardware (such as ASIC or FPGA).It is appreciated that, computer, treater, microprocessor controller or programmable hardware comprise can store or receive software or computer code storage assembly (such as, RAM, ROM, flash memory etc.), when described software or computer code are by computer, treater or hardware access and execution, it is achieved treatment process described here. In addition, when the code for realizing the process shown in this accessed by multi-purpose computer, multi-purpose computer is converted to the special purpose computer for performing the process shown in this by the execution of code.

The above; it is only the specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, any it is familiar with those skilled in the art in the technical scope that the present invention discloses; change can be expected easily or replace, all should be encompassed within protection scope of the present invention. Therefore, protection scope of the present invention should be as the criterion with the protection domain of described claim.

Claims (16)

1. one kind for analyzing the method for the service data of wind power generating set, it is characterised in that, described method comprises:
Obtain the primary data relevant to the operation of wind power generating set,
And extract the attribute relevant to the change of wind power generating set running status from described primary data;
Described primary data is sampled, obtains the sampled data of multiple running status classification;
The sampled data of described multiple running status classification is carried out statistical study by attribute according to extracting respectively, obtain multiple operation characteristic amounts of the predetermined condition meeting described multiple running status classification, the variation diagram of the operation characteristic amount specified in status time period is drawn, to characterize the change of described wind power generating set its operation characteristic amount under prescribed conditions according to the multiple operation characteristic amount obtained.
2. method according to claim 1, it is characterised in that, described primary data comprises the service data of wind power generating set, state data and the unit fault data that gather in predetermined timing statistics section.
3. method according to claim 2, it is characterised in that, described primary data is also included in statistic data relevant to described service data, state data and unit fault data in described timing statistics section.
4. method according to claim 3, it is characterised in that, the process of the described attribute relevant to the change of wind power generating set running status from the extraction of described primary data comprises:
According to principal component analytical method and factor-analysis approach, each attribute in described primary data is screened and returned choosing;
To screening and return the attribute selected to analyze according to genetic algorithm, select the attribute that the change of the running status to wind power generating set is relevant.
5. method according to claim 4, it is characterised in that, described method also comprises:
Use the sampled data of described multiple running status classification and described primary data, respectively multiple operation characteristic amounts of the wind power generating set of the described predetermined condition meeting described multiple running status classification are checked according to Wilcoxen signed rank test algorithm, to reject the operation characteristic amount that Wilcoxen statistic does not meet normal distribution.
6. method according to any one of claim 2~5, it is characterised in that, described method also comprises:
The service data of the wind power generating set of the collection of scheduled duration and state data before acquisition,
The value of the relevant attribute of the change of the running status to wind power generating set is calculated according to described service data and state data,
Value according to the attribute relevant to the change of the running status of wind power generating set calculated respectively variation diagram with the operation characteristic amount in described appointment status time period determine the running status that described wind power generating set is current.
7. method according to any one of claim 3~5, it is characterised in that, described described primary data to be sampled, the process of the sampled data obtaining multiple running status classification comprises:
According to described multiple running status classification, described primary data is carried out stratified sampling, to obtain the sampled data of described multiple running status classification.
8. method according to any one of Claims 1 to 5, it is characterized in that, by performing the step in described method based on the data platform of the distributed framework of HADOOP, wherein, store described primary data and sampled data with HIVE form, the attribute information of extraction is stored in relational database.
9. one kind for analyzing the system of the service data of wind power generating set, it is characterised in that, described system comprises:
Primary data acquisition device, for obtaining the primary data relevant to the operation of wind power generating set;
Attributes extraction device, for extracting the attribute relevant to the change of wind power generating set running status from described primary data;
Sampling device, for described primary data being sampled, obtains the sampled data of multiple running status classification;
Feature apparatus for establishing, for carrying out statistical study according to the attribute extracted respectively to the sampled data of described multiple running status classification, to obtain multiple operation characteristic amounts of the wind power generating set of the predetermined condition meeting described multiple running status classification;
Model apparatus for establishing, for drawing the variation diagram of the operation characteristic amount specified in status time period, to characterize the change of described wind power generating set its operation characteristic amount under prescribed conditions according to the multiple operation characteristic amount obtained.
10. system according to claim 9, it is characterised in that, described primary data comprises the service data of wind power generating set, state data and the unit fault data that gather in predetermined timing statistics section.
11. systems according to claim 10, it is characterised in that, described primary data is also included in statistic data relevant to described service data, state data and unit fault data in described timing statistics section.
12. systems according to claim 11, it is characterised in that, described attributes extraction device comprises:
Attribute primary election unit, for screening according to principal component analytical method and factor-analysis approach each attribute in described primary data and return choosing;
The selected unit of attribute, for screening and returning the attribute selected to analyze according to genetic algorithm, selects the attribute that the change of the running status to wind power generating set is relevant.
13. systems according to claim 12, it is characterised in that, described system also comprises:
Characteristic quantity checking device, for using the sampled data of described multiple running status classification and described primary data, respectively multiple operation characteristic amounts of the wind power generating set of the described predetermined condition meeting described multiple running status classification are checked according to Wilcoxen signed rank test algorithm, to reject the operation characteristic amount that Wilcoxen statistic does not meet normal distribution.
14. systems according to any one of claim 10~13, it is characterised in that, described system also comprises:
State determination device, for the service data of wind power generating set of collection and the state data of scheduled duration before obtaining, calculate the value of the relevant attribute of the change of the running status with wind power generating set according to described service data and state data, and the value changing relevant attribute according to the running status to wind power generating set of calculating respectively variation diagram with the operation characteristic amount in described appointment status time period determine the running status that described wind power generating set is current.
15. systems according to any one of claim 11~13, it is characterised in that, described sampling device for described primary data being carried out stratified sampling according to described multiple running status classification, to obtain the sampled data of described multiple running status classification.
16. systems according to any one of claim 9~13, it is characterized in that, described system is by the data platform based on the distributed framework of HADOOP, wherein, described primary data and sampled data are stored with HIVE form, and the attribute information that described attributes extraction device extracts is stored in relational database.
CN201511029436.4A 2015-12-31 2015-12-31 For analyzing the method and system of the operation data of wind power generating set CN105677791B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201511029436.4A CN105677791B (en) 2015-12-31 2015-12-31 For analyzing the method and system of the operation data of wind power generating set

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201511029436.4A CN105677791B (en) 2015-12-31 2015-12-31 For analyzing the method and system of the operation data of wind power generating set

Publications (2)

Publication Number Publication Date
CN105677791A true CN105677791A (en) 2016-06-15
CN105677791B CN105677791B (en) 2019-03-08

Family

ID=56298429

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201511029436.4A CN105677791B (en) 2015-12-31 2015-12-31 For analyzing the method and system of the operation data of wind power generating set

Country Status (1)

Country Link
CN (1) CN105677791B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106640547A (en) * 2016-11-24 2017-05-10 东北电力大学 Method and system for monitoring state of wind generating set

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012149901A1 (en) * 2011-05-03 2012-11-08 北京中瑞泰科技有限公司 Similarity curve-based device malfunction early-warning and optimization method and system
CN103234767A (en) * 2013-04-21 2013-08-07 蒋全胜 Nonlinear fault detection method based on semi-supervised manifold learning
CN103323772A (en) * 2012-03-21 2013-09-25 北京光耀能源技术股份有限公司 Wind driven generator operation state analyzing method based on neural network model
CN103674234A (en) * 2013-12-23 2014-03-26 北京天源科创风电技术有限责任公司 State early warning method and system for abnormal vibration of wind generating set

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012149901A1 (en) * 2011-05-03 2012-11-08 北京中瑞泰科技有限公司 Similarity curve-based device malfunction early-warning and optimization method and system
CN103323772A (en) * 2012-03-21 2013-09-25 北京光耀能源技术股份有限公司 Wind driven generator operation state analyzing method based on neural network model
CN103234767A (en) * 2013-04-21 2013-08-07 蒋全胜 Nonlinear fault detection method based on semi-supervised manifold learning
CN103674234A (en) * 2013-12-23 2014-03-26 北京天源科创风电技术有限责任公司 State early warning method and system for abnormal vibration of wind generating set

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106640547A (en) * 2016-11-24 2017-05-10 东北电力大学 Method and system for monitoring state of wind generating set

Also Published As

Publication number Publication date
CN105677791B (en) 2019-03-08

Similar Documents

Publication Publication Date Title
Pinson et al. On‐line assessment of prediction risk for wind power production forecasts
Hochachka et al. Data‐mining discovery of pattern and process in ecological systems
Beccali et al. Forecasting daily urban electric load profiles using artificial neural networks
Moreira et al. SPI-based drought category prediction using loglinear models
Liu et al. A systematic approach for dynamic security assessment and the corresponding preventive control scheme based on decision trees
KR101316486B1 (en) Error detection method and system
Jain et al. Data mining techniques: a survey paper
Dickson et al. Mapping the probability of large fire occurrence in northern Arizona, USA
Feng et al. A data-driven multi-model methodology with deep feature selection for short-term wind forecasting
Hu et al. A hybrid forecasting approach applied to wind speed time series
CN102509178B (en) Distribution network device status evaluating system
Knape et al. Effects of weather and climate on the dynamics of animal population time series
Bianchi et al. Short-term electric load forecasting using echo state networks and PCA decomposition
CN102130783B (en) Intelligent alarm monitoring method of neural network
JP5301717B1 (en) Equipment condition monitoring method and apparatus
Zhang et al. Feature selection for high-dimensional machinery fault diagnosis data using multiple models and radial basis function networks
JP3489279B2 (en) Data analyzer
Robbirt et al. Comparing IUCN and probabilistic assessments of threat: do IUCN red list criteria conflate rarity and threat?
Long et al. Improved diagnostics for the incipient faults in analog circuits using LSSVM based on PSO algorithm with Mahalanobis distance
Li et al. Applications of Bayesian methods in wind energy conversion systems
Kusiak et al. Analysis of wind turbine vibrations based on SCADA data
Bhaskar et al. Wind speed forecasting: Present status
Petre A decision tree for weather prediction
Anders et al. Innovations in power systems reliability
CN103996077B (en) A kind of electrical equipment malfunction Forecasting Methodology based on multidimensional time-series

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