CN105930900B - The Forecasting Methodology and system of a kind of hybrid wind power generation - Google Patents
The Forecasting Methodology and system of a kind of hybrid wind power generation Download PDFInfo
- Publication number
- CN105930900B CN105930900B CN201610300971.7A CN201610300971A CN105930900B CN 105930900 B CN105930900 B CN 105930900B CN 201610300971 A CN201610300971 A CN 201610300971A CN 105930900 B CN105930900 B CN 105930900B
- Authority
- CN
- China
- Prior art keywords
- wind
- wind power
- forecast model
- hybrid
- power
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas 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
Abstract
The present invention discloses the Forecasting Methodology and system of a kind of hybrid wind power generation, and the Forecasting Methodology of the hybrid wind power generation includes:The historical data of the wind direction of wind power plant, wind speed and corresponding wind power output power is obtained, and historical data is sampled to obtain sample data;Discriminatory analysis is carried out to the statistical property of sample data, obtain wind frequency concentration and wind power output power difference reaches the wind direction of discrepancy threshold, and fuzzy hierarchy clustering procedure is used according to acquired wind direction and its with the corresponding relation of wind speed, wind power output power, sample data is divided into three classes;Every class sample is trained using neural network algorithm, the specific wind-power electricity generation forecast model of three classes is correspondingly formed, then merges processing, establish the hybrid wind power generation forecast model for being predicted to wind-power electricity generation production capacity.Therefore, model prediction is targetedly carried out to different wind directions, wind speed by implementing of the invention can realize, and the precision of prediction of wind-power electricity generation power can be improved.
Description
Technical field
The present invention relates to the technical field of wind power prediction, the Forecasting Methodology of more particularly to a kind of hybrid wind power generation and
System.
Background technology
When carrying out wind-power electricity generation, wind power output power is not only influenceed wind park by wind speed, and wind direction is also to neglect
Depending on factor, but due to the effect of atmospheric pressure, wind speed and direction can all change at any time, thus the output of wind-driven generator
Power also has the characteristics that fluctuation and randomness, and the fluctuation of this power output is unfavorable for the flat of regional power grid overall operation
Stability and security.Therefore it is effective to facilitate power operation department to carry out, it is necessary to make rational prediction to wind power output power
Scheduling.
At present, in terms of short-term wind-power electricity generation prediction, two kinds of sides are mainly predicted using physical model prediction and statistical model
Formula.Wherein:
1) physical model is predicted, mainly obtains wind speed, wind direction, gas according to the prediction result of numerical weather forecast system
The weather datas such as pressure, temperature, then calculated according to the information such as contour, roughness, barrier, thermal stratification around wind power plant
The information such as wind speed, wind direction to wind-powered machine unit hub height, wind power plant is finally calculated according to the power curve of wind power plant
Power output.Here, physical modeling, including the landform of wind field, surface vegetation and roughness, week are carried out to wind power plant location
Barrier etc. is enclosed, while also the firm height of the wheel of blower fan in itself, power curve, machine driving and control strategy etc. are built
Mould.The parameter of other this method input forecasts (NWP) model for digital meteorology.
But because the physical model equation of physical model prediction lacks constraint and the weather forecast renewal frequency of elasticity
Low reason, it is set not adapt to short-term wind-force prediction.In addition, although physical model Forecasting Methodology does not need historical data,
Wind power plant is gone into operation and can be carried out, but needs accurate NWP data and the details in wind power plant location, input parameter compared with
It is more, and the collection and processing of NWP data are also relatively complicated.
2) statistical model is predicted, this method does not consider the physical process of wind speed change, is found out according to historical statistical data
Weather conditions and the relation of output of wind electric field, then according to measured data and numerical value data of weather forecast to Power Output for Wind Power Field
It is predicted.The essence of this method is the input (NWP historical statistical datas, measured data) in system between wind power
One vertical mapping relations is built, usually linear relationship.This relation can be showed with the form of function, such as regression analysis
Method, exponential smoothing, time series method, Kalman filtering method, grey method etc., are all based on linear model.These moulds
Type is predicted by catching in data with the information of time and space correlation.
But predict that data sheet one used can to short-term or ultra-short term wind power prediction result for statistical model
To meet required precision.But for the prediction of longer time, prediction result precision is often inadequate, and need prolonged
Measurement data and substantial amounts of data processing work and extra training.In addition, it is bad for abrupt information processing, in training rank
Seldom there are rare weather conditions in section, and also is difficult to Accurate Prediction.
The content of the invention
In view of this, the purpose of the embodiment of the present invention is the Forecasting Methodology and system for proposing a kind of hybrid wind power generation,
It can realize and model prediction is targetedly carried out to different wind directions, wind speed, and the pre- of wind-power electricity generation power can be improved
Survey precision.
For further, the Forecasting Methodology of the hybrid wind power generation includes:Obtain the wind direction of wind power plant, wind speed and corresponding
The historical data of wind power output power, and the historical data is sampled to obtain sample data;To the sample data
Statistical property carries out discriminatory analysis, obtains that wind frequency is concentrated and wind power output power difference reaches the wind direction of discrepancy threshold, and according to
Acquired wind direction and its fuzzy hierarchy clustering procedure is used with the corresponding relation of wind speed, wind power output power, by the sample number
According to being divided into three classes;Every class sample is trained using neural network algorithm, it is pre- to be correspondingly formed the specific wind-power electricity generation of three classes
Model is surveyed, the specific wind-power electricity generation forecast model of this three class is merged into processing, is established for being carried out to wind-power electricity generation production capacity
The hybrid wind power generation forecast model of prediction.
Alternatively, in certain embodiments, the Forecasting Methodology of above-mentioned hybrid wind power generation also includes:Obtain test
Sample data, the hybrid wind power generation forecast model established is debugged;Exported according to hybrid wind power generation forecast model
Test result, the hybrid wind power generation forecast model is modified;Wherein, the sample data of the test is from described
Sample and obtain in the historical data of the wind direction of wind power plant, wind speed and corresponding wind power output power.
Alternatively, in certain embodiments, the Forecasting Methodology of above-mentioned hybrid wind power generation also includes:Utilize the mixing wind
Power generating forecast model, the output production capacity to wind power plant are tested;When being predicted, the wind of input data is judged
To according to identified wind direction, wind-power electricity generation forecast model corresponding to calling calculates the wind power output power of the wind power plant
Predicted value, test result corresponding to output.
Alternatively, in certain embodiments, the generation type of the hybrid wind power generation forecast model also includes:Respectively
Sample weights parameter corresponding to the specific wind-power electricity generation forecast model configuration of three classes;Weighed according to the sample of each forecast model
Weight parameter, the forecast model established according to wind direction is merged, establishes hybrid wind power generation forecast model.
Alternatively, in certain embodiments, the Forecasting Methodology of above-mentioned hybrid wind power generation also includes:To the mixing wind
Power generating forecast model is iterated, and the hybrid wind power generation forecast model is gradually restrained, until output result becomes
In stable.
To realize the above method, the embodiment of the present invention also proposes a kind of forecasting system of hybrid wind power generation, the system bag
Include:
Decimation blocks, for the history number to the wind direction of acquired wind power plant, wind speed and corresponding wind power output power
According to being sampled to obtain sample data;
Analysis module, for carrying out discriminatory analysis to the statistical property of the sample data, obtain wind frequency concentration and wind-powered electricity generation
Difference of Output Power reaches the wind direction of discrepancy threshold;
Processing module, for using mould according to acquired wind direction and its with the corresponding relation of wind speed, wind power output power
Paste layer time clustering procedure, three classes are divided into by the sample data;
Model just models block, and for being trained using neural network algorithm to every class sample, it is specific to be correspondingly formed three classes
Wind-power electricity generation forecast model;
Mixed processing module, for the specific wind-power electricity generation forecast model of this three class to be merged into processing, formed and be used for
The hybrid wind power generation forecast model being predicted to wind-power electricity generation production capacity.
Alternatively, in certain embodiments, the forecasting system of above-mentioned hybrid wind power generation also includes:Test correcting module,
For obtaining the sample data of test, the hybrid wind power generation forecast model established is debugged, corrected and optimized.
Alternatively, in certain embodiments, the forecasting system of above-mentioned hybrid wind power generation also includes:Data input device,
For obtaining test data;Test device, the hybrid wind power generation forecast model is configured with, for according to acquired test
Data, judge the wind direction of input data, according to identified wind direction, call corresponding in the hybrid wind power generation forecast model
Wind-power electricity generation forecast model, the output production capacity to wind power plant are tested;Output device, for obtaining the mixing wind-force
Generating forecast model tests the predicted value of the wind power output power of the obtained wind power plant, and test result corresponding to output.
Alternatively, in certain embodiments, the mixed processing module also includes:Parameter configuration unit, for being respectively
Sample weights parameter corresponding to the specific wind-power electricity generation forecast model configuration of three classes;Merging treatment unit, for according to every
The sample weights parameter of individual forecast model, the forecast model established according to wind direction is merged, it is pre- to establish hybrid wind power generation
Survey model;Iterative optimization unit, for being iterated to the hybrid wind power generation forecast model, make the mixing wind-force
Generating forecast model is gradually restrained, until tending towards stability.
Alternatively, in certain embodiments, the hybrid wind power generation forecast model further comprises wind direction recognition unit,
For judging the wind direction of input data;Model call unit, for the wind direction determined by, call the hybrid wind power generation
Corresponding wind-power electricity generation forecast model in forecast model, the output production capacity to wind power plant are tested.
Relative to prior art, various embodiments of the present invention have advantages below:
After the technical scheme of the embodiment of the present invention, pass through wind direction, wind speed and wind power output power to wind park etc.
The statistical analysis of historical data, the historical data of the wind direction of wind power plant, wind speed and wind power output power is sampled, obtains sample
Notebook data, by fuzzy hierarchy clustering procedure, sample data is divided into three classes according to wind direction, then using Hopfield neutral nets
Algorithm is trained to every class sample data, establishes the wind-power electricity generation forecast model of feature wind direction respectively, to three class forecast models
Processing is merged, builds a hybrid wind power generation forecast model based on different wind directions.Then, established mixing is utilized
Wind-power electricity generation forecast model, by analyzing the wind direction feature of wind park different times, the production capacity of wind park is predicted.To wind
During electric output power is predicted, while consider the influence of wind direction and the aspect of wind-force two, and according to wind direction, wind-force with
The corresponding relation of wind power output power, is carried out fuzzy hierarchy cluster.So, can be by one group of hits for having noise redundancy
According to being divided into the corresponding several parts of sample datas that internal association be present, then these sample datas are modeled respectively, inhomogeneity
The internal characteristicses of other data just can be fully demonstrated by different models, so as to targetedly be carried out to different wind directions, wind speed
Model prediction.
In addition, when being predicted, the wind direction of input data is first judged, it is defeated to obtain wind-powered electricity generation in forecast model corresponding to substitution
Go out the predicted value of power, so as to accurately carry out short-term wind-electricity prediction, improve the precision of wind power prediction.
The more features and advantage of the embodiment of the present invention will be explained in embodiment afterwards.
Brief description of the drawings
The accompanying drawing for forming a part of the embodiment of the present invention is used for providing further understanding the embodiment of the present invention, the present invention
Schematic description and description be used for explain the present invention, do not form inappropriate limitation of the present invention.In the accompanying drawings:
Fig. 1 is a kind of Forecasting Methodology schematic flow sheet of hybrid wind power generation provided in an embodiment of the present invention;
Fig. 2 is the operation principle schematic diagram of Hopfield neural network algorithms in the embodiment of the present invention;
Fig. 3 is the contrast schematic diagram of dynamic wind rose and electric power rose figure in the embodiment of the present invention;
Fig. 4 is the level incidence relation schematic diagram between the other class that sample data is different in the embodiment of the present invention;
Fig. 5 is the classification schematic diagram of sample data in the embodiment of the present invention;
Fig. 6 is the prediction result contrast schematic diagram of different models in the embodiment of the present invention;
Fig. 7 is a kind of composition schematic diagram of the forecasting system of hybrid wind power generation provided in an embodiment of the present invention;
Fig. 8 is the composition schematic diagram of data modeling and correcting device in the embodiment of the present invention.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation describes, it is clear that described embodiment is only part of the embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art are obtained every other under the premise of creative work is not made
Embodiment, belong to the scope of protection of the invention.
It should be noted that in the case where not conflicting, the feature in the embodiment of the present invention and embodiment can be mutual group
Close.
Influenceed while due to wind power output power by wind speed and direction, it is different particularly under identical wind speed
Wind power output power under wind direction has a long way to go, and therefore, does not consider that traditional Individual forecast model error of wind direction is very big, there is mirror
In this, the present invention provides a kind of mixing wind power output power forecast model of the two factors of comprehensive wind direction and wind speed, so as to carry
The precision of prediction of high wind-power electricity generation power.
Below in conjunction with the accompanying drawings, various embodiments of the present invention are described further:
Embodiment of the method
Shown in reference picture 1, it is a kind of Forecasting Methodology flow signal for hybrid wind power generation that the embodiment of the present invention proposes
Figure, the Forecasting Methodology of the hybrid wind power generation comprise the following steps:
S100:The historical data of the wind direction of wind power plant, wind speed and corresponding wind power output power is obtained, and to the history
Data are sampled to obtain sample data;
S102:Discriminatory analysis is carried out to the statistical property of the sample data, obtains wind frequency concentration and wind power output power
Difference reaches the wind direction of discrepancy threshold, and is adopted according to acquired wind direction and its with the corresponding relation of wind speed, wind power output power
With fuzzy hierarchy clustering procedure, the sample data is divided into three classes;
S104:Every class sample is trained using neural network algorithm, it is pre- to be correspondingly formed the specific wind-power electricity generation of three classes
Model is surveyed, the specific wind-power electricity generation forecast model of this three class is merged into processing, is established for being carried out to wind-power electricity generation production capacity
The hybrid wind power generation forecast model of prediction.
In the present embodiment, hybrid wind power generation forecast model is pre-established, passes through the wind direction, wind speed and wind-powered electricity generation to wind park
Found after the statistical analysis of the historical datas such as power output, height output frequency correspond on high frequency wind direction, based on this, this implementation
Example is first sampled to the historical data of the wind direction of wind power plant, wind speed and wind power output power, obtains sample data, by fuzzy
Hierarchical clustering method, sample data is divided into three classes according to wind direction, then using Hopfield neural network algorithms to every class sample
Data are trained, and establish the wind-power electricity generation forecast model of feature wind direction respectively, and processing, structure are merged to three class forecast models
Build a hybrid wind power generation forecast model based on different wind directions.Then, established hybrid wind power generation is utilized to predict mould
Type, by analyzing the wind direction feature of wind park different times, the production capacity of wind park is predicted.The present embodiment is according to wind direction pair
Input data is classified, and when being predicted, first judges the wind direction of input data, wind is obtained in forecast model corresponding to substitution
The predicted value of electric output power, so as to accurately carry out short-term wind-electricity prediction, improve the precision of wind power prediction.
It is pointed out that above-mentioned discrepancy threshold can be configured according to the actual conditions of wind power plant, for screening
The wind direction of sample data.
As an alternative embodiment, the Forecasting Methodology of above-mentioned hybrid wind power generation can also include the steps of:
S106:The sample data of test is obtained, the hybrid wind power generation forecast model established is debugged;
S108:The test result exported according to hybrid wind power generation forecast model, mould is predicted to the hybrid wind power generation
Type is modified.
It should be noted that the sample data of the test is from the wind direction of the wind power plant, wind speed and corresponding wind-powered electricity generation
Sampling obtains in the historical data of power output.
Optionally, may be used also based on any of the above one embodiment, the Forecasting Methodology embodiment of above-mentioned hybrid wind power generation
With including following step:
S110:Using the hybrid wind power generation forecast model, the output production capacity to wind power plant is tested;
S112:When being predicted, the wind direction of input data is judged, according to identified wind direction, wind-force corresponding to calling
Generating forecast model calculates the predicted value of the wind power output power of the wind power plant, test result corresponding to output.
In the present embodiment, difference and feature according to wind direction carry out classification model construction, during prediction, discriminatory analysis to input data
The wind direction of input data, according to identified wind direction, by forecast model corresponding to the substitution of corresponding data, obtain wind-powered electricity generation output
The predicted value of power, so as to realize accurately short-term wind-electricity prediction.
It is then pre- using hybrid wind power generation by first establishing hybrid wind power generation forecast model in the various embodiments described above
Model is surveyed to be predicted wind-power electricity generation according to short-term wind speed, wind direction data.Here, it is pre- for hybrid wind power generation is explained further
Model is surveyed, its forming process is once illustrated below, above-mentioned steps S104 includes following processing procedure:
S1041:Sample weights parameter corresponding to the respectively described specific wind-power electricity generation forecast model configuration of three classes;
S1042:According to the sample weights parameter of each forecast model, the forecast model established according to wind direction is closed
And establish hybrid wind power generation forecast model;
Alternatively, the process for the forecast model for establishing hybrid wind power generation in above-mentioned steps S104 may also include:
S1043:The hybrid wind power generation forecast model is iterated, makes the hybrid wind power generation prediction mould
Type gradually restrains, until output result tends towards stability.
Such as:Reference picture 2, it is the operation principle schematic diagram of Hopfield neural network algorithms, as general nerve net
A kind of derivation of network system, algorithm in itself independently of data outside, different data form different moulds by the processing of algorithm
Type, it is trained by the way that screen three different classes of training set datas are put into Hopfield neutral nets, meeting
Draw the different forecast model of three classes.The only corresponding data set of these models is related, so can be by before it is fuzzy
Other 30% test set data are divided into three classes by Clustering Model by same principle, and each class testing collection data are put into correspondingly
Model in be predicted, just can draw corresponding prediction power output.
If necessary to judge the prediction precision of overall mixed model, then need by the predicted value of three class difference models with it is right
Answer the actual value of sample to make comparisons, obtain its mean square error, and it has just been obtained to the prediction of entirety by certain Weight
Precision.Wherein, the weight of each model predication value mean square error is accounted for the proportion of integrated testability collection data by all kinds of test set data
It is determined that so as to strengthen the fault-tolerant ability of model and generalization ability so that hybrid wind power generation forecast model can be adapted to more
The outer data of sample.In addition, it is necessary to be important to note that, during the prediction of reality, if a new sample point is transfused to model,
Then judge that it for A classes, is then directly predicted by A class models by fuzzy clustering, and obtain a result, the present embodiment passes through power
Weight is integrated to weigh the whole structure of whole model cluster.
Compared with existing wind-power electricity generation forecast model, the hybrid wind power generation forecast model energy of the various embodiments described above use
It is enough to obtain higher precision of prediction.The advantages of Forecasting Methodology embodiment of above-mentioned hybrid wind power generation, is further analysed below
Illustrate, generally, above-described embodiment is mainly using following handling process:
1) modeling sample data are collected
The historical data for obtaining wind park wind direction, wind speed and wind-powered electricity generation output is sampled, and obtains sample data.To whole sample
This statistical property is judged, the wind frequency on different directions is drawn using polar coordinate system, finds out wind frequency most concentration and wind-powered electricity generation is defeated
Go out power difference significantly several directions come establish classification generating forecast model.
Here, because wind-power electricity generation output is not only related to angle, also it is related to wind direction, therefore, by by different wind
Upward particular prediction model merges, and can obtain more accurate wind-force discovery forecast model, i.e. hybrid wind power generation is pre-
Survey model, the hybrid wind power generation forecast model can by distinguishing different wind directions, establish respectively wind speed with output generated output it
Between relation.
2) sample data classification model construction
It is otherwise overlapping in order to establish wind direction, the wind speed model corresponding with power output, it is necessary to which sample is divided
Sample point will influence the independence and accuracy of each model.Therefore, in order to avoid this problem hybrid wind power generation is predicted
Sample is first carried out classification processing by model before model foundation is carried out with fuzzy hierarchy clustering procedure, is beaten to establish accurate model
Basis.
In above-described embodiment, data mining is carried out using fuzzy hierarchy clustering procedure, sample data is divided into three according to wind direction
Class, the height of each of which layer represent the distance between two types data space, distance two kinds of data spaces of bigger expression
Between border it is more obvious, can select border most obvious from result, have minimum uniformity a few class sample groups.For example,
Sample can be divided into two parts, 70% sample is used for training pattern, and 30% sample is used for test model;Several classes that will be selected
Sample group with obvious border is imported in Hopfield algorithms, the model on corresponding different wind directions is respectively trained out, then use
Remaining 30% sample is tested the model of foundation.
Above-described embodiment establishes different forecast models for different classes of sample data, for example with Hopfield god
Specific wind-power electricity generation forecast model is established through network algorithm, passes through study of the Hopfield neural network algorithms to sample characteristics
Come the forecast model set up, the short-term fluctuation of wind-power electricity generation output can not only be reflected, additionally it is possible to overcome other short-term pre-
Survey the weakness of model anti-noise ability difference.
3) hybrid wind power generation forecast model is established
According to the sample weights of each model, forecast model on each wind direction established in back is merged, built
Found a new hybrid prediction model.The sample of no progress fuzzy classification polymerization processing is imported into Hopfield models,
As original predictive model.Iterate hybrid prediction model and original predictive model, model is gradually restrained, finally tends to be steady
It is fixed, both results are subjected to contrast it can be found that the precision of mixed model is higher.
4) it is predicted
When being predicted using hybrid wind power generation forecast model, the wind direction of input data is judged, according to identified wind
The predicted value of the wind power output power of the wind power plant is calculated to, wind-power electricity generation forecast model corresponding to calling, corresponding to output
Test result.
Herein, with reference to an example, the Forecasting Methodology embodiment of above-mentioned hybrid wind power generation is once illustrated:
Reference picture 3, it is the contrast schematic diagram of dynamic wind rose and electric power rose figure, can be preliminary by dynamic wind rose
Judge that the wind frequency in northeast, the southeast and direction northwest is maximum.For wind direction is for wind-force and electrical output power, northeast
The wind-force in direction is larger and the power output of wind-powered electricity generation is also higher;Next to that the wind from northwest and southeast both direction, although
The wind-force in the two directions is generally also larger, but its corresponding power output is smaller.Therefore, it is necessary in modeling by its point
Set the exam worry.So, present problem just come, in training sample some data between the different classification of this three class,
Intuitively effectively it can not be made a distinction.In order to reach the method for quantifying to use fuzzy clustering with objective effect, this example
By training sample according to it:The corresponding relation of wind direction, wind-force and power output is divided into three classes, why selects fuzzy hierarchy here
Cluster be because using in data we have found that not only some sample points it be difficult to differentiate between two kinds of different classifications,
And there is also level association between different other classes, as shown in Figure 4.Clustered by fuzzy hierarchy, it can be determined that each sample
Which kind of this point belongs to actually, and the correlation degree between class and class how, so original training sample is just effective
It divide into different classifications.As shown in figure 5, the difference of these sample point categories is put forward, three groups of new samples are just obtained,
And a kind of specific, rational wind-powered electricity generation input/output relation is all correspond to per a kind of sample data, so can by this three
Group data establish regression model, obtain the wind-powered electricity generation forecast model based on different wind speed and directions.
Reference picture 6, it is the prediction result pair of hybrid wind power generation forecast model and the archetype before mixed processing
Compare schematic diagram.In this example, as a result the model established respectively on different wind directions with Hopfield models is shown in northeastward
Wind-force prediction error it is minimum, and the error in direction northwest is maximum, is finally hybrid prediction model by three model combinations.With
Archetype is contrasted (result is as shown in Figure 6):The mean square deviation of hybrid wind power generation forecast model is more square than archetype
Difference is small, and the former is 18.32, and the latter 35.1;The R side of hybrid wind power generation forecast model is 0.73, compared to archetype
0.7 closer to 1, that is to say, that independent variable can preferably explain dependent variable in hybrid wind power generation forecast model.This
Outside, although the run time of hybrid prediction model is slightly longer, its difference is only 0.38 second, has no effect on practical operation.
Therefore from prediction result, hybrid wind power generation forecast model can obtain higher precision, and can apply to reality
Operation.
To sum up, the mixing wind-powered electricity generation based on fuzzy hierarchy cluster and Hopfield neutral nets described in the various embodiments described above is pre-
The advantages of surveying model is:During being predicted to wind power output power, while consider two aspects of wind direction and wind-force
Influence, and according to the corresponding relation of wind direction, wind-force and wind power output power, carried out fuzzy hierarchy cluster.So, can incite somebody to action
One group of sampled data for having noise redundancy is divided into the corresponding several parts of sample datas that internal association be present, then to these sample numbers
According to being modeled respectively, the internal characteristicses of different classes of data just can be fully demonstrated by different models, so as to targetedly
Carry out model prediction, the precision of prediction also thus increase.
It should be noted that for foregoing embodiment of the method, in order to be briefly described, therefore it is all expressed as a series of
Combination of actions, but those skilled in the art should know, the present invention is not limited by described sequence of movement, because according to
According to the present invention, some steps can use other orders or carry out simultaneously.Secondly, those skilled in the art should also know,
Embodiment described in this description belongs to preferred embodiment, and involved action is not necessarily essential to the invention.
System embodiment
To realize the above method, the present embodiment proposes a kind of forecasting system of hybrid wind power generation, as shown in Figure 7 and Figure 8,
The forecasting system embodiment of the hybrid wind power generation includes:Data input device 101, it is configured with hybrid wind power generation forecast model
Test device 102 and output device 103.
In the forecasting system of the hybrid wind power generation, data input device 101 is used to obtain test data;Test device
102 are used for the test data that is inputted according to data input device 101, judge the wind direction of input data, according to identified wind direction,
Wind-power electricity generation forecast model corresponding with wind direction in the hybrid wind power generation forecast model 1021 is called, to wind power plant
Output production capacity is tested;Output device 103 is used to obtain the wind that the hybrid wind power generation forecast model is tested to obtain
The predicted value of the wind power output power of electric field, and test result corresponding to output.
In above-described embodiment, the forecasting system of the hybrid wind power generation may also include:Data modeling and correcting device 104,
For establishing hybrid wind power generation forecast model 1021 and amendment or optimization hybrid wind power generation forecast model 1021, therefore,
Data modeling and correcting device 104 further comprise consisting of:
1) decimation blocks 201, for going through to the wind direction of acquired wind power plant, wind speed and corresponding wind power output power
History data are sampled to obtain sample data;
2) analysis module 202, for carrying out discriminatory analysis to the statistical property of the sample data, obtain wind frequency concentrate and
Wind power output power difference reaches the wind direction of discrepancy threshold;
3) processing module 203, for being adopted according to acquired wind direction and its with the corresponding relation of wind speed, wind power output power
With fuzzy hierarchy clustering procedure, the sample data is divided into three classes;
4) model just models block 204, for being trained using neural network algorithm to every class sample, is correspondingly formed three classes
Specific wind-power electricity generation forecast model;
5) mixed processing module 205, for the specific wind-power electricity generation forecast model of this three class to be merged into processing, formed
For the hybrid wind power generation forecast model being predicted to wind-power electricity generation production capacity.
As an alternative embodiment, in above-described embodiment, the forecasting system of hybrid wind power generation may also include:
6) correcting module 206 is tested, for obtaining the sample data of test, the hybrid wind power generation established is predicted
Model is debugged, corrected.
For further, in an alternative embodiment, above-mentioned mixed processing module 205 may also include consisting of:
Parameter configuration unit 51, for being respectively sample corresponding to the specific wind-power electricity generation forecast model configuration of three class
Weight parameter;
Merging treatment unit 52, it is pre- by being established according to wind direction for the sample weights parameter according to each forecast model
Survey model to merge, establish hybrid wind power generation forecast model;
Iterative optimization unit 53, for being iterated to the hybrid wind power generation forecast model, make the mixing
Wind-power electricity generation forecast model is gradually restrained, until output result tends towards stability.
As an alternative embodiment, above-mentioned hybrid wind power generation forecast model 1021 further comprises consisting of
Module:
Wind direction recognition unit, for judging the wind direction of input data;
Model call unit, for the wind direction determined by, call corresponding in the hybrid wind power generation forecast model
Wind-power electricity generation forecast model, the output production capacity to wind power plant tests.
Obviously, those skilled in the art should be understood that the prediction of the hybrid wind power generation of the above-mentioned embodiment of the present invention
Each module of system or each step of Forecasting Methodology of hybrid wind power generation can realize that they can collect with general computing device
In on single computing device, or be distributed on the network that multiple computing devices are formed, alternatively, they can use tricks
The executable program code of device is calculated to realize, it is thus possible to be stored in storage device by computing device to perform,
They are either fabricated to each integrated circuit modules respectively or the multiple modules or step in them are fabricated to single collection
Realized into circuit module.So, the present invention is not restricted to any specific hardware and software combination.The storage device is non-
Volatile memory, such as:ROM/RAM, flash memory, magnetic disc, CD etc..
The foregoing is only embodiments of the invention, be not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent substitution and improvements made etc., it should be included in the scope of the protection.
Claims (10)
- A kind of 1. Forecasting Methodology of hybrid wind power generation, it is characterised in that including:The historical data of the wind direction of wind power plant, wind speed and corresponding wind power output power is obtained, and the historical data is carried out Sampling obtains sample data;Discriminatory analysis is carried out to the statistical property of the sample data, wind frequency concentration is obtained and wind power output power difference reaches poor The wind direction of different threshold value, and use fuzzy hierarchy according to acquired wind direction and its with the corresponding relation of wind speed, wind power output power Clustering procedure, the sample data is divided into three classes;Every class sample is trained using neural network algorithm, is correspondingly formed the specific wind-power electricity generation forecast model of three classes, will The specific wind-power electricity generation forecast model of this three class merges processing, establishes the mixing for being predicted to wind-power electricity generation production capacity Wind-power electricity generation forecast model.
- 2. the Forecasting Methodology of hybrid wind power generation according to claim 1, it is characterised in that also include:The sample data of test is obtained, the hybrid wind power generation forecast model established is debugged;The test result exported according to hybrid wind power generation forecast model, is repaiied to the hybrid wind power generation forecast model Just;Wherein, the sample data of the test going through from the wind direction of the wind power plant, wind speed and corresponding wind power output power Sampling obtains in history data.
- 3. the Forecasting Methodology of hybrid wind power generation according to claim 1, it is characterised in that also include:Using the hybrid wind power generation forecast model, the output production capacity to wind power plant is tested;When being predicted, the wind direction of input data is judged, according to identified wind direction, wind-power electricity generation corresponding to calling predicts mould Type calculates the predicted value of the wind power output power of the wind power plant, test result corresponding to output.
- 4. the Forecasting Methodology of hybrid wind power generation according to claim 1, it is characterised in that the hybrid wind power generation is pre- Surveying the generation type of model also includes:Sample weights parameter corresponding to the respectively described specific wind-power electricity generation forecast model configuration of three classes;According to the sample weights parameter of each forecast model, the forecast model established according to wind direction is merged, establishes mixing Wind-power electricity generation forecast model.
- 5. the Forecasting Methodology of the hybrid wind power generation according to any one of Claims 1-4, it is characterised in that this method is also Including:The hybrid wind power generation forecast model is iterated, the hybrid wind power generation forecast model is gradually received Hold back, until output result tends towards stability.
- A kind of 6. forecasting system of hybrid wind power generation, it is characterised in that including:Decimation blocks, for entering to the historical data of the wind direction of acquired wind power plant, wind speed and corresponding wind power output power Line sampling obtains sample data;Analysis module, for carrying out discriminatory analysis to the statistical property of the sample data, obtain wind frequency concentration and wind-powered electricity generation exports Power difference reaches the wind direction of discrepancy threshold;Processing module, for using obscuring layer according to acquired wind direction and its with the corresponding relation of wind speed, wind power output power Secondary clustering procedure, the sample data is divided into three classes;Model just models block, for being trained using neural network algorithm to every class sample, is correspondingly formed the specific wind of three classes Power generating forecast model;Mixed processing module, for the specific wind-power electricity generation forecast model of this three class to be merged into processing, formed for wind The hybrid wind power generation forecast model that power generating production capacity is predicted.
- 7. the forecasting system of hybrid wind power generation according to claim 6, it is characterised in that the system also includes:Correcting module is tested, for obtaining the sample data of test, the hybrid wind power generation forecast model established is carried out Debugging, amendment and optimization.
- 8. the forecasting system of hybrid wind power generation according to claim 6, it is characterised in that the system also includes:Data input device, for obtaining test data;Test device, the hybrid wind power generation forecast model is configured with, for according to acquired test data, judging to input The wind direction of data, according to identified wind direction, call corresponding wind-power electricity generation prediction in the hybrid wind power generation forecast model Model, the output production capacity to wind power plant are tested;Output device, for the wind-powered electricity generation output work for the wind power plant for obtaining the hybrid wind power generation forecast model to test to obtain The predicted value of rate, and test result corresponding to output.
- 9. the forecasting system of the hybrid wind power generation according to any one of claim 6 to 8, it is characterised in that the mixing Processing module also includes:Parameter configuration unit, for being respectively sample weights ginseng corresponding to the specific wind-power electricity generation forecast model configuration of three class Number;Merging treatment unit, for the sample weights parameter according to each forecast model, the forecast model that will be established according to wind direction Merge, establish hybrid wind power generation forecast model;Iterative optimization unit, for being iterated to the hybrid wind power generation forecast model, send out the mixing wind-force Electric forecast model is gradually restrained, until tending towards stability.
- 10. the forecasting system of hybrid wind power generation according to claim 9, it is characterised in that the hybrid wind power generation Forecast model further comprises:Wind direction recognition unit, for judging the wind direction of input data;Model call unit, for the wind direction determined by, call corresponding wind in the hybrid wind power generation forecast model Power generating forecast model, the output production capacity to wind power plant are tested.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610300971.7A CN105930900B (en) | 2016-05-09 | 2016-05-09 | The Forecasting Methodology and system of a kind of hybrid wind power generation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610300971.7A CN105930900B (en) | 2016-05-09 | 2016-05-09 | The Forecasting Methodology and system of a kind of hybrid wind power generation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105930900A CN105930900A (en) | 2016-09-07 |
CN105930900B true CN105930900B (en) | 2018-02-27 |
Family
ID=56835381
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610300971.7A Active CN105930900B (en) | 2016-05-09 | 2016-05-09 | The Forecasting Methodology and system of a kind of hybrid wind power generation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105930900B (en) |
Families Citing this family (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106529741B (en) * | 2016-12-12 | 2019-12-06 | 国网电力科学研究院 | Ultra-short-term wind power prediction method based on spatial correlation characteristics |
CN108596415B (en) * | 2017-12-15 | 2023-11-24 | 创新先进技术有限公司 | Model integration method and device |
CN110276468A (en) * | 2018-03-14 | 2019-09-24 | 北京金风科创风电设备有限公司 | Method and device for predicting generated power of wind generating set |
CN108898251B (en) * | 2018-06-29 | 2021-10-08 | 上海电力学院 | Offshore wind farm power prediction method considering meteorological similarity and power fluctuation |
CN109597901B (en) * | 2018-11-15 | 2021-11-16 | 韶关学院 | Data analysis method based on biological data |
CN111353629B (en) * | 2018-12-24 | 2023-09-29 | 顺丰科技有限公司 | Frequency updating method and device |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102184453A (en) * | 2011-05-16 | 2011-09-14 | 上海电气集团股份有限公司 | Wind power combination predicting method based on fuzzy neural network and support vector machine |
KR20150082804A (en) * | 2014-01-08 | 2015-07-16 | 김상훈 | Vertical wind turbines |
JP2015203387A (en) * | 2014-04-16 | 2015-11-16 | 株式会社日立製作所 | Wind turbine generator system |
CN105389634A (en) * | 2015-12-01 | 2016-03-09 | 广东智造能源科技研究有限公司 | Combined short-term wind power prediction system and method |
-
2016
- 2016-05-09 CN CN201610300971.7A patent/CN105930900B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102184453A (en) * | 2011-05-16 | 2011-09-14 | 上海电气集团股份有限公司 | Wind power combination predicting method based on fuzzy neural network and support vector machine |
KR20150082804A (en) * | 2014-01-08 | 2015-07-16 | 김상훈 | Vertical wind turbines |
JP2015203387A (en) * | 2014-04-16 | 2015-11-16 | 株式会社日立製作所 | Wind turbine generator system |
CN105389634A (en) * | 2015-12-01 | 2016-03-09 | 广东智造能源科技研究有限公司 | Combined short-term wind power prediction system and method |
Non-Patent Citations (2)
Title |
---|
"短期风速和风电功率预测模型的研究";方江晓;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20110915;第C042-213页 * |
"风电场输出功率组合算法的研究";吴丹丹;《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》;20131215;第C042-533页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105930900A (en) | 2016-09-07 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105930900B (en) | The Forecasting Methodology and system of a kind of hybrid wind power generation | |
CN109614997A (en) | A kind of stealing Risk Forecast Method and device based on deep learning | |
CN108375808A (en) | Dense fog forecasting procedures of the NRIET based on machine learning | |
CN112069567A (en) | Method for predicting compressive strength of concrete based on random forest and intelligent algorithm | |
CN106021771A (en) | Method and device for diagnosing faults | |
CN111369070A (en) | Envelope clustering-based multimode fusion photovoltaic power prediction method | |
CN104408562A (en) | Photovoltaic system generating efficiency comprehensive evaluation method based on BP (back propagation) neural network | |
CN108428017A (en) | Wind power interval prediction method based on core extreme learning machine quantile estimate | |
CN113901977A (en) | Deep learning-based power consumer electricity stealing identification method and system | |
CN106295911A (en) | A kind of grid branch parameter evaluation method based on chromatographic assays | |
CN108921359A (en) | A kind of distribution gas density prediction technique and device | |
CN113516305B (en) | Scene-target interaction water network regional water resource scheduling intelligent decision method and system | |
CN112990330B (en) | User energy abnormal data detection method and device | |
CN104112062B (en) | The acquisition methods of wind-resources distribution based on interpolation method | |
CN112053005A (en) | Machine learning fusion method for subjective and objective rainfall forecast | |
CN108269197A (en) | Wind turbines power characteristic appraisal procedure and device | |
CN112784491A (en) | Urban charging grid point planning method facing high-elasticity power grid based on LSTM and IQPSO | |
CN116432123A (en) | Electric energy meter fault early warning method based on CART decision tree algorithm | |
CN115526258A (en) | Power system transient stability evaluation method based on Spearman correlation coefficient feature extraction | |
CN115100819A (en) | Landslide hazard early warning method and device based on big data analysis and electronic equipment | |
CN113011455B (en) | Air quality prediction SVM model construction method | |
CN116842459B (en) | Electric energy metering fault diagnosis method and diagnosis terminal based on small sample learning | |
CN109508820A (en) | Campus electricity demand forecasting modeling method based on differentiation modeling | |
CN108988347A (en) | A kind of adjusting method and system that power grid Transient Voltage Stability sample set classification is unbalance | |
CN110807014B (en) | Cross validation based station data anomaly discrimination method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
CB03 | Change of inventor or designer information |
Inventor after: Song Xiaohua Inventor after: Zhang Yulin Inventor after: Long Yun Inventor after: Zhang Xubei Inventor after: Li Leming Inventor before: Song Xiaohua Inventor before: Zhang Yulin Inventor before: Li Leming |
|
COR | Change of bibliographic data | ||
GR01 | Patent grant | ||
GR01 | Patent grant |