CN109102102A - Based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method - Google Patents
Based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method Download PDFInfo
- Publication number
- CN109102102A CN109102102A CN201810660529.4A CN201810660529A CN109102102A CN 109102102 A CN109102102 A CN 109102102A CN 201810660529 A CN201810660529 A CN 201810660529A CN 109102102 A CN109102102 A CN 109102102A
- Authority
- CN
- China
- Prior art keywords
- photovoltaic
- svr
- phase space
- power output
- short term
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000003064 k means clustering Methods 0.000 claims abstract description 7
- 238000003780 insertion Methods 0.000 claims description 4
- 230000037431 insertion Effects 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 3
- 238000005065 mining Methods 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 238000012360 testing method Methods 0.000 claims description 3
- 230000000295 complement effect Effects 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims 1
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 238000010248 power generation Methods 0.000 description 14
- 239000011159 matrix material Substances 0.000 description 9
- 238000012549 training Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000005619 thermoelectricity Effects 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Marketing (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Probability & Statistics with Applications (AREA)
- Game Theory and Decision Science (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Photovoltaic Devices (AREA)
Abstract
The invention proposes a kind of based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, based on photovoltaic plant weather station actual monitoring historical data, clustering processing is carried out to photovoltaic plant historical sample using K-means clustering algorithm, establishes the photovoltaic plant short term power hybrid prediction model for being based on multivariate phase space reconstruction and support vector regression (SVR).Irradiation intensity and temperature-time sequence are decomposed and reconstructed first with CC method phase space reconfiguration, then irradiation intensity, the temperature after reconstruct are predicted and tracked using SVR, and the reliable input as power prediction model.Finally it is fitted using the nonlinearity that SVR model then realizes photovoltaic plant power.
Description
Technical field
The present invention relates to a kind of based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, belongs to electricity
Force system field of automation technology.
Background technique
With the rapid development of economic society, energy demand increasingly increases.Tradition is with the non-renewable energy such as thermoelectricity and petroleum
Supply mode based on the consumption of source is difficult to meet, and can also produce serious influence to climatic environment, the new energy especially sun
The effective way of substitution fossil energy can be become.The output power of photovoltaic plant is affected by weather conditions, randomness and
Fluctuation is stronger.The large-scale grid-connected coordinated control to electric system and management and running bring huge challenge, light
The effective means that volt power generation prediction exactly solves the above problems.
The influence factor of photovoltaic plant power output mainly includes geographical location and meteorological condition.After photovoltaic plant is built up, ground
Position is managed, arranges arrangement mode, system effectiveness is determined.Therefore, photovoltaic generation power fluctuation main source gas on site
As condition.Currently, the weather station of photovoltaic plant can monitor including irradiation intensity, temperature, wind speed, wind direction etc. it is multiple it is meteorological because
Element, these factors suffer from certain influence to the output power of photovoltaic battery panel, wherein what is played a major role is irradiation intensity
And temperature.Irradiation intensity and photovoltaic system output power have the relevance of height, and it is higher to irradiate stronger output power.And in phase
With under the conditions of, the raising of temperature will lead to the decline of photovoltaic output power.
Irradiation intensity is all larger with the degree of association of photovoltaic power station power generation power under any weather conditions, and temperature also can be to light
Volt power output has direct influence.Simultaneously as being needed in modeling process between temperature and irradiation intensity there are stronger coupling
It takes in.As can be seen that all kinds of meteorologic factors all have a certain impact to photovoltaic power output from conspicuousness.But from correlation
Property angle, irradiation intensity and temperature and photovoltaic power output have that there is stronger relevances.
It can be seen that photovoltaic power generation output forecasting is the important support to large-scale photovoltaic interconnection technology, it is to guarantee power grid security
One of effective ways of stable operation.Method proposed by the invention is exactly a kind of based on multivariate phase space reconstruction and SVR
Photovoltaic power output short term prediction method.Compared to conventional method this method comprehensively considered influence photovoltaic power output critical environments because
Element: irradiation intensity and temperature.By the phase space reconfiguration to irradiation intensity and temperature-time sequence, prediction day has sufficiently been excavated
Relationship between weather condition and weather history situation realizes the prediction of irradiation intensity, temperature and photovoltaic power output.
Summary of the invention
It is an object of the invention to: in view of the defects existing in the prior art, propose a kind of based on multivariate phase space reconstruction
With the photovoltaic power output short term prediction method of SVR, the nonlinearity fitting of photovoltaic plant power is realized.
In order to reach the goals above, the present invention provides a kind of classification hyperspectral imagery sides based on the special inquiry learning of bands of a spectrum
Method includes the following steps:
Based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, method includes the following steps:
Step 1 carries out clustering to photovoltaic plant historical data using K-means clustering algorithm；
Step 2 carries out photovoltaic plant original power, irradiation intensity and temperature-time sequence using CC method phase space reconfiguration
It decomposes and reconstructs, the high and low frequency information of abundant mining data；
Step 3, using SVR model to after reconstruct irradiation intensity and temperature predicted and tracked, and it is pre- as power
Survey the reliable input of model；
Step 4 then realizes that the nonlinearity fitting of photovoltaic plant power obtains prediction result using CC-SVR model.
Preferably, clustering tool is carried out to photovoltaic plant historical data using K-means clustering algorithm in the step 1
Steps are as follows for gymnastics work:
Step 1.1, cluster centre is initialized.According to actual needs or experience randomly selects K sample from sample set and makees
For the initial cluster center of algorithm, it is denoted asIterative steps l=0 at this time.
Step 1.2, a sample is randomly selected from sample set, is denoted as Xn。
Step 1.3, sample X is calculatednAt a distance from each cluster centre, determine apart from nearest cluster centre, such as formula (1) institute
Show:
Step 1.4, by sample XnIt is divided into and adjusts cluster centre after nearest cluster set with cluster centre, such as formula
(2) shown in:
In formula, niFor the number of samples of the i-th class, ΩiFor the sample set of the i-th class.
Step 1.5, detecting whether that all samples all operate terminates, if meeting formula (1) minimum or no longer changing, iteration
Terminate；Otherwise second step is jumped to.
Preferably, CC method phase space reconfiguration modeling procedure in the step 2 are as follows:
Step 2.1, the sub- sequence that original time time series { x (k), k=1,2 ..., N } is resolved into t complementary overhangs
Column, t are reconstruct time delay, it may be assumed that
Wherein N is the integral multiple of t.
Step 2.2 calculates correlation integral to each subsequence:
M is phase space reconstruction quantity of state number.According to BDS statistical conclusions, m=2,3,4,5, corresponding m=2 are taken, 3,4,5, r
=0.5 σ, σ, 1.5 σ, 2 σ, σ is time series standard deviation.
Step 2.3 calculates test statistics using piecemeal Average Strategy:
Step 2.4 calculates residual quantity:
△ S (m, t)=max { S (m, rj,t)}-min{S(m,rj,t)} (6)
△ S (m, t) has measured the maximum deviation of S (m, N, r, t) pair radius r.When △ S (m, t) is minimum value, weigh at this time
Point in structure phase space is close to being uniformly distributed, and the track of reconfiguration system is fully deployed in phase space, and the correlation of time series is most
Close to zero.Therefore △ S (m, t)~t curve reflects the autocorrelation performance of original series.
Step 2.5 calculates:
WithReflect the autocorrelation of former time series.ConsiderValue can just be born,Value is always positive
Number, comprehensively considersWithDefine index:
First local minizing point corresponding to t be optimum delay τ, Scor(t) global minimum is corresponding
The i.e. best insertion window t of tω, tω=(m-1) τ, so that it is determined that the embedding dimension m and time delay τ.
Preferably, SVR modeling procedure in the step 3 are as follows:
Construction optimal hyperlane problem is converted into optimization problem:
Constraint condition are as follows:
yi((w·xi)+b)≥1-ξi, i=1,2 ..., l (11)
W is optimal hyperlane normal vector, and b is threshold value, and C is punishment parameter, ξiFor slack variable.It can use Lagrange
Multiplier method solves the above problem.If expanding to nonlinear problem, it can use mapping phi (x) and reflect the sample in lower dimensional space
It penetrates as in higher dimensional space, at this time objective function are as follows:
αiFor Lagrange multiplier.
The input sample space reflection of linearly inseparable to higher dimensional space is realized linear regression by SVR, is obtained non-linear time
Return functionIt is solved using method of Lagrange multipliers.Under normal circumstances, Gauss in the selection of kernel function
Kernel function K (xk,νi)=exp [- | | xk-vi||/(2σ2)] more, wherein σ is Gauss nuclear parameter.
Preferably, CC-SVR photovoltaic power ultra-short term prediction model modeling procedure is based in the step 4 are as follows:
Step 4.1 is based on CC method to irradiation intensity { rt, t=1,2 ... M } and temperature { wt, t=1,2 ... M history
Time series carries out phase space reconfiguration with the embedding dimension m and time delay τ, after obtaining reconstruct
Phase space matrix are as follows:
Step 4.2, based on after reconstruct irradiation intensity and temperature matrices using SVR establish irradiation intensity and temperature
The prediction model of degree.Assuming that training sample has k+1, then the input square that irradiation intensity and temperature are trained
Battle array X1And X2Respectively shown in formula (15)-(16), output matrix Y1And Y2Respectively formula (17)-
(18) shown in.Forecast period is by X'1=[rk+1-(m-1)τ,rk+1-(m-2)τ,...rk+1] and
X'2=[wk+1-(m-1)τ,wk+1-(m-2)τ,...wk+1] input as SVR model, then SVR model at this time is defeated
Y out1'=rk+2And Y2'=wk+2The as prediction result of irradiation intensity and temperature is denoted as sequence { rtAnd { wt}。
Y1=[r2+(m-1)τ r3+(m-1)τ ... rk+1]T (17)
Y2=[w2+(m-1)τ w3+(m-1)τ ... wk+1]T (18)
Step 4.3 establishes photovoltaic power generation output forecasting model using SVR, it is assumed that training sample has k+1, then
The input matrix of photovoltaic power output training is X3, output matrix Y3As shown in formula (19)-(20).In advance
The survey stage is by X'3=[pk+1-(m-1)τ,pk+1-(m-2)τ,...pk+1] input as SVR model, then SVR at this time
The output Y of model3'=pk+2For the prediction result of photovoltaic power output, it is denoted as sequence { pt}。
Y3=[p2+(m-1)τ p3+(m-1)τ ... pk+1]T (20)。
Preferably, the result that the step 4 obtains has comprehensively considered the critical environments factor for influencing photovoltaic power output: irradiation is strong
Degree and temperature.By the phase space reconfiguration to irradiation intensity and temperature-time sequence, the weather condition of prediction day has sufficiently been excavated
With the relationship between weather history situation, the prediction of irradiation intensity, temperature and photovoltaic power output is realized.
Detailed description of the invention
The present invention will be further described below with reference to the drawings.
Fig. 1 is CC-SVR photovoltaic power ultra-short term prediction model modeling procedure figure in the present invention.
Fig. 2 is in the weaker situation of fluctuation in the present inventionScor(t) and the relation curve of t.
Fig. 3 is in the stronger situation of fluctuation in the present inventionScor(t) and the relation curve of t.
Fig. 4 is the weaker weather irradiation intensity prediction curve of fluctuation in the present invention.
Fig. 5 is the weaker weather temperature prediction curve of fluctuation in the present invention.
Fig. 6 is the weaker weather photovoltaic power generation output forecasting curve of fluctuation in the present invention.
Fig. 7 is the stronger weather irradiation intensity prediction curve of fluctuation in the present invention.
Fig. 8 is the stronger weather temperature prediction curve of fluctuation in the present invention.
Fig. 9 is the stronger weather photovoltaic power generation output forecasting curve of fluctuation in the present invention.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description.
The present invention provides a kind of based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, the party
Method the following steps are included:
Step 1 carries out clustering to photovoltaic plant historical data using K-means clustering algorithm；
Step 2 carries out photovoltaic plant original power, irradiation intensity and temperature-time sequence using CC method phase space reconfiguration
It decomposes and reconstructs, the high and low frequency information of abundant mining data；
Step 3, using SVR model to after reconstruct irradiation intensity and temperature predicted and tracked, and it is pre- as power
Survey the reliable input of model；
Step 4 then realizes that the nonlinearity fitting of photovoltaic plant power obtains prediction result using CC-SVR model；
The step 1K-means clustering algorithm are as follows:
d1=[tmin,tmax,tmean,smin,smax,smean,x1min,x1max,x1mean]
Wherein p (t) is the photovoltaic power output of t moment, and △ t indicates sampling time interval, X (t)=[x (t), x (t- △
t),...,x(t-(D-1)△t)]TIt is the power after standardization, pmax, pmaxThe maximum and minimum value of photovoltaic power output is respectively indicated,
E=[1,1 ..., 1]T∈RN×1, x1min,x1max,x1meanRespectively indicate maximum value, the minimum value peace that photovoltaic is contributed in one day
Mean value.tmin,tmax,tmean,smin,smax,smeanRespectively indicate temperature and the maximum value of irradiation intensity, minimum value peace in one day
Mean value.
The quantity of state reconstructed in step 2 phase space is represented by
Xi=[x (i), x (i+ τ) ..., x (i+ (m-1) τ)]
Wherein, m is the embedding dimension, and τ is time delay.
SVR model in the step 3 are as follows:
Constraint condition are as follows: yi((w·xi)+b)≥1-ξi, i=1,2 ..., l
W is optimal hyperlane normal vector, and b is threshold value, and C is punishment parameter, ξiFor slack variable.It can use Lagrange
Multiplier method solves the above problem.If expanding to nonlinear problem, it can use mapping phi (x) and reflect the sample in lower dimensional space
It penetrates as in higher dimensional space, at this time objective function are as follows:
αiFor Lagrange multiplier.
The step 4CC-SVR photovoltaic power ultra-short term prediction result are as follows:
Y3=[p2+(m-1)τ p3+(m-1)τ ... pk+1]T
Wherein Y3'=pk+2For the prediction result of photovoltaic power output, it is denoted as sequence { pt}
Mould process are as follows:
(1) based on CC method to irradiation intensity { rt, t=1,2 ... M } and temperature { wt, t=1,2 ... M historical time
Sequence carries out phase space reconfiguration with the embedding dimension m and time delay τ, the phase space matrix after being reconstructed are as follows:
(2) based on after reconstruct irradiation intensity and temperature matrices establish using SVR the prediction model of irradiation intensity and temperature.
Assuming that training sample has k+1, then the input matrix X that irradiation intensity and temperature are trained1And X2It is defeated respectively shown in formula (3)-(4)
Matrix Y out1And Y2Respectively shown in formula (5)-(6).Forecast period is by X'1=[rk+1-(m-1)τ,rk+1-(m-2)τ,...rk+1] and X'2
=[wk+1-(m-1)τ,wk+1-(m-2)τ,...wk+1] input as SVR model, then the output Y of SVR model at this time1'=rk+2And Y2'
=wk+2The as prediction result of irradiation intensity and temperature is denoted as sequence { rtAnd { wt}。
Y1=[r2+(m-1)τ r3+(m-1)τ ... rk+1]T (5)
Y2=[w2+(m-1)τ w3+(m-1)τ ... wk+1]T (6)
(3) photovoltaic power generation output forecasting model is established using SVR, it is assumed that training sample there are k+1, then the training of photovoltaic power output is defeated
Entering matrix is X3, output matrix Y3As shown in formula (7)-(8).Forecast period is by X'3=[pk+1-(m-1)τ,pk+1-(m-2)τ,
...pk+1] input as SVR model, then the output Y of SVR model at this time3'=pk+2For the prediction result of photovoltaic power output, it is denoted as
Sequence { pt}。
Y3=[p2+(m-1)τ p3+(m-1)τ ... pk+1]T (8)
Case study on implementation
To verify having based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method proposed by the present invention
Effect property herein carries out the photovoltaic plant power output situation in the case of two kinds of different fluctuations using the programming of MATLAB 8.3 modeling pre-
It surveys.To mould based on the implementation of certain city's photovoltaic electric station grid connection point output power and photovoltaic plant weather station monitoring historical data
Type is verified, data sampling period 5min, and it is weaker to choose fluctuation from April, 2016 in September, 2017 for data sample
Relatively strong (on June 7th, the 2016) two types of (on July 21st, 2016) and fluctuation carry out real data modeling and test.
1. photovoltaic power generation output forecasting in the weaker situation of fluctuation
(1) irradiation intensity and the selection of temperature-time sequence Parameters for Phase Space Reconstruction based on CC method
According to fig. 2,Scor(t) with the relation curve of t, irradiation intensity and temperature in the weaker situation of fluctuation can be determined
Time delay τ=7 for spending phase space reconfiguration and most preferably insertion window tω=7, according to tω=(m-1) τ can determine that embedded space is tieed up
Number m=2.
(2) photovoltaic power generation output forecasting in the weaker situation of fluctuation
Irradiation intensity and temperature are all relatively stable under the weaker weather condition of fluctuation, and trend comparison is obvious.Fig. 3-Fig. 4 institute
It is shown as the prediction result of fluctuation weaker weather photovoltaic plant irradiation intensity and temperature.Fig. 5 is the prediction of irradiation intensity and temperature
As a result the resulting photovoltaic power generation output forecasting curve of SVR regression model is inputted.
As it can be seen that, since temperature and irradiation intensity fluctuation are little, variation tendency is more under the weaker weather condition of fluctuation
Obviously, ideal prediction result can be obtained using phase space reconfiguration and SVR combination.Fig. 5 also indicates that accurate spoke
The input for being used as SVR regression model according to intensity and temperature prediction result can obtain preferable power prediction result.
2. photovoltaic power generation output forecasting in the stronger situation of fluctuation
(1) irradiation intensity and the selection of temperature-time sequence Parameters for Phase Space Reconstruction based on CC method
According to Fig. 6,Scor(t) with the relation curve of t, irradiation intensity and temperature in the stronger situation of fluctuation can be determined
Time delay τ=8 for spending phase space reconfiguration and most preferably insertion window tω=8, according to tω=(m-1) τ can determine that embedded space is tieed up
Number m=2.
(2) photovoltaic power generation output forecasting in the stronger situation of fluctuation
Such as the cloudy this kind of stronger situation of fluctuation of sleet, irradiation intensity is lower but overall variation trend is still more bright
It is aobvious.The method proposed in this paper for combining CC method phase space reconfiguration and SVR also may be implemented accurately to predict.Fig. 7-figure
8 show the prediction result of fluctuation stronger weather photovoltaic plant irradiation intensity and temperature.Fig. 9 is in the stronger situation of fluctuation
Photovoltaic power generation output forecasting curve.
For the prediction effect of evaluation photovoltaic power generation power output, average absolute percentage error ε is generallyd useMAPEAnd root-mean-square error
εRMSETo measure the departure degree between global error degree and predicted value and true value.
Precision of prediction in the case of the fluctuation of Table I difference
As can be seen that CC-SVR model error in the case where gently i.e. fluctuation is weaker for irradiation intensity variation from Table I
It is relatively small；And in the case where irradiation intensity variation is acutely the stronger situation of fluctuation relatively, prediction error can also rise with it.But
On the whole, photovoltaic plant short term prediction method proposed by the present invention is relatively reasonable regardless of can reach for which kind of weather condition
Prediction effect.
Finally it should be noted that: embodiment described above is only the preferred embodiment of the present invention rather than protects model to it
The limitation enclosed, although the application is described in detail referring to above-described embodiment, those of ordinary skill in the art are answered
When understanding: still can carry out various changes, modification to the specific embodiment of application after those skilled in the art read the application
Perhaps equivalent replacement but these changes, modification or equivalent replacement, are applying within pending claims.
Claims (9)
1. based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is characterised in that: including following step
It is rapid:
Step 1 carries out clustering to photovoltaic plant historical data using K-means clustering algorithm；
Step 2 decomposes photovoltaic plant original power, irradiation intensity and temperature-time sequence using CC method phase space reconfiguration
And reconstruct, the high and low frequency information of abundant mining data；
Step 3, using SVR model to after reconstruct irradiation intensity and temperature predicted and tracked, and as power prediction mould
The reliable input of type；
Step 4 then realizes that the nonlinearity fitting of photovoltaic plant power obtains prediction result using CC-SVR model.
2. it is according to claim 1 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: carrying out clustering concrete operations step to photovoltaic plant historical data using K-means clustering algorithm in the step 1
It is rapid as follows:
Step 1.1, initialization cluster centre, according to actual needs or experience randomly selects K sample as calculation from sample set
The initial cluster center of method, is denoted asIterative steps l=0 at this time；
Step 1.2 randomly selects a sample from sample set, is denoted as Xn；
Step 1.3 calculates sample XnAt a distance from each cluster centre, determine apart from nearest cluster centre, as shown in formula (1):
Step 1.4, by sample XnIt is divided into and adjusts cluster centre after nearest cluster set with cluster centre, such as formula (2) institute
Show:
In formula, niFor the number of samples of the i-th class, ΩiFor the sample set of the i-th class；
Step 1.5 detects whether that all samples all operate and terminates, if it is minimum or no longer change, iteration knot to meet formula (1)
Beam；Otherwise second step is jumped to.
3. it is according to claim 2 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: in the step 1, K-means clustering algorithm are as follows:
d1=[tmin,tmax,tmean,smin,smax,smean,x1min,x1max,x1mean]
Wherein p (t) is the photovoltaic power output of t moment, and △ t indicates sampling time interval, X (t)=[x (t), x (t- △ t) ..., x
(t-(D-1)△t)]TIt is the power after standardization, pmax, pmaxRespectively indicate photovoltaic power output maximum and minimum value, E=[1,
1,...,1]T∈RN×1, x1min,x1max,x1meanMaximum value, minimum value and the average value that photovoltaic is contributed in one day are respectively indicated,
tmin,tmax,tmean,smin,smax,smeanRespectively indicate temperature and the maximum value of irradiation intensity, minimum value and average value in one day.
4. it is according to claim 1 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: in the step 2, the quantity of state reconstructed in phase space is represented by
Xi=[x (i), x (i+ τ) ..., x (i+ (m-1) τ)]
Wherein, m is the embedding dimension, and τ is time delay.
5. it is according to claim 4 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: in the step 2, CC method phase space reconfiguration models specific steps are as follows:
Step 2.1, the subsequence that original time time series { x (k), k=1,2 ..., N } is resolved into t complementary overhangs, t
To reconstruct time delay, it may be assumed that
Wherein N is the integral multiple of t；
Step 2.2 calculates correlation integral to each subsequence:
M is phase space reconstruction quantity of state number, according to BDS statistical conclusions, takes m=2,3,4,5, corresponding m=2,3,4,5, r=
0.5 σ, σ, 1.5 σ, 2 σ, σ is time series standard deviation；
Step 2.3 calculates test statistics using piecemeal Average Strategy:
Step 2.4 calculates residual quantity:
△ S (m, t)=max { S (m, rj,t)}-min{S(m,rj,t)} (6)
△ S (m, t) has measured the maximum deviation of S (m, N, r, t) pair radius r.When △ S (m, t) is minimum value, phase is reconstructed at this time
Point in space is close to being uniformly distributed, and the track of reconfiguration system is fully deployed in phase space, and the correlation of time series is closest
In zero.Therefore △ S (m, t)~t curve reflects the autocorrelation performance of original series；
Step 2.5 calculates:
WithReflect the autocorrelation of former time series.ConsiderValue can just be born,Value is always positive number, comprehensive
It closes and considersWithDefine index:
First local minizing point corresponding to t be optimum delay τ, Scor(t) the corresponding t of global minimum
I.e. most preferably insertion window tω, tω=(m-1) τ, so that it is determined that the embedding dimension m and time delay τ.
6. it is according to claim 1 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: in the step 3, SVR model are as follows:
Constraint condition are as follows: yi((w·xi)+b)≥1-ξi, i=1,2 ..., l
W is optimal hyperlane normal vector, and b is threshold value, and C is punishment parameter, ξiFor slack variable.
7. it is according to claim 6 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: being solved by method of Lagrange multipliers；If expanding to nonlinear problem, using mapping phi (x) by lower dimensional space
In sample be mapped as in higher dimensional space, objective function at this time are as follows:
αiFor Lagrange multiplier.
8. it is according to claim 1 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: in the step 4, CC-SVR photovoltaic power ultra-short term prediction result are as follows:
Y3=[p2+(m-1)τ p3+(m-1)τ ... pk+1]T
Wherein Y3'=pk+2For the prediction result of photovoltaic power output, it is denoted as sequence { pt}。
9. it is according to claim 8 based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method, it is special
Sign is: in the step 4, the result of acquisition comprehensively considered influence photovoltaic power output critical environments factor: irradiation intensity and
Temperature.By the phase space reconfiguration to irradiation intensity and temperature-time sequence, sufficiently excavated prediction day weather condition with go through
Relationship between history weather condition realizes the prediction of irradiation intensity, temperature and photovoltaic power output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810660529.4A CN109102102A (en) | 2018-06-25 | 2018-06-25 | Based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810660529.4A CN109102102A (en) | 2018-06-25 | 2018-06-25 | Based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109102102A true CN109102102A (en) | 2018-12-28 |
Family
ID=64844936
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810660529.4A Pending CN109102102A (en) | 2018-06-25 | 2018-06-25 | Based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109102102A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111797496A (en) * | 2020-05-21 | 2020-10-20 | 中国电力科学研究院有限公司 | New energy station generated output time sequence construction method and device |
| CN112132328A (en) * | 2020-09-04 | 2020-12-25 | 国网上海市电力公司 | Photovoltaic output power ultra-short-term local emotion reconstruction neural network prediction method |
| CN113111572A (en) * | 2021-03-23 | 2021-07-13 | 合肥工业大学 | Method and system for predicting residual life of aircraft engine |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104732296A (en) * | 2015-04-01 | 2015-06-24 | 贵州电力试验研究院 | Modeling method for distributed photovoltaic output power short-term prediction model |
| CN107563565A (en) * | 2017-09-14 | 2018-01-09 | 广西大学 | A kind of short-term photovoltaic for considering Meteorology Factor Change decomposes Forecasting Methodology |
| CN107766990A (en) * | 2017-11-10 | 2018-03-06 | 河海大学 | A kind of Forecasting Methodology of photovoltaic power station power generation power |
| CN107944604A (en) * | 2017-11-10 | 2018-04-20 | 中国电力科学研究院有限公司 | A kind of weather pattern recognition methods and device for photovoltaic power prediction |
-
2018
- 2018-06-25 CN CN201810660529.4A patent/CN109102102A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104732296A (en) * | 2015-04-01 | 2015-06-24 | 贵州电力试验研究院 | Modeling method for distributed photovoltaic output power short-term prediction model |
| CN107563565A (en) * | 2017-09-14 | 2018-01-09 | 广西大学 | A kind of short-term photovoltaic for considering Meteorology Factor Change decomposes Forecasting Methodology |
| CN107766990A (en) * | 2017-11-10 | 2018-03-06 | 河海大学 | A kind of Forecasting Methodology of photovoltaic power station power generation power |
| CN107944604A (en) * | 2017-11-10 | 2018-04-20 | 中国电力科学研究院有限公司 | A kind of weather pattern recognition methods and device for photovoltaic power prediction |
Non-Patent Citations (2)
| Title |
|---|
| 邢校萄: "支持向量机在短期光伏发电功率预测中的应用", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 陈健: "基于多变量相空间重构的投资组合策略研究", 《中国优秀硕士学位论文全文数据库 经济与管理科学辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111797496A (en) * | 2020-05-21 | 2020-10-20 | 中国电力科学研究院有限公司 | New energy station generated output time sequence construction method and device |
| CN111797496B (en) * | 2020-05-21 | 2023-05-23 | 中国电力科学研究院有限公司 | New energy station power generation output time sequence construction method and device |
| CN112132328A (en) * | 2020-09-04 | 2020-12-25 | 国网上海市电力公司 | Photovoltaic output power ultra-short-term local emotion reconstruction neural network prediction method |
| CN113111572A (en) * | 2021-03-23 | 2021-07-13 | 合肥工业大学 | Method and system for predicting residual life of aircraft engine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | A comparison of day-ahead photovoltaic power forecasting models based on deep learning neural network | |
| CN110070226A (en) | Photovoltaic power prediction technique and system based on convolutional neural networks and meta learning | |
| Liu et al. | Fault diagnosis approach for photovoltaic array based on the stacked auto-encoder and clustering with IV curves | |
| Fadare | The application of artificial neural networks to mapping of wind speed profile for energy application in Nigeria | |
| CN110909919A (en) | Photovoltaic power prediction method of depth neural network model with attention mechanism fused | |
| Moayyed et al. | A Cyber-Secure generalized supermodel for wind power forecasting based on deep federated learning and image processing | |
| Ciulla et al. | Forecasting the cell temperature of PV modules with an adaptive system | |
| Zhang et al. | Optimal graph structure based short-term solar PV power forecasting method considering surrounding spatio-temporal correlations | |
| CN106228278A (en) | Photovoltaic power prognoses system | |
| CN110929953A (en) | Photovoltaic power station ultra-short term output prediction method based on cluster analysis | |
| CN109102102A (en) | Based on the photovoltaic of multivariate phase space reconstruction and SVR power output short term prediction method | |
| Aliberti et al. | Forecasting Short-term Solar Radiation for Photovoltaic Energy Predictions. | |
| He et al. | A power forecasting approach for PV plant based on irradiance index and LSTM | |
| CA3231095A1 (en) | Systems and methods for load forecasting for improved forecast results based on tuned weather data | |
| CN116796168A (en) | CNN-BiLSTM high-altitude multi-factor power transmission line audible noise prediction method based on multi-head attention mechanism | |
| CN115409369A (en) | Comprehensive energy system reliability evaluation method based on mechanism and data hybrid driving | |
| CN117293809A (en) | Multi-time space scale new energy generation power prediction method based on large model | |
| CN112183877A (en) | Photovoltaic power station fault intelligent diagnosis method based on transfer learning | |
| Putra et al. | Modeling of high uncertainty photovoltaic generation in quasi dynamic power flow on distribution systems: A case study in Java Island, Indonesia | |
| Li et al. | Assessing urban micro-climates with vertical and horizontal building morphological cutting deep transfer learning neural networks | |
| Chen et al. | Probabilistic Prediction of Photovoltaic Power Using Bayesian Neural Network-LSTM Model | |
| CN105741192A (en) | Short-term wind speed combined forecasting method for wind turbine cabin of wind power plant | |
| Xiong et al. | Combined model for short-term wind power prediction based on deep neural network and long short-term memory | |
| Tran et al. | Short-term solar power generation forecasting using edge ai | |
| Yang et al. | Multi-scale convolutional echo state network with an effective pre-training strategy for solar irradiance forecasting |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181228 |
|
| RJ01 | Rejection of invention patent application after publication |