CN106503867A - A kind of genetic algorithm least square wind power forecasting method - Google Patents
A kind of genetic algorithm least square wind power forecasting method Download PDFInfo
- Publication number
- CN106503867A CN106503867A CN201611003309.1A CN201611003309A CN106503867A CN 106503867 A CN106503867 A CN 106503867A CN 201611003309 A CN201611003309 A CN 201611003309A CN 106503867 A CN106503867 A CN 106503867A
- Authority
- CN
- China
- Prior art keywords
- genetic algorithm
- vector machine
- data
- supporting vector
- square method
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000002068 genetic effect Effects 0.000 title claims abstract description 37
- 238000013277 forecasting method Methods 0.000 title claims abstract description 8
- 239000013598 vector Substances 0.000 claims abstract description 39
- 238000000034 method Methods 0.000 claims abstract description 29
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000005457 optimization Methods 0.000 claims abstract description 15
- 238000012360 testing method Methods 0.000 claims abstract description 12
- 238000005259 measurement Methods 0.000 claims abstract description 4
- 238000009434 installation Methods 0.000 claims abstract description 3
- 238000012843 least square support vector machine Methods 0.000 claims description 2
- 239000003086 colorant Substances 0.000 abstract description 4
- 230000006870 function Effects 0.000 description 30
- 238000012706 support-vector machine Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 7
- 230000008901 benefit Effects 0.000 description 4
- 230000005611 electricity Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000000052 comparative effect Effects 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 230000007935 neutral effect Effects 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000005684 electric field Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000008303 genetic mechanism Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000003278 mimic effect Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 108090000623 proteins and genes Proteins 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Public Health (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of genetic algorithm least square wind power forecasting method, sets up genetic algorithm least square method supporting vector machine forecast model using actual measurement wind speed is collected, determines modeling input used, output variable；Initial data is normalized, using the sample data of the data of genetic algorithm optimization parameter, the training of least square method supporting vector machine forecast model and test；To genetic algorithm and least square method supporting vector machine prediction model parameterses Initialize installation, training pattern, evolved by genetic algorithm many generations and obtain the least square method supporting vector machine prediction model parameterses for optimizing, set up least square method supporting vector machine forecast model；Wind speed short-term forecast is done to test sample with least square method supporting vector machine forecast model.The present invention carries out parameter optimization with genetic algorithm to LSSVM models, establishes the wind speed information forecast model based on GA LSSVM, can realize the accurate prediction of data with flying colors.
Description
Technical field
The invention belongs to technical field of wind power, and in particular to a kind of genetic algorithm least square wind power forecasting method.
Background technology
Wind power prediction/wind farm power prediction WPP (Wind Power Prediction) (also has some domestic specialties
Magazine is referred to as Wind Energy Prediction) wind power prediction refers to wind energy turbine set wind turbine power generation power prediction.
Existing wind-powered electricity generation Predicting Technique major part is modified using numerical weather forecast, for historical data is not using
Foot, the accuracy of heavy dependence numerical weather forecast.
Content of the invention
The purpose of the present invention is to propose to a kind of method of prediction wind-powered electricity generation ultra-short term power, with genetic algorithm to least square
Vector machine (LSSVM) model carries out parameter optimization, establishes the wind based on GA-LSSVM (genetic algorithm least square support vector machines)
Fast information prediction model, can realize the accurate prediction of data with flying colors, show very strong in terms of prediction wind farm wind velocity
Superiority.
The purpose of the present invention is achieved through the following technical solutions：
A kind of genetic algorithm least square wind power forecasting method, sets up genetic algorithm using actual measurement wind speed is collected
Least square method supporting vector machine forecast model, comprises the following steps that：
Step one, determination modeling input used, output variable：An air speed data, one day institute were gathered every 10 minutes
There are data for one group, six groups of data are a cycle, the wind speed of continuous 5 days used as training sample, surveyed by conduct in 1 day afterwards in the past
Examination；
Step 2, initial data is normalized, using the data of genetic algorithm optimization parameter, least square
Hold the sample data of vector machine forecast model training and test；
Step 3, to genetic algorithm and least square method supporting vector machine prediction model parameterses Initialize installation：Using adopting
The data that collection comes, carry out binary coding, produce just for population, i.e., initial least square method supporting vector machine model, then train
Model, is evolved by genetic algorithm many generations and obtains the least square method supporting vector machine prediction model parameterses for optimizing, set up a most young waiter in a wineshop or an inn
Take advantage of SVM prediction model；
Step 4, that the least square method supporting vector machine forecast model obtained with step 3 does wind speed short-term to test sample is pre-
Survey；
Step 5, the precision for being verified acquired results by the fitness function for setting, if undesirable, are set again
Determine genetic algorithm parameter, three re -training of return to step.
The present invention carries out parameter optimization with genetic algorithm to LSSVM models, establishes the wind speed based on GA-LSSVM and believes
Breath forecast model, by simulation analysis：Predicting the outcome for GA-LSSVM models is better than conventional RBFNN models, and logical
The contrast of precision and error is crossed, is absolutely proved that GA-LSSVM models are effective and feasible, the accurate of data can be realized with flying colors
Prediction, while also show machine learning algorithm shows very strong superiority in terms of prediction wind farm wind velocity.
Description of the drawings
Fig. 1 is method of the present invention flow chart
General principle figures of the Fig. 2 for SVMs
Fig. 3 is that SVMs is vectorial through kernel function transition diagram
Fig. 4 is SVMs two-dimensional space example
Fig. 5 is SVMs insensitive loss function and slack variable
Fig. 6 is LSSVM model error handling principle figures
Fig. 7 is genetic algorithm optimization flow chart
Fig. 8 is the predicting wind speed of wind farm Comparative result schematic diagram using GA-LSSVM models
Fig. 9 is the predicting wind speed of wind farm result relative error schematic diagram using GA-LSSVM models
Figure 10 is using RBFNN model predicting wind speed of wind farm Comparative result schematic diagrames
Figure 11 is using RBFNN model predicting wind speed of wind farm result relative error schematic diagrames
Figure 12 is predicting wind speed of wind farm Comparative result schematic diagram of another anemometer tower using GA-LSSVM models
Figure 13 is predicting wind speed of wind farm resultant error schematic diagram of another anemometer tower using GA-LSSVM models
Specific embodiment
Technical scheme is discussed in detail below in conjunction with accompanying drawing：
Principle background
SVMs (Support Vector Machine, SVM) is a kind of machine learning method, and he is by training sample
Notebook data, obtains corresponding forecast model.SVM is structural risk minimization relative to the basic advantage of other Forecasting Methodologies, and
And for sample is little, the high data processing prediction of dimension has the advantage of oneself.Just because of having the above advantage, a lot of experts
The application mode of SVM is being studied, and therefore SVM is rapidly progressed.SVM can be used to predict that precision of prediction is just depended on
Ruleization constant C (representing the degree of mistake) and slack variable ξt, C and ξtChange with the different of input data, directly affect
Precision of prediction.How C and ξ is so affectedtValue, be exactly various algorithm improvement directions.In order that SVM is advantageously in meter
Calculate, have scholar that least square correlation principle is incorporated in SVMs, least square method supporting vector machine (Least
Squares Support Vector Machine, LSSVM) it is developed.LSSVM is the least square based on SVM standards
Formula.LSSVM has two features, and one is through theory deduction by the inequality computing of SVMs, is converted into equation computing,
There to be detailed derivation later；Two is when supporting vector is found, and only focuses on those nonzero informations.Obviously, have above
Two kinds of changes, the calculating process of LSSVM are simplified, and efficiency is greatly improved.
Based on the thought of statistical learning, SVM is used only for sort research.SVM can pass through to solve a convex double optimization
Problem, the recurrence for being applied to nonlinear function are calculated.Fig. 2 is the general principle figure of a SVMs.In form, with
One feed forward type neutral net has a lot of similarities, the output layer of the same input layer with information and information.The area of the two
It is not neuron that SVM kernel functions instead of neutral net.The operation principle of the two is also not quite similar, one 3 layers of feedforward
Two transmission functions of ANN input and output.But SVM only one of which kernel functions, the effect of the kernel function is by low-dimensional input data
Vector space (be sometimes Infinite-dimensional) of the vector transformation to more higher-dimension.The vectorial process for going to high latitude space through kernel function is such as
Shown in Fig. 3.After kernel function conversion, SVM may be selected by the optimized algorithm (such as quadratic programming) of some classes return executing or
Classified calculating.
As shown in figure 4, the vector in grey box is supporting vector, the linear separation vector of SVM and support in two-dimensional space
Difference between vector.Supporting vector is exactly the vector that can determine minimum safe distance.Can also as seen from the figure, support to
Amount determines optimal classification interval.
Assume { (Xt,yt) (t=1,2 ..., it is n) a given group data set, wherein Xt=(xt1,xt2,…,xtk) be
There is the input vector of k variable, ytIt is the output data of corresponding t, can be defined as：
Wherein,<,>Dot product is represented, W is weight vectors, and b is biasing,It is input vector XtIt is transformed into higher-dimension
The mapping function of feature space.Then corresponding optimization problem can be obtained：
Wherein, C is regular constant, represents the degree of mistake.ξtAnd ξt* it is slack variable, for weighing as shown in Figure 5
Training points above and below desired value error.Width is that the ε-insensitive loss function of 2 ε is defined as：
By introducing Lagrange multiplier, the problem in formula (1.2) is changed into：
Wherein, αt,η t andIt is Lagrange multiplier.Lagrangian is respectively to W, b, ξtWithCarry out partially micro-
Point calculate and by result zero setting, i.e.,：
To in former Lagrangian formula (1.4), optimization problem is converted into following problem to substitution formula (1.5)：
According to KKT (Karush-Kuhn-Tucker) condition, the continuous item comprising Lagrange's multiplier can in optimal solution
To eliminate.This means that equation below is set up：
Formula (1.7) shows, for all of sample, its Lagrange multiplier is considered non-supporting vector equal to zero, and is
Number non-zero is regarded as supporting vector.Meanwhile, when slack variable ξtWithWhen being zero, the value of b can be obtained, i.e.,：
Finally, the nonlinear function estimation formulas of SVM can be write as：
Formula (1.9) can be written to：
Wherein,It is called kernel function, conventional kernel function such as formula (1.11)～(1.14).
(1) linear kernel function：
K(X,Xt)=＜ X, Xt＞ ... ... ... ... ... ... (1.11)
(2) Polynomial kernel function：
K(X,Xt)=(＜ X, Xt＞+p)d, d ∈ N, p ＞ 0 ... ... ... ... (1.12)
(3) gaussian kernel function：
(4) S types kernel function (Sigmoid)：
K(X,Xt)=tanh (c ＜ X, Xt＞+d), c ＞ 0, d ＞ 0 ... ... (1.14)
Gaussian kernel function is one of most powerful nonlinear function estimation.For SVM[39,40]Classified calculating, decision-making letter
Number equation (1.10) is by output category result, rather than regression result.Multiclass classification problem can be viewed as having multiple predetermined thresholds
The recurrence of value is calculated.SVM also has because the limitation of the optimized algorithm of its own when doing and classifying and return calculating
Shortcoming.
LSSVM and SVM have same structure.It has input layer and containing the defeated of single or multiple input/output datas
Go out layer, hidden layer includes the kernel that low-dimensional input data is transformed into higher-dimension, and input vector is characterized through the kernel conversion, change
The false high-dimensional vector in space, so as to have the separability of high-dimensional space vector, has counting for low dimensional space again
The property calculated, but after conversion, the operation principle of LSSVM is different from SVM.It is different from the substandard inequality constraints of SVM, LSSVM
It is the thought based on equality constraint.Make original problem linear from the problem of quadratic programming be converted into linear KKT systems one group
The problem of equation.The recurrence of LSSVM is calculated and was suggested in 2002 later, its thought be using all of training data as
Hold vector.
By formula (1.1), for the corresponding optimization problems of LSSVM are：
Wherein, etIt is the error variance of t, γ is adjustable constant, similar to the C of SVM.
Obtain Lagrangian as follows：
Wherein, αtIt is Lagrange multiplier, Lagrangian is to original variable W, b, etAnd αtPartial derivative can be by formula
(1.17) obtain.
Obtain Lagrangian as follows：
Wherein, αtIt is Lagrange multiplier, Lagrangian is to original variable W, b, etAnd αtPartial derivative can be by formula
(1.17) obtain.
It is replaced by formula (1.17), equivalence formula can be obtained：
Another kind of equation of formula (1.18) is：
Wherein,Y=[y1,…,yn] T, 1N=[1 ...,
1] T, α=[α1,α2,…,αn]T.
Thus we can obtain following LSSVM forecast models：
Wherein, α and b can be obtained by formula (1.19), kernel function K (X, Xt) adopt gaussian kernel function.It can be seen that error is punished
Penalty parameter γ is the important parameter of the precision for affecting LSSVM.
Genetic algorithm (Genetic Algorithms, GA) also becomes evolution algorithm, is mimic biology genetic mechanism and reaches
A kind of evolutional heuristic search algorithm method of that text.The theory of biological evolution principle of " survival of the fittest in natural selection " is drawn by it
Enter to find in the coded strings group body that optimized parameter is formed, individuality is screened by selected fitness function, make adaptation
The high individuality of degree is retained, by heredity in duplication, the new colony of intersection and variation composition, new colony both inherited
The information of previous generation, the possibility of the high hereditary offspring of fitness are big, and the low meeting of fitness is progressively eliminated.So constantly repeat
Fitness screening is carried out to new population, the high individual amount of fitness is more and more in colony, set in advance until meeting
Condition, algorithm terminate, and at this moment, the probability highest that fitness highest individuality is stayed in population, so that obtain optimal solution.Heredity is calculated
Method is conducive to computer disposal, and can arrive globally optimal solution.
Technical solution of the present invention
The present invention provides a kind of genetic algorithm least square wind power forecasting method, as shown in figure 1, being received using system
Collect actual measurement wind speed and set up genetic algorithm least square method supporting vector machine (GA-LSSVM) forecast model, comprise the following steps that：
Step one, determination modeling input used, output variable：An air speed data, one day institute were gathered every 10 minutes
There are data for one group, six groups of data are a cycle, the wind speed of continuous 5 days used as training sample, surveyed by conduct in 1 day afterwards in the past
Examination.
Step 2, initial data is normalized, in order to be calculated.Using genetic algorithm optimization parameter
The sample data of data, LSSVM training and test.
Step 3, GA algorithms and LSSVM parameter initializations are arranged.Using collection come data, carry out binary system volume
Code, produces just for population, i.e., initial LSSVM models, and then training pattern, is evolved by GA many generations and obtain the LSSVM for optimizing ginsengs
Number, sets up LSSVM forecast models.
Wind speed short-term forecast made by step 4, the model obtained with step 3 to test sample.
Step 5, the precision for being verified acquired results by the fitness function for setting, if undesirable, are set again
Determine GA parameters, three re -training of return to step.Meet precision, or reach operation times set in advance, then it is assumed that find most
Excellent parameter, substitutes in LSSVM and is predicted computing.
The present invention selects gaussian kernel function as kernel function, thus have two important parameters it needs to be determined that, i.e. LSSVM moulds
σ in the middle of type in error punishment parameter γ and kernel function, will optimize the two parameters below using GA.
1. initial population 20,100 generation of genetic algebra；
2. LSSVM parameters will be optimized using GA, first have to binary coding be carried out to γ and σ.The two common group
Into the binary coding of 20, genetic algorithm computing is participated in.
3. determine fitness function, determine to obtain whether result meets optimum.
Wind speed information prediction and evaluation standard
Wind energy turbine set finally can electricity volume fine or not closely coupled with what output of wind electric field predicted the outcome, Forecasting Methodology quality
Judgement, the height of precision is evaluated by the evaluation criterion that predicts.It is listed below main error judgment formula：
1. mean absolute percentage error (Mean Absolute Percent Error, MAPE)
2. root-mean-square error (Root Mean Square Error, RMSE)
3. mean absolute error (Mean Absolute Error, MAE)
4. relative error (Relative Error)
5. impartial coefficient (EC)
In order to preferably show the similarity degree predicted the outcome with actual output of wind electric field, we define impartial coefficient
(abbreviation EC below).So-called impartial coefficient, from definition as can be seen that expression be predict the outcome similar to actual result
Degree, is a kind of simple method for judging prediction order of accuarcy, from defined formula as can be seen that impartial coefficient is naturally larger than 0
It is less than 1.EC values are bigger, represent and predict the outcome closer to actual conditions.If generally it is believed that EC>0.85, it is possible to see
Work is preferably prediction, if EC>0.9 prediction for being taken as satisfaction.
The quality that predicts the outcome of GA-LSSVM has many indexs, and we choose relative error (Relative Error) here
As the fitness function of checking gene quality, in formula：yiWithThe real-valued and predicted value of respectively sample；N is test specimens
This number present invention carrys out the quality of evaluation algorithm from multiple errors.
Simulation analysis
By above deriving analysis, the predicting wind speed of wind farm model of GA-LSSVM is established, in order to verify model
Validity, carries out contrast test with RBFNN.Test data comes from the measured data of Jilin certain wind-power electricity generation enterprise inside the province,
Time is the September And October of 2015, frequency acquisition be 10 minutes once, have chosen ten distant ventilation measuring point data,
Data processing is also carried out in the algorithm in the lump, i.e., when wind speed is more than 12m/s, model is defaulted as 12m/s, and this is in experiment below
Embody in output.This chapter emulation data use the data in 5 days on the -5th on the 1st October as the training sample of LSSVM models, use
The constructed good LSSVM model predictions wind speed information of the 6th day.GA parts carry out cross validation's optimization to parameter, initial kind
The size of group is set as 20, and maximum iterations is set as 100.
Fig. 8, Fig. 9 are the predicting wind speed of wind farm results using GA-LSSVM models, and give genetic algorithm searching
Optimal value of the parameter.
Fig. 8 is the wind farm power prediction comparing result using GA-LSSVM models, parameter γ of the LSSVM that GA optimizes=
0.7162 and σ=0.0643.Fig. 9 is the relative error of each point, and histogram gives the direct feel of error, it can be seen that by mistake
Difference all very littles, average relative error is 10.1%, and maximum relative error is 48%, and degree of fitting is 94.1%.With in the world 5%
Also there is a big difference for error, but improves compared with neutral net a lot.
In order to contrast GA-LSSVM,, as training sample the equally applicable October No. 1-5, the 6th day used as test specimens for we
This, such as Figure 10, Figure 11 average relative error is 13.5, and maximum relative error is 51%, and degree of fitting is 92.4%.As can be seen that
GA-LSSVM improves precision than RBFNN really.
Figure 12, Figure 13 remain the wind farm power prediction comparing result using GA-LSSVM models, and only data are adopted
With the data of another anemometer tower collection of contemporaneity, it can be seen that result is very different, the parameter of the LSSVM that GA optimizes
γ=1.291 and σ=0.12.Mean error has reached 16%, and degree of fitting is 92%.
Parameter optimization is carried out to LSSVM models with genetic algorithm, the wind speed information based on GA-LSSVM is established and is predicted
Model, by simulation analysis：Predicting the outcome for GA-LSSVM models is better than conventional RBFNN models, and passes through precision
With the contrast of error, absolutely prove that GA-LSSVM models proposed by the present invention are effective and feasible, data can be realized with flying colors
Accurate prediction.Also show machine learning algorithm simultaneously very strong superiority is shown in terms of prediction wind farm wind velocity.
Claims (2)
1. a kind of genetic algorithm least square wind power forecasting method, it is characterised in that built using actual measurement wind speed is collected
Vertical genetic algorithm least square method supporting vector machine forecast model, comprises the following steps that：
Step one, determination modeling input used, output variable：An air speed data, one day all number were gathered every 10 minutes
According to for one group, six groups of data are a cycle, and the wind speed of continuous 5 days used as training sample, tested by conduct in 1 day afterwards in the past；
Step 2, initial data is normalized, using the data of genetic algorithm optimization parameter, least square support to
The training of amount machine forecast model and the sample data of test；
Step 3, to genetic algorithm and least square method supporting vector machine prediction model parameterses Initialize installation：Using collection come
Data, carry out binary coding, produce just for population, i.e., initial least square method supporting vector machine model, then training pattern,
Evolved by genetic algorithm many generations and obtain the least square method supporting vector machine prediction model parameterses for optimizing, set up least square support
Vector machine forecast model；
Step 4, the least square method supporting vector machine forecast model obtained with step 3 do wind speed short-term forecast to test sample；
Step 5, the precision for verifying acquired results by the fitness function for setting, if undesirable, reset something lost
Propagation algorithm parameter, three re -training of return to step.
2. a kind of genetic algorithm least square wind power forecasting method as claimed in claim 1, it is characterised in that the step
In rapid three, select gaussian kernel function as kernel function, worked as using genetic algorithm optimization least square method supporting vector machine forecast model
σ in middle error punishment parameter γ and kernel function：
1) initial population 20,100 generation of genetic algebra；
2) binary coding is carried out to γ and σ, the two collectively constitutes the binary coding of 20, participates in genetic algorithm computing；
3) determine fitness function, determine to obtain whether result meets optimum.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611003309.1A CN106503867A (en) | 2016-11-14 | 2016-11-14 | A kind of genetic algorithm least square wind power forecasting method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611003309.1A CN106503867A (en) | 2016-11-14 | 2016-11-14 | A kind of genetic algorithm least square wind power forecasting method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106503867A true CN106503867A (en) | 2017-03-15 |
Family
ID=58324589
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611003309.1A Pending CN106503867A (en) | 2016-11-14 | 2016-11-14 | A kind of genetic algorithm least square wind power forecasting method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106503867A (en) |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106960260A (en) * | 2017-03-27 | 2017-07-18 | 深圳汇创联合自动化控制有限公司 | A kind of wind power forecasting system for being easy to power scheduling |
| CN108564204A (en) * | 2018-03-23 | 2018-09-21 | 西安理工大学 | Least square method supporting vector machine power predicating method based on maximal correlation entropy criterion |
| CN108764534A (en) * | 2018-05-07 | 2018-11-06 | 湖北工业大学 | A kind of Wind power forecasting method based on improvement population |
| CN109242133A (en) * | 2018-07-11 | 2019-01-18 | 北京石油化工学院 | A kind of data processing method and system of earth's surface disaster alarm |
| CN109711435A (en) * | 2018-12-03 | 2019-05-03 | 三峡大学 | A kind of support vector machines on-Line Voltage stability monitoring method based on genetic algorithm |
| CN110889564A (en) * | 2019-12-16 | 2020-03-17 | 吉林大学 | Short-term power load prediction method based on MKPCA-RBFNN |
| CN111091878A (en) * | 2019-11-07 | 2020-05-01 | 上海大学 | Method for rapidly predicting perovskite dielectric constant |
| CN111985796A (en) * | 2020-08-07 | 2020-11-24 | 华中科技大学 | Method for predicting concrete structure durability based on random forest and intelligent algorithm |
| CN112069656A (en) * | 2020-08-07 | 2020-12-11 | 湖北交投十巫高速公路有限公司 | Durable concrete mix proportion multi-objective optimization method based on LSSVM-NSGAII |
| CN112132185A (en) * | 2020-08-26 | 2020-12-25 | 上海大学 | Method for rapidly predicting band gap of double perovskite oxide based on data mining |
| CN112800541A (en) * | 2021-01-26 | 2021-05-14 | 南京航空航天大学 | Aero-engine load spectrum task segment modeling method based on support vector machine regression |
| CN113054881A (en) * | 2021-02-22 | 2021-06-29 | 江苏大学 | Method for constructing outer rotor stator coreless bearingless permanent magnet synchronous motor controller |
| CN113420825A (en) * | 2021-07-07 | 2021-09-21 | 国能龙源蓝天节能技术有限公司 | Abnormal data detection method based on support vector machine and electronic equipment |
| CN114004158A (en) * | 2021-11-02 | 2022-02-01 | 西安电子科技大学 | Sea surface electromagnetic scattering prediction method based on genetic algorithm optimization support vector machine |
| CN114354534A (en) * | 2021-12-30 | 2022-04-15 | 中国航空油料有限责任公司 | Method for establishing aviation kerosene property prediction model by utilizing binary linear classifier |
| CN116861324A (en) * | 2023-07-27 | 2023-10-10 | 西南石油大学 | Method for calculating viscosity of thickened oil emulsion based on support vector machine |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102938093A (en) * | 2012-10-18 | 2013-02-20 | 安徽工程大学 | Wind power forecasting method |
| CN104899665A (en) * | 2015-06-19 | 2015-09-09 | 国网四川省电力公司经济技术研究院 | Wind power short-term prediction method |
-
2016
- 2016-11-14 CN CN201611003309.1A patent/CN106503867A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102938093A (en) * | 2012-10-18 | 2013-02-20 | 安徽工程大学 | Wind power forecasting method |
| CN104899665A (en) * | 2015-06-19 | 2015-09-09 | 国网四川省电力公司经济技术研究院 | Wind power short-term prediction method |
Non-Patent Citations (2)
| Title |
|---|
| 刘爱国 等: "《基于GA优化SVM的风电功率的超短期预测》", 《电力系统保护与控制》 * |
| 裴瑞平 等: "《基于GA-LSSVM的短期风功率预测》", 《自动化与仪器仪表》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106960260A (en) * | 2017-03-27 | 2017-07-18 | 深圳汇创联合自动化控制有限公司 | A kind of wind power forecasting system for being easy to power scheduling |
| CN108564204A (en) * | 2018-03-23 | 2018-09-21 | 西安理工大学 | Least square method supporting vector machine power predicating method based on maximal correlation entropy criterion |
| CN108564204B (en) * | 2018-03-23 | 2022-01-07 | 西安理工大学 | Least square support vector machine electricity quantity prediction method based on maximum correlation entropy criterion |
| CN108764534A (en) * | 2018-05-07 | 2018-11-06 | 湖北工业大学 | A kind of Wind power forecasting method based on improvement population |
| CN108764534B (en) * | 2018-05-07 | 2021-12-03 | 湖北工业大学 | Wind power generation power prediction method based on improved particle swarm |
| CN109242133A (en) * | 2018-07-11 | 2019-01-18 | 北京石油化工学院 | A kind of data processing method and system of earth's surface disaster alarm |
| CN109242133B (en) * | 2018-07-11 | 2022-03-22 | 北京石油化工学院 | Data processing method and system for ground disaster early warning |
| CN109711435A (en) * | 2018-12-03 | 2019-05-03 | 三峡大学 | A kind of support vector machines on-Line Voltage stability monitoring method based on genetic algorithm |
| CN111091878A (en) * | 2019-11-07 | 2020-05-01 | 上海大学 | Method for rapidly predicting perovskite dielectric constant |
| CN110889564A (en) * | 2019-12-16 | 2020-03-17 | 吉林大学 | Short-term power load prediction method based on MKPCA-RBFNN |
| CN112069656A (en) * | 2020-08-07 | 2020-12-11 | 湖北交投十巫高速公路有限公司 | Durable concrete mix proportion multi-objective optimization method based on LSSVM-NSGAII |
| CN112069656B (en) * | 2020-08-07 | 2024-01-12 | 湖北交投十巫高速公路有限公司 | LSSVM-NSGAII durable concrete mixing ratio multi-objective optimization method |
| CN111985796A (en) * | 2020-08-07 | 2020-11-24 | 华中科技大学 | Method for predicting concrete structure durability based on random forest and intelligent algorithm |
| CN112132185A (en) * | 2020-08-26 | 2020-12-25 | 上海大学 | Method for rapidly predicting band gap of double perovskite oxide based on data mining |
| CN112132185B (en) * | 2020-08-26 | 2023-07-18 | 上海大学 | Method for rapidly predicting double perovskite oxide band gap based on data mining |
| CN112800541A (en) * | 2021-01-26 | 2021-05-14 | 南京航空航天大学 | Aero-engine load spectrum task segment modeling method based on support vector machine regression |
| CN113054881A (en) * | 2021-02-22 | 2021-06-29 | 江苏大学 | Method for constructing outer rotor stator coreless bearingless permanent magnet synchronous motor controller |
| CN113420825A (en) * | 2021-07-07 | 2021-09-21 | 国能龙源蓝天节能技术有限公司 | Abnormal data detection method based on support vector machine and electronic equipment |
| CN114004158A (en) * | 2021-11-02 | 2022-02-01 | 西安电子科技大学 | Sea surface electromagnetic scattering prediction method based on genetic algorithm optimization support vector machine |
| CN114354534A (en) * | 2021-12-30 | 2022-04-15 | 中国航空油料有限责任公司 | Method for establishing aviation kerosene property prediction model by utilizing binary linear classifier |
| CN116861324A (en) * | 2023-07-27 | 2023-10-10 | 西南石油大学 | Method for calculating viscosity of thickened oil emulsion based on support vector machine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106503867A (en) | A kind of genetic algorithm least square wind power forecasting method | |
| CN108320016B (en) | Short-term prediction method for building energy consumption | |
| CN109142171B (en) | Urban PM10 concentration prediction method based on feature expansion and fusing with neural network | |
| CN108921339B (en) | Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine | |
| CN106971240A (en) | The short-term load forecasting method that a kind of variables choice is returned with Gaussian process | |
| CN106371427A (en) | Industrial process fault classification method based on analytic hierarchy process and fuzzy fusion | |
| CN103488869A (en) | Wind power generation short-term load forecast method of least squares support vector machine | |
| CN110059852A (en) | A kind of stock yield prediction technique based on improvement random forests algorithm | |
| CN103679263A (en) | Thunder and lightning approach forecasting method based on particle swarm support vector machine | |
| CN102185735A (en) | Network security situation prediction method | |
| CN109143408B (en) | Dynamic region combined short-time rainfall forecasting method based on MLP | |
| CN105095494A (en) | Method for testing categorical data set | |
| CN109919236A (en) | A kind of BP neural network multi-tag classification method based on label correlation | |
| CN114444378A (en) | Short-term power prediction method for regional wind power cluster | |
| CN113762387B (en) | Multi-element load prediction method for data center station based on hybrid model prediction | |
| CN112232561A (en) | Power load probability prediction method based on constrained parallel LSTM quantile regression | |
| CN109858665A (en) | Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO | |
| Baran et al. | Calibration of wind speed ensemble forecasts for power generation | |
| Liu et al. | An RBF neural network based on improved black widow optimization algorithm for classification and regression problems | |
| CN104217296A (en) | Listed company performance comprehensive evaluation method | |
| CN110738363A (en) | photovoltaic power generation power prediction model and construction method and application thereof | |
| CN104794546A (en) | Wind power climbing forecasting method based on deep confidence network classifying method | |
| CN114004153A (en) | Penetration depth prediction method based on multi-source data fusion | |
| CN116245259B (en) | Photovoltaic power generation prediction method and device based on depth feature selection and electronic equipment | |
| CN116663404A (en) | Flood forecasting method and system coupling artificial intelligence and Bayesian theory |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170315 |
|
| WD01 | Invention patent application deemed withdrawn after publication |