CN110059891A - A kind of photovoltaic plant output power predicting method based on VMD-SVM-WSA-GM built-up pattern - Google Patents
A kind of photovoltaic plant output power predicting method based on VMD-SVM-WSA-GM built-up pattern Download PDFInfo
- Publication number
- CN110059891A CN110059891A CN201910351027.8A CN201910351027A CN110059891A CN 110059891 A CN110059891 A CN 110059891A CN 201910351027 A CN201910351027 A CN 201910351027A CN 110059891 A CN110059891 A CN 110059891A
- Authority
- CN
- China
- Prior art keywords
- output power
- sequence
- photovoltaic plant
- prediction
- single order
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- 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—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 invention belongs to technical field of photovoltaic power generation, specifically related to a kind of photovoltaic plant output power predicting method of base VMD-SVM-WSA-GM built-up pattern, main structure as prediction technique, the each mode done after VMD decomposition the invention proposes a kind of pair of photovoltaic plant history output power data sequence establishes support vector machines prediction model respectively, then it sums the new construction of reconstruct, the learning ability and generalization ability of support vector machines can be given full play to, precision of prediction and convergence rate are improved.
Description
Technical field
The invention belongs to technical field of photovoltaic power generation, and in particular to a kind of based on time series and intelligent optimization algorithm
Photovoltaic plant output power predicting method, be related specifically to it is a kind of based on variation mode decomposition (VMD), support vector machines (SVM),
The photovoltaic plant output power predicting method of whale colony optimization algorithm (WSA) and several model combinations of gray model (GM).
Background technique
Photovoltaic is the abbreviation of solar photovoltaic generation system, is a kind of photovoltaic effect using solar cell semiconductor material
It answers, solar radiation energy is converted directly into a kind of new power generating system of electric energy.Photovoltaic technology has many advantages: in addition to straight
It obtains except the illumination taken, does not consume the energy medium of any other form；The equipment such as large rotating machinery are not needed；Arrangement spirit
It is living, it can use existing vacant lot and roof etc..
Due to the lasting upheaval of the political economy situation of main petroleum, gas exporter in the world, each main energy sources disappear
Alternative solution is all actively being sought by consumption state, to reduce to outer energy interdependency, ensures the safety of national economy.Exactly such
Under background, solar energy obtains more and more extensive utilization as a kind of cleaning, cheap, sufficient and safety energy form.Data
It has been shown that, in by the end of November, 2017 by, China's photovoltaic add up installed capacity up to 125,790,000 kilowatts, increase by 67% on a year-on-year basis, add up installation
Capacity accounts for the specific gravity of total electricity installation up to 7.5%；2017, photovoltaic annual electricity generating capacity head surpassed 100,000,000,000 kilowatt hours.International Energy Agency
" the Chinese ad hoc report of world energy outlook 2017 " thinks that energy structure in China will gradually be transformed into clean electric power generation, wherein the sun
Energy photovoltaic will become the most economical generation mode of China, will with the low-carbon installed capacity that waterpower, wind energy and photovoltaic are led
It increases rapidly, the 60% of total installation of generating capacity will be accounted for the year two thousand forty.
Due to being influenced by complicated factors such as solar irradiation intensity and environment, photovoltaic plant output power have typically with
Machine and fluctuation, and this uncertain stability that will directly affect grid-connected rear electric system.Thus, to one section following
The output power of photovoltaic plant, which carries out Accurate Prediction, in time is very important, and is that technical staff is urgently to be resolved in related fields
The problem of.
Summary of the invention
The object of the present invention is to provide a kind of photovoltaic plant output power based on VMD-SVM-WSA-GM built-up pattern is pre-
Survey method realizes photovoltaic generating system and electric system to improve the precision of photovoltaic generating system short-term output power prediction
Scientific dispatch decision and optimization operation.
In order to solve the above-mentioned technical problem, the embodiment of the present invention the following technical schemes are provided:
A kind of photovoltaic plant output power predicting method based on VMD-SVM-WSA-GM built-up pattern, which is characterized in that
According to photovoltaic plant history output power data, its generated output value in future time period is predicted comprising the steps of:
Step 1, target photovoltaic plant history output power data are obtained, the history output power data are with setting time
For unit, time series is arranged in by the sampling time.
Step 2, the history output power data sequence is pre-processed, pretreatment measure includes changing and one at equal intervals
Rank Accumulating generation, specific as follows:
Change the data for some output power sampled point of completion missing at equal intervals, takes sampled point front and back to number in fixed step size
According to average value, it is preferred that the step-length is taken as 3~6.
Single order Accumulating generation is used to eliminate the influence of enchancement factor in photovoltaic plant history output power data sequence, described
History output power data sequence is X(0), single order Accumulating generation sequence is X(1), it is as follows:
X(0)=(x(0)(1),x(0)(2),…,x(0)(n))
X(1)=(x(1)(1),x(1)(2),…,x(1)(n))
Wherein, x(0)It (j) is the element in photovoltaic plant history output power data sequence, j is that the element is exported in history
Serial number in power data sequence, j ∈ [1, n]；x(1)It (k) is the element in single order Accumulating generation sequence, k is the element one
Serial number in rank Accumulating generation sequence, k ∈ [1, n].
Step 3, the single order Accumulating generation sequence of the history output power data is used into variation mode decomposition method VMD
Resolve into the subsequence of finite bandwidth, each subsequence corresponds to different mode, respectively mode 1, mode 2 ..., mode q.
Preferably, the Decomposition order q of VMD needs external setting-up, and the present invention uses frequency spectrum analysis method, passes through the history
The mode of main frequency quantity, determines the Decomposition order q of VMD in the single order Accumulating generation sequence spectrum figure of output power data.
Step 4, support vector machines prediction mould is established respectively to each subsequence decomposited in step 3 by VMD
Type.
Kernel function K (the x of support vector machinesi, x) and type is determined as Radial basis kernel function RBF, as follows:
Wherein, xiAny point and a certain center in space are respectively represented with x, σ is the width parameter of Radial basis kernel function.
Step 5, it is sought using parameter of the whale colony optimization algorithm WSA to each support vector machines prediction model
Excellent, SVM parameter to be optimized is the width parameter σ of punishment parameter C, sensitivity loss parameter ε and Radial basis kernel function.
Step 6, summation reconstruct is carried out to the prediction result of each SVM model, it is tired obtains photovoltaic plant output power single order
Add prediction of result value sequence, is denoted asIt is as follows:
Step 7, the single order Accumulating generation sequence X of the history output power data is calculated(1)With photovoltaic plant output power
Single order accumulation result predicts value sequenceThe difference of the two is denoted as error sequence E, as follows:
E=(e (1), e (2) ..., e (n))
Step 8, error sequence E is predicted using GM (1, N) model, obtains the prediction value sequence of error, is denoted as
It is as follows:
Step 9, the output power single order accumulation result of the prediction is calculatedWith the prediction value sequence of errorThe two
Difference, and carry out regressive reduction obtains the prediction value sequence Y of target photovoltaic plant output power, as follows:
Y=(y (1), y (2) ..., y (n))
Wherein, y (1) is the 1st element in the prediction value sequence of target photovoltaic plant output power,With
The 1st element respectively in the prediction value sequence of photovoltaic plant output power single order accumulation result prediction value sequence and error；y
It (k) is k-th of element in the prediction value sequence of target photovoltaic plant output power, k is serial number,WithPoint
Not Wei photovoltaic plant output power single order accumulation result predict value sequence in kth and k-1 element,WithRespectively
For the kth and k-1 element in the prediction value sequence of error.
Step 10, relative error is calculated, the photovoltaic plant output power based on VMD-SVM-WSA-GM built-up pattern is examined
The precision of prediction of prediction technique.
In a kind of above-mentioned photovoltaic plant output power predicting method based on VMD-SVM-WSA-GM built-up pattern, institute
State the detailed process for carrying out optimizing in step 5 to the parameter of support vector machines prediction model using whale colony optimization algorithm WSA
It is as follows:
Step 501, the mathematical expression of support vector machines prediction model parameters optimization problem is established, as follows:
s.t.Cmin≤Cp≤Cmax
εmin≤εp≤εmax
σmin≤σp≤σmax
1≤p≤q
Wherein, Cp、εpAnd σpThe punishment parameter of respectively p-th SVM prediction model, sensitive loss parameter and radial base core
The parameter of function, CminAnd Cmax、εminAnd εmax、σminAnd σmaxPunishment parameter C, sensitivity loss parameter ε and radial base core letter respectively
Value lower limit/upper limit of several width parameter σ.
The optimization aim of the Parametric optimization problem is to minimize the sum of error sequence E, i.e. history output power data
Single order Accumulating generation sequence X(1)Value sequence is predicted with photovoltaic plant output power single order accumulation resultRespective components in the two
The sum of difference minimum；The constraint condition of the Parametric optimization problem be the value range that should meet of parameter of SVM prediction model about
Beam；The optimized variable of the Parametric optimization problem is the row vector that the parameter of q SVM prediction model is arranged to make up, as follows:
C1ε1σ1C2ε2σ2…Cpεpσp…Cqεqσq
Step 502, whale group Ω in whale group algorithm is initialized, population scale popsize is taken as 100, and maximum is evolved generation
Number maxgen is taken as 10000.
Step 503, whale position in whale group is initialized.
Step 504, each whale individual of first evaluation, calculates its fitness value.
Step 505, to the current individual Ω in whale groupi, the whale W of " more excellent and nearest " is found, if W exists,
ΩiW is shifted to, and calculates Ω after movementiFitness value.
Step 506, judge whether to have traversed all Ω in current whale groupi, it is to continue in next step；It is no, jump to step
Rapid 505.
Step 507, judge whether to have reached preset maximum evolutionary generation maxgen, be to continue in next step；It is no,
Evolutionary generation gos to step 505 from increasing 1.
Step 508, optimization calculates and terminates, and exports the Optimal Parameters of each SVM prediction model.
Compared with traditional technology, the invention has the following advantages that
1, innovative Combined model forecast scheme: since photovoltaic plant output power shows strong randomness and fluctuation
Property, great error will be present using single model prediction.Of the invention innovative application variation mode decomposition VMD, support to
The Predicting Technique scheme of amount machine SVM, whale colony optimization algorithm WSA and the combination of tetra- kinds of models of gray model GM, give full play to each mould
The advantage of type method is adapted to the demand of the accurate forecasting problem of photovoltaic plant output power.
2, compared with common empirical mode decomposition EMD, variation mode decomposition VMD is existing when EMD can be overcome to decompose
Modal overlap phenomenon and end effect；The present invention also utilizes the single order Accumulating generation in gray theory before executing VMD and decomposing
Weaken the influence of enchancement factor in photovoltaic plant history output power data sequence.
3, as the main structure of prediction technique, the invention proposes a kind of pair of photovoltaic plant history output power data sequences
Each mode that column do after VMD decomposition establishes support vector machines prediction model respectively, the new construction for reconstruct of then summing,
The learning ability and generalization ability of support vector machines can be given full play to, precision of prediction and convergence rate are improved.
4, the parameter of the support vector machines prediction model then passes through the complete optimization problem mathematical model of building and solves
It obtains, method for solving uses meta-heuristic searching algorithm --- the whale colony optimization algorithm WSA of the newest proposition of academia.
5, finally, the present invention is also proposed using gray model GM (1,1) to by the support vector machines model prediction
The error of the photovoltaic plant output power obtained with summation reconstruct is predicted, and is exporting final photovoltaic plant output power
The method for subtracting the error before predicted value, is further noted that precision of prediction.
Advantageous effects of the invention:
The mentioned method of the present invention is realized by VMD-SVM-WSA-GM built-up pattern to photovoltaic plant short-term output power
Accurate prediction, facilitate it is grid-connected after electric system scientific dispatch decision, improve the stability of Operation of Electric Systems.
Detailed description of the invention
Fig. 1 is a kind of photovoltaic plant output based on VMD-SVM-WSA-GM built-up pattern that the embodiment of the present invention one provides
The general frame of power forecasting method.
Fig. 2 is application whale colony optimization algorithm WSA provided by Embodiment 2 of the present invention to support vector machines prediction model
Parameter carry out optimizing flow chart.
Fig. 3 is mentioned obtained by built-up pattern predicts photovoltaic power by what the embodiment of the present invention three provided using the present invention
The comparison diagram of regression forecasting data and initial data.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, with reference to the accompanying drawings and detailed description
The present invention is described in further detail.
Embodiment one
As shown in Figure 1, a kind of photovoltaic plant output based on VMD-SVM-WSA-GM built-up pattern mentioned for the present invention
The general frame of power forecasting method.It should be noted that the method be a kind of Time Series Forecasting Methods, with based on influence
The method of factor (such as solar irradiation intensity, environment temperature, wind speed factor) prediction photovoltaic system output power has the area of essence
Not.Prediction technique based on influence factor is to pass through prediction model using multiple influence factor data sequences as multidimensional input quantity
Obtain corresponding photovoltaic plant output power predicted value.And the mentioned Time Series Forecasting Methods of the present invention be set a length as
" the period window " of m is tieed up input quantity for m output power value before predicted time point as m, is worked as by prediction model
Preceding photovoltaic plant output power predicted value；Before predicting next time point, " period window " is also correspondingly slided to the right
One chronomere, then take m output power value in " the period window " as input quantity, above process iteration carry out until
Prediction terminates.Preferably, the length of " period window " is taken as 6~10 in the present invention.
A kind of specific steps of the photovoltaic plant output power predicting method based on VMD-SVM-WSA-GM built-up pattern are such as
Under:
Step 1, target photovoltaic plant history output power data are obtained, the history output power data were with 1 hour
(1h) is unit, is arranged in time series by the sampling time.According to the practical feelings of target photovoltaic plant output power forecasting problem
Condition, history output power data can also be less than 1 hour or a few hours are sample frequency.
Step 2, the history output power data sequence is pre-processed, pretreatment measure includes changing and one at equal intervals
Rank Accumulating generation, specific as follows:
Change the data for some output power sampled point of completion missing at equal intervals, takes sampled point front and back to number in fixed step size
According to average value, it is preferred that the step-length is taken as 3~6.
Single order Accumulating generation is used to eliminate the influence of enchancement factor in photovoltaic plant history output power data sequence, described
History output power data sequence is X(0), single order Accumulating generation sequence is X(1), it is as follows:
X(0)=(x(0)(1),x(0)(2),…,x(0)(n))
X(1)=(x(1)(1),x(1)(2),…,x(1)(n))
Wherein, x(0)It (j) is the element in photovoltaic plant history output power data sequence, j is that the element is exported in history
Serial number in power data sequence, j ∈ [1, n]；x(1)It (k) is the element in single order Accumulating generation sequence, k is the element one
Serial number in rank Accumulating generation sequence, k ∈ [1, n].
Step 3, the single order Accumulating generation sequence of the history output power data is used into variation mode decomposition method VMD
Resolve into the subsequence of finite bandwidth, each subsequence corresponds to different mode, respectively mode 1, mode 2 ..., mode q.
Preferably, the Decomposition order q of VMD needs external setting-up, and the present invention uses frequency spectrum analysis method, passes through the history
The mode of main frequency quantity, determines the Decomposition order q of VMD in the single order Accumulating generation sequence spectrum figure of output power data.
Step 4, support vector machines prediction mould is established respectively to each subsequence decomposited in step 3 by VMD
Type.
Preferably, the kernel function K (x of support vector machinesi, x) and type is determined as Radial basis kernel function RBF, as follows:
Wherein, xiAny point and a certain center in space are respectively represented with x, σ is the width parameter of Radial basis kernel function.
Step 5, it is sought using parameter of the whale colony optimization algorithm WSA to each support vector machines prediction model
Excellent, SVM parameter to be optimized is the parameter σ of punishment parameter C, sensitivity loss parameter ε and Radial basis kernel function.
Step 6, summation reconstruct is carried out to the prediction result of each SVM model, it is tired obtains photovoltaic plant output power single order
Add prediction of result value sequence, is denoted asIt is as follows:
Step 7, the single order Accumulating generation sequence X of the history output power data is calculated(1)With photovoltaic plant output power
Single order accumulation result predicts value sequenceThe difference of the two is denoted as error sequence E, as follows:
E=(e (1), e (2) ..., e (n))
Step 8, error sequence E is predicted using GM (1, N) model, obtains the prediction value sequence of error, is denoted as
It is as follows:
Step 9, the output power single order accumulation result of the prediction is calculatedWith the prediction value sequence of errorThe two
Difference, and carry out regressive reduction obtains the prediction value sequence Y of target photovoltaic plant output power, as follows:
Y=(y (1), y (2) ..., y (n))
Wherein, y (1) is the 1st element in the prediction value sequence of target photovoltaic plant output power,With
The 1st element respectively in the prediction value sequence of photovoltaic plant output power single order accumulation result prediction value sequence and error；y
It (k) is k-th of element in the prediction value sequence of target photovoltaic plant output power, k is serial number,WithPoint
Not Wei photovoltaic plant output power single order accumulation result predict value sequence in kth and k-1 element,WithRespectively
For the kth and k-1 element in the prediction value sequence of error.
Step 10, relative error is calculated, the photovoltaic plant output power based on VMD-SVM-WSA-GM built-up pattern is examined
The precision of prediction of prediction technique.
It should be noted that the photovoltaic plant output power historical data that the present invention will acquire is divided into training set and test
Collection, in which: training set mentions the built-up pattern based on VMD-SVM-WSA-GM for establishing the present invention, and test set is for testing inspection
The precision of prediction of the built-up pattern is tested, after training process and test process are completed, can be carried out using the built-up pattern
Prediction work.1~step 10 of abovementioned steps is the entire flow for training set and training process, to test process and is applied
Cheng Eryan, it is only necessary to according to the established model group comprising q support vector machines prediction model, input m correspondence respectively
The modal components of the output power of different moments obtain the predicted value subsequence of q different modalities, then summation reconstruct, then plus
On the error prediction value that is obtained via gray model GM (1,1), finally obtain the prediction value sequence of photovoltaic plant output power.
Embodiment two
As shown in Fig. 2, to apply whale colony optimization algorithm WSA to the ginseng of support vector machines prediction model in the present invention
Number carries out the flow chart of optimizing, is described in detail below:
Step 501, the mathematical expression of support vector machines prediction model parameters optimization problem is established, as follows:
s.t.Cmin≤Cp≤Cmax
εmin≤εp≤εmax
σmin≤σp≤σmax
1≤p≤q
Wherein, Cp、εpAnd σpThe punishment parameter of respectively p-th SVM prediction model, sensitive loss parameter and radial base core
The parameter of function, CminAnd Cmax、εminAnd εmax、σminAnd σmaxPunishment parameter C, sensitivity loss parameter ε and radial base core letter respectively
Value lower limit/upper limit of several width parameter σ.
The optimization aim of the Parametric optimization problem is to minimize the sum of error sequence E, i.e. history output power data
Single order Accumulating generation sequence X(1)Value sequence is predicted with photovoltaic plant output power single order accumulation resultRespective components in the two
The sum of difference minimum；The constraint condition of the Parametric optimization problem be the value range that should meet of parameter of SVM prediction model about
Beam；The optimized variable of the Parametric optimization problem is the row vector that the parameter of q SVM prediction model is arranged to make up, as follows:
C1ε1σ1C2ε2σ2…Cpεpσp…Cqεqσq
Step 502, whale group Ω in whale group algorithm is initialized, population scale popsize is taken as 100, and maximum is evolved generation
Number maxgen is taken as 10000.
Step 503, whale position in whale group is initialized.
Step 504, each whale individual of first evaluation, calculates its fitness value.
Step 505, to the current individual Ω in whale groupi, the whale W of " more excellent and nearest " is found, if W exists,
ΩiW is shifted to, and calculates Ω after movementiFitness value.
Step 506, judge whether to have traversed all Ω in current whale groupi, it is to continue in next step；It is no, jump to step
Rapid 505.
Step 507, judge whether to have reached preset maximum evolutionary generation maxgen, be to continue in next step；It is no,
Evolutionary generation gos to step 505 from increasing 1.
Step 508, optimization calculates and terminates, and exports the Optimal Parameters of each SVM prediction model.
Embodiment three
Fig. 3 is mentioned obtained by built-up pattern predicts photovoltaic power by what the embodiment of the present invention three provided using the present invention
The comparison diagram of regression forecasting data and initial data.
In order to verify the validity of institute climbing form type and method of the present invention, it is applied to the output work of certain practical photovoltaic plant
Rate forecasting problem.Using 480 groups of output power data of the photovoltaic plant as data set.Every group of data correspond to photovoltaic plant every
Output power sampled value in 30 minutes (min) since night photovoltaic plant output power is ignored, thus need to be rejected daily
Data corresponding with the period at night in 24 hours, only take daytime 12 hours output powers, and data daily in this way are 24 groups.
480 group data sets are further divided into training set and test set, separately include 360 groups and 120 groups of data.The specific mistake of application
Journey are as follows: this paper institute climbing form type is trained with training set first, then the test and verification on test set.
From the figure 3, it may be seen that institute's climbing form type and method obtain according to the present invention in 5 days corresponding with test set (120 groups of data)
To regression forecasting data compared with initial data, maintain high consistency, regression forecasting data have reproduced original well
The fluctuation pattern of beginning data, and the two deviation very little.The deviation of preceding 20 groups of data has been intercepted below, as shown in table 1.
The deviation of 1 regression forecasting data of table and initial data (containing absolute error and relative error)
As can be seen from the above table, the precision of prediction of institute's climbing form type and method of the present invention is very high, in above-mentioned 20 groups of data,
In addition to first three groups, the relative error of remaining all data is respectively less than 3.67%, and has the relative error of 10 groups of data to be less than
1%.This shows the validity and superiority of institute's climbing form type and method of the present invention.
In conclusion method disclosed by the invention is directed to the strong randomness and fluctuation that photovoltaic system output power shows
Feature provides a kind of prediction technique based on VMD-SVM-WSA-GM built-up pattern, realizes and export in short term to photovoltaic plant
The accurate prediction of power facilitates the scientific dispatch decision of grid-connected rear electric system, improves the stabilization of Operation of Electric Systems
Property.
Those of ordinary skill in the art be further appreciated that implement the method for the above embodiments be can
It is completed with instructing relevant hardware by program, the program can store in computer-readable storage medium,
Described storage medium, including ROM/RAM, disk, CD etc..
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Made any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.
Claims (2)
1. a kind of photovoltaic plant output power predicting method based on VMD-SVM-WSA-GM built-up pattern, which is characterized in that root
According to photovoltaic plant history output power data, its generated output value in future time period is predicted comprising the steps of:
Step 1, target photovoltaic plant history output power data are obtained, the history output power data are single with setting time
Position, is arranged in time series by the sampling time；；
Step 2, the history output power data sequence is pre-processed, pretreatment measure includes changing to tire out with single order at equal intervals
Add generation, specific as follows:
Change the data for some output power sampled point of completion missing at equal intervals, takes sampled point front and back to data in fixed step size
Average value, it is preferred that the step-length is taken as 3~6；
Single order Accumulating generation is used to eliminate the influence of enchancement factor in photovoltaic plant history output power data sequence, the history
Output power data sequence is X(0), single order Accumulating generation sequence is X(1), it is as follows:
X(0)=(x(0)(1),x(0)(2),…,x(0)(n))
X(1)=(x(1)(1),x(1)(2),…,x(1)(n))
Wherein, x(0)It (j) is the element in photovoltaic plant history output power data sequence, j is the element in history output power
Serial number in data sequence, j ∈ [1, n]；x(1)It (k) is the element in single order Accumulating generation sequence, k is that the element is tired in single order
Add the serial number in formation sequence, k ∈ [1, n]；
Step 3, the single order Accumulating generation sequence of the history output power data is decomposed using variation mode decomposition method VMD
At the subsequence of finite bandwidth, each subsequence corresponds to different mode, respectively mode 1, mode 2 ..., mode q；
Preferably, the Decomposition order q of VMD needs external setting-up, and the present invention uses frequency spectrum analysis method, exported by the history
The mode of main frequency quantity, determines the Decomposition order q of VMD in the single order Accumulating generation sequence spectrum figure of power data；
Step 4, support vector machines prediction model is established to each subsequence decomposited in step 3 by VMD respectively, is propped up
Hold the kernel function K (x of vector machinei, x) and type is determined as Radial basis kernel function RBF, as follows:
Wherein, xiAny point and a certain center in space are respectively represented with x, σ is the width parameter of Radial basis kernel function；
Step 5, optimizing is carried out using parameter of the whale colony optimization algorithm WSA to each support vector machines prediction model, to
The SVM parameter of optimization is the width parameter σ of punishment parameter C, sensitivity loss parameter ε and Radial basis kernel function；
Step 6, summation reconstruct is carried out to the prediction result of each SVM model, obtains the cumulative knot of photovoltaic plant output power single order
Fruit predicts value sequence, is denoted asIt is as follows:
Step 7, the single order Accumulating generation sequence X of the history output power data is calculated(1)With photovoltaic plant output power single order
Accumulation result predicts value sequenceThe difference of the two is denoted as error sequence E, as follows:
E=(e (1), e (2) ..., e (n))
Step 8, error sequence E is predicted using GM (1, N) model, obtains the prediction value sequence of error, is denoted asIt is as follows:
Step 9, the output power single order accumulation result of the prediction is calculatedWith the prediction value sequence of errorThe difference of the two
Value, and carries out regressive reduction, obtains the prediction value sequence Y of target photovoltaic plant output power, as follows:
Y=(y (1), y (2) ..., y (n))
Wherein, y (1) is the 1st element in the prediction value sequence of target photovoltaic plant output power,WithRespectively
Photovoltaic plant output power single order accumulation result predicts the 1st element in value sequence and the prediction value sequence of error；Y (k) is
K-th of element in the prediction value sequence of target photovoltaic plant output power, k is serial number,WithRespectively
Photovoltaic plant output power single order accumulation result predicts kth and k-1 element in value sequence,WithRespectively miss
Kth and k-1 element in the prediction value sequence of difference；
Step 10, relative error is calculated, the photovoltaic plant output power prediction based on VMD-SVM-WSA-GM built-up pattern is examined
The precision of prediction of method.
2. a kind of photovoltaic plant output power prediction based on VMD-SVM-WSA-GM built-up pattern according to claim 1
Method, which is characterized in that using whale colony optimization algorithm WSA to the parameter of support vector machines prediction model in the step 5
The detailed process for carrying out optimizing is as follows:
Step 501, the mathematical expression of support vector machines prediction model parameters optimization problem is established, as follows:
s.t.Cmin≤Cp≤Cmax
εmin≤εp≤εmax
σmin≤σp≤σmax
1≤p≤q
Wherein, Cp、εpAnd σpThe punishment parameter of respectively p-th SVM prediction model, sensitive loss parameter and Radial basis kernel function
Parameter, CminAnd Cmax、εminAnd εmax、σminAnd σmaxDistinguish punishment parameter C, sensitivity loss parameter ε and Radial basis kernel function
Value lower limit/upper limit of width parameter σ；
The optimization aim of the Parametric optimization problem is to minimize the sum of error sequence E, the i.e. single order of history output power data
Accumulating generation sequence X(1)Value sequence is predicted with photovoltaic plant output power single order accumulation resultRespective components difference in the two
The sum of minimum；The constraint condition of the Parametric optimization problem is the value range constraint that the parameter of SVM prediction model should meet；Institute
The row vector that the optimized variable of Parametric optimization problem is arranged to make up for the parameter of q SVM prediction model is stated, as follows:
C1ε1σ1C2ε2σ2…Cpεpσp…Cqεqσq
Step 502, whale group Ω in whale group algorithm is initialized, population scale popsize is taken as 100, maximum evolutionary generation
Maxgen is taken as 10000；
Step 503, whale position in whale group is initialized；
Step 504, each whale individual of first evaluation, calculates its fitness value；
Step 505, to the current individual Ω in whale groupi, the whale W of " more excellent and nearest " is found, if W exists, ΩiIt moves
To W, and calculate Ω after movementiFitness value；
Step 506, judge whether to have traversed all Ω in current whale groupi, it is to continue in next step；It is no, it gos to step
505；
Step 507, judge whether to have reached preset maximum evolutionary generation maxgen, be to continue in next step；It is no, it evolves
Algebra gos to step 505 from increasing 1；
Step 508, optimization calculates and terminates, and exports the Optimal Parameters of each SVM prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910351027.8A CN110059891B (en) | 2019-04-28 | 2019-04-28 | Photovoltaic power station output power prediction method based on VMD-SVM-WSA-GM combined model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910351027.8A CN110059891B (en) | 2019-04-28 | 2019-04-28 | Photovoltaic power station output power prediction method based on VMD-SVM-WSA-GM combined model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110059891A true CN110059891A (en) | 2019-07-26 |
| CN110059891B CN110059891B (en) | 2023-06-06 |
Family
ID=67321358
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910351027.8A Active CN110059891B (en) | 2019-04-28 | 2019-04-28 | Photovoltaic power station output power prediction method based on VMD-SVM-WSA-GM combined model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110059891B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062533A (en) * | 2019-12-16 | 2020-04-24 | 国家能源集团谏壁发电厂 | Fan fault prediction method based on whale optimization algorithm optimization weighted least square support vector machine |
| CN111795826A (en) * | 2020-06-29 | 2020-10-20 | 南京航空航天大学 | Fault diagnosis method for abnormal oil injection of small two-stroke piston engine |
| CN112613650A (en) * | 2020-12-14 | 2021-04-06 | 燕山大学 | Novel three-stage short-term power load prediction method |
| CN113052330A (en) * | 2021-03-18 | 2021-06-29 | 淮北师范大学 | Ox penis effect weakening method based on VMD-SVM algorithm |
| CN113326969A (en) * | 2021-04-29 | 2021-08-31 | 淮阴工学院 | Short-term wind speed prediction method and system based on improved whale algorithm for optimizing ELM |
| CN113688869A (en) * | 2021-07-21 | 2021-11-23 | 广东工业大学 | Photovoltaic data missing reconstruction method based on generation countermeasure network |
| CN116167508A (en) * | 2023-01-16 | 2023-05-26 | 南通电力设计院有限公司 | Short-term photovoltaic output rapid prediction method and system based on meteorological factor decomposition |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018082132A1 (en) * | 2016-11-04 | 2018-05-11 | 江南大学 | Short-period prediction method for output power of energy system |
| CN108985521A (en) * | 2018-08-02 | 2018-12-11 | 广东工业大学 | Output power predicting method, device, equipment and the medium of photovoltaic generating system |
-
2019
- 2019-04-28 CN CN201910351027.8A patent/CN110059891B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018082132A1 (en) * | 2016-11-04 | 2018-05-11 | 江南大学 | Short-period prediction method for output power of energy system |
| CN108985521A (en) * | 2018-08-02 | 2018-12-11 | 广东工业大学 | Output power predicting method, device, equipment and the medium of photovoltaic generating system |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062533A (en) * | 2019-12-16 | 2020-04-24 | 国家能源集团谏壁发电厂 | Fan fault prediction method based on whale optimization algorithm optimization weighted least square support vector machine |
| CN111795826A (en) * | 2020-06-29 | 2020-10-20 | 南京航空航天大学 | Fault diagnosis method for abnormal oil injection of small two-stroke piston engine |
| CN111795826B (en) * | 2020-06-29 | 2021-12-21 | 南京航空航天大学 | Fault diagnosis method for abnormal oil injection of small two-stroke piston engine |
| CN112613650A (en) * | 2020-12-14 | 2021-04-06 | 燕山大学 | Novel three-stage short-term power load prediction method |
| CN113052330A (en) * | 2021-03-18 | 2021-06-29 | 淮北师范大学 | Ox penis effect weakening method based on VMD-SVM algorithm |
| CN113052330B (en) * | 2021-03-18 | 2022-08-02 | 淮北师范大学 | Ox penis effect weakening method based on VMD-SVM algorithm |
| CN113326969A (en) * | 2021-04-29 | 2021-08-31 | 淮阴工学院 | Short-term wind speed prediction method and system based on improved whale algorithm for optimizing ELM |
| CN113326969B (en) * | 2021-04-29 | 2023-06-20 | 淮阴工学院 | Short-term wind speed prediction method and system based on improved whale optimization algorithm and ELM |
| CN113688869A (en) * | 2021-07-21 | 2021-11-23 | 广东工业大学 | Photovoltaic data missing reconstruction method based on generation countermeasure network |
| CN113688869B (en) * | 2021-07-21 | 2022-05-27 | 广东工业大学 | Photovoltaic data missing reconstruction method based on generation countermeasure network |
| CN116167508A (en) * | 2023-01-16 | 2023-05-26 | 南通电力设计院有限公司 | Short-term photovoltaic output rapid prediction method and system based on meteorological factor decomposition |
| CN116167508B (en) * | 2023-01-16 | 2024-03-15 | 南通电力设计院有限公司 | Short-term photovoltaic output rapid prediction method and system based on meteorological factor decomposition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110059891B (en) | 2023-06-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059891A (en) | A kind of photovoltaic plant output power predicting method based on VMD-SVM-WSA-GM built-up pattern | |
| CN109858673A (en) | A kind of photovoltaic generating system power forecasting method | |
| CN109242204A (en) | Ultra-short term wind speed forecasting method based on optimal VMD and Synchronous fluorimetry | |
| CN102129511B (en) | System for forecasting short-term wind speed of wind power station based on MATLAB | |
| CN103106544B (en) | A kind of photovoltaic generation prognoses system based on T-S Fuzzy neutral net | |
| Guo et al. | Study on short-term photovoltaic power prediction model based on the Stacking ensemble learning | |
| CN109948833A (en) | A kind of Hydropower Unit degradation trend prediction technique based on shot and long term memory network | |
| CN106295899B (en) | Wind power probability density Forecasting Methodology based on genetic algorithm Yu supporting vector quantile estimate | |
| CN104638672B (en) | Determining method of photovoltaic transmission power limit considering variable correlation | |
| CN107292516B (en) | Load reliability assessment method considering load classification and energy scheduling | |
| CN107103154A (en) | A kind of photovoltaic module model parameter identification method | |
| Zhang et al. | Wavelet-based neural network with genetic algorithm optimization for generation prediction of PV plants | |
| CN114123294B (en) | Multi-target photovoltaic single-phase grid-connected capacity planning method considering three-phase unbalance | |
| Chen et al. | An improved brain storm optimization for a hybrid renewable energy system | |
| CN107276093A (en) | The Probabilistic Load computational methods cut down based on scene | |
| CN111401659A (en) | Ultra-short-term or short-term photovoltaic power generation power prediction method based on case reasoning | |
| CN116128211A (en) | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene | |
| CN110276478A (en) | Short-term wind power forecast method based on segmentation ant group algorithm optimization SVM | |
| Zhang et al. | Operation state assessment of wind power system based on pso+ ahp—fce | |
| CN109800924A (en) | A kind of Optimal Configuration Method and device mixing renewable energy system | |
| CN114595891A (en) | Power distribution network voltage and power flow boundary crossing risk assessment method, system and equipment | |
| Shen et al. | PV power interval prediction based on EEMD-LSTM method | |
| Mouton et al. | Assessment of methods to measure power system flexibility: a review | |
| CN107895206A (en) | A kind of multistep wind energy Forecasting Methodology based on singular spectrum analysis and local sensitive hash | |
| Zeng et al. | Photovoltaic Power Generation Prediction Based on Correlation Analysis and GA-BP Neural Network |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |