CN110059891B - Photovoltaic power station output power prediction method based on VMD-SVM-WSA-GM combined model - Google Patents
Photovoltaic power station output power prediction method based on VMD-SVM-WSA-GM combined model Download PDFInfo
- Publication number
- CN110059891B CN110059891B CN201910351027.8A CN201910351027A CN110059891B CN 110059891 B CN110059891 B CN 110059891B CN 201910351027 A CN201910351027 A CN 201910351027A CN 110059891 B CN110059891 B CN 110059891B
- Authority
- CN
- China
- Prior art keywords
- output power
- sequence
- photovoltaic power
- power station
- svm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000010248 power generation Methods 0.000 claims abstract description 10
- 238000012706 support-vector machine Methods 0.000 claims description 47
- 238000009825 accumulation Methods 0.000 claims description 44
- 238000005457 optimization Methods 0.000 claims description 36
- 241000283153 Cetacea Species 0.000 claims description 34
- 230000006870 function Effects 0.000 claims description 18
- 238000000354 decomposition reaction Methods 0.000 claims description 14
- 238000005070 sampling Methods 0.000 claims description 11
- 230000035945 sensitivity Effects 0.000 claims description 9
- 230000008569 process Effects 0.000 claims description 8
- 238000007781 pre-processing Methods 0.000 claims description 6
- 239000013598 vector Substances 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 3
- 230000009467 reduction Effects 0.000 claims description 3
- 238000010183 spectrum analysis Methods 0.000 claims description 3
- 230000009191 jumping Effects 0.000 claims 1
- 230000001502 supplementing effect Effects 0.000 claims 1
- 238000012360 testing method Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 238000003860 storage Methods 0.000 description 2
- 229910052799 carbon Inorganic materials 0.000 description 1
- 238000005265 energy consumption Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 239000003345 natural gas Substances 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013082 photovoltaic technology Methods 0.000 description 1
- 230000005855 radiation Effects 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Images
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—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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Operations Research (AREA)
- Primary Health Care (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Medical Informatics (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Control Of Electrical Variables (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Photovoltaic Devices (AREA)
Abstract
The invention belongs to the technical field of photovoltaic power generation, and particularly relates to a photovoltaic power station output power prediction method based on a VMD-SVM-WSA-GM combined model.
Description
Technical Field
The invention belongs to the technical field of photovoltaic power generation, and particularly relates to a photovoltaic power station output power prediction method based on a time sequence and an intelligent optimization algorithm, in particular to a photovoltaic power station output power prediction method based on a combination of several models, namely a Variational Modal Decomposition (VMD), a Support Vector Machine (SVM), a whale swarm optimization algorithm (WSA) and a Gray Model (GM).
Background
Photovoltaic is a short term of solar photovoltaic power generation system, and is a novel power generation system for directly converting solar radiation energy into electric energy by utilizing the photovoltaic effect of solar cell semiconductor materials. Photovoltaic technology offers many advantages: no other forms of energy medium are consumed except for the directly acquired illumination; equipment such as large-scale rotating machinery is not needed; the arrangement is flexible, and the existing open space, roof and the like can be utilized.
Due to the continuous fluctuation of political and economic situations of major oil and natural gas export countries in the world, each major energy consumption country is actively seeking an alternative scheme to reduce the dependence on external energy and ensure the safety of national folk life. It is against this background that solar energy is becoming increasingly widely utilized as a clean, inexpensive, plentiful and safe energy source. The data shows that by the end of 11 months in 2017, the accumulated installed capacity of the photovoltaic in China reaches 12579 kilowatts, the accumulated installed capacity is increased by 67% in the same ratio, and the accumulated installed capacity accounts for 7.5% of the total installed power; in 2017, photovoltaic annual energy production exceeded 1000 hundred million kilowatt-hours. The international energy agency, world energy prospect 2017 China special report, believes that China energy structure will gradually convert to clean power generation, wherein solar photovoltaic will become the most economical power generation mode in China, and the low-carbon installed capacity led by waterpower, wind energy and solar photovoltaic will rapidly increase, and will account for 60% of the total installed capacity by 2040 years.
The output power of the photovoltaic power station has typical randomness and fluctuation due to the influence of complex factors such as solar irradiation intensity, environment and the like, and the uncertainty directly influences the stability of the grid-connected power system. Therefore, it is very necessary to accurately predict the output power of the photovoltaic power station in a future period of time, which is a problem to be solved by those skilled in the relevant field.
Disclosure of Invention
The invention aims to provide a photovoltaic power station output power prediction method based on a VMD-SVM-WSA-GM combined model, so as to improve the accuracy of short-term output power prediction of a photovoltaic power generation system and realize scientific scheduling decision and optimized operation of the photovoltaic power generation system and a power system.
In order to solve the technical problems, the embodiment of the invention provides the following technical scheme:
the photovoltaic power station output power prediction method based on the VMD-SVM-WSA-GM combined model is characterized by predicting the power generation value of the photovoltaic power station in a future period according to historical output power data of the photovoltaic power station, and comprises the following steps:
step 1, acquiring historical output power data of a target photovoltaic power station, wherein the historical output power data are arranged into a time sequence according to sampling time by taking set time as a unit. .
Step 2, preprocessing the historical output power data sequence, wherein preprocessing measures comprise equidistant generation and first-order accumulation generation, and specifically comprise the following steps:
the equidistant data used for complementing the missing data of a certain output power sampling point is obtained by taking the average value of the data in a given step length before and after the sampling point, and preferably, the step length is 3-6.
Generating a first-order accumulation for eliminating the influence of random factors in a historical output power data sequence of the photovoltaic power station, wherein the historical output power data sequence is X (0) The first-order accumulation generation sequence is X (1) The following are provided:
X (0) ＝(x (0) (1),x (0) (2),…,x (0) (n))
X (1) ＝(x (1) (1),x (1) (2),…,x (1) (n))
wherein x is (0) (j) For the element in the historical output power data sequence of the photovoltaic power station, j is the sequence number of the element in the historical output power data sequence, j is [1, n ]]；x (1) (k) For the first-order accumulation generating sequence, k is the sequence number of the element in the first-order accumulation generating sequence, and k is E [1, n]。
And 3, decomposing the first-order accumulation generation sequence of the historical output power data into subsequences with limited bandwidth by adopting a variation mode decomposition method VMD, wherein each subsequence corresponds to different modes, namely a mode 1, a mode 2, a mode … and a mode q.
Preferably, the number q of decomposition layers of the VMD needs to be set externally, and the invention adopts a frequency spectrum analysis method to determine the number q of decomposition layers of the VMD in a mode of generating the number of main frequencies in a sequence frequency spectrum chart by first-order accumulation of the historical output power data.
And 4, respectively establishing a Support Vector Machine (SVM) prediction model for each subsequence decomposed by the VMD in the step 3.
Kernel function K (x) of support vector machine i X) type is determined as radial basis function RBF as follows:
wherein x is i And x represents any point in space and a certain center, respectively, and sigma is the width parameter of the radial basis function.
And 5, optimizing parameters of each support vector machine SVM prediction model by using a whale swarm optimization algorithm WSA, wherein the SVM parameters to be optimized are a penalty parameter C, a sensitivity loss parameter epsilon and a width parameter sigma of a radial basis function.
Step 6, carrying out summation reconstruction on the prediction results of each SVM model to obtain a first-order accumulation result prediction value sequence of the output power of the photovoltaic power station, and marking the first-order accumulation result prediction value sequence as
The following are provided:
step 7, calculating a first-order accumulation generation sequence X of the historical output power data (1) And photovoltaic power station output power first-order accumulation result predicted value sequence
The difference between the two is noted as error sequence E as follows:
E＝(e(1),e(2),…,e(n))
step 8, predicting the error sequence E by using a GM (1, N) model to obtainTo the predicted value sequence of the error, recorded as
The following are provided:
step 9, calculating the first-order accumulation result of the predicted output power
And error prediction value sequence +.>
And carrying out subtraction reduction on the difference value of the two to obtain a predicted value sequence Y of the output power of the target photovoltaic power station, wherein the predicted value sequence Y is as follows:
Y＝(y(1),y(2),…,y(n))
wherein y (1) is the 1 st element in the predicted value sequence of the output power of the target photovoltaic power station,
and->
The 1 st element in the predicted value sequence of the first-order accumulated result and the predicted value sequence of the error of the output power of the photovoltaic power station are respectively; y (k) is the kth element in the predicted value sequence of the output power of the target photovoltaic power station, k is the serial number, ">
And->
Respectively, the first-order accumulation results of the output power of the photovoltaic power station are pre-determinedK and k-1 elements of the measured sequence,/and>
and->
K and k-1 elements in the predicted value sequence of the error, respectively.
And step 10, calculating a relative error, and checking the prediction accuracy of the photovoltaic power station output power prediction method based on the VMD-SVM-WSA-GM combined model.
In the above method for predicting output power of photovoltaic power station based on VMD-SVM-WSA-GM combined model, the detailed flow of optimizing parameters of support vector machine SVM prediction model by applying whale swarm optimization algorithm WSA in step 5 is as follows:
step 501, establishing a mathematical expression of a support vector machine SVM prediction model parameter optimization problem, as follows:
s.t.C min ≤C p ≤C max
ε min ≤ε p ≤ε max
σ min ≤σ p ≤σ max
1≤p≤q
wherein C is p 、ε p Sum sigma p Penalty parameter, sensitivity loss parameter and parameter of radial basis function of p-th SVM predictive model respectively, C min And C max 、ε min And epsilon max 、σ min Sum sigma max The penalty parameter C, the sensitivity loss parameter epsilon and the width parameter sigma of the radial basis function are respectively lower/upper valued limits.
The optimization objective of the parameter optimization problem is to minimize the sum of error sequences E, i.e. the first-order accumulation of historical output power data generates a sequence X (1) First-order accumulation of output power of photovoltaic power stationResult predictor sequence
The sum of the difference values of the corresponding components is the smallest; the constraint condition of the parameter optimization problem is the value range constraint which is satisfied by the parameters of the SVM prediction model; the optimization variables of the parameter optimization problem are row vectors formed by parameter arrangement of q SVM prediction models, and the row vectors are as follows:
C 1 ε 1 σ 1 C 2 ε 2 σ 2 …C p ε p σ p …C q ε q σ q
step 502, initializing whale shoal omega in whale shoal algorithm, taking the population size pop as 100, and taking the maximum evolutionary algebra maxgen as 10000.
Step 503, initializing whale positions in a whale population.
Step 504, each whale individual is first evaluated and its fitness value is calculated.
Step 505, omega for the current individual in whale population i Searching for "preferred and most recent" whale W, if W is present, Ω i Move to W and calculate omega after the move i Is used for the adaptation value of the (c).
Step 506, determining whether all Ω -fish in the current whale are traversed i If yes, continuing the next step; no, go to step 505.
Step 507, judging whether a preset maximum evolution algebra maxgen is reached, if yes, continuing the next step; if not, the evolution algebra self-increases by 1 and jumps to step 505.
And step 508, after the optimization calculation is finished, outputting the optimization parameters of each SVM prediction model.
Compared with the prior art, the invention has the following advantages:
1. innovative combined model prediction scheme: because the output power of the photovoltaic power station presents strong randomness and fluctuation, a single model prediction is adopted to have a great error. The invention creatively applies the prediction technical scheme of combination of four models of variational modal decomposition VMD, support vector machine SVM, whale swarm optimization algorithm WSA and gray model GM, fully exerts the advantages of each model method, and meets the requirement of accurate prediction of the photovoltaic power station output power.
2. Compared with the common EMD, the VMD can overcome the modal aliasing phenomenon and the end-point effect existing in EMD decomposition; the method also utilizes first-order accumulation in gray theory to generate and weaken the influence of random factors in the historical output power data sequence of the photovoltaic power station before VMD decomposition is executed.
3. As a main structure of the prediction method, the invention provides a novel structure which is characterized in that each mode after VMD decomposition of the historical output power data sequence of the photovoltaic power station respectively establishes a support vector machine SVM prediction model and then sums up and reconstructs, so that the learning capacity and generalization capacity of the support vector machine can be fully exerted, and the prediction precision and convergence speed are improved.
4. The parameters of the SVM predictive model are obtained by constructing a complete optimization problem mathematical model, and the solving method adopts a meta heuristic search algorithm, namely a whale group optimization algorithm WSA, which is newly proposed in academia.
5. Finally, the invention also provides a method for predicting the error of the output power of the photovoltaic power station by using the gray model GM (1, 1) and by using the SVM model to predict and sum and reconstruct, and subtracting the error before outputting the final predicted value of the output power of the photovoltaic power station, thereby further improving the prediction precision.
The beneficial technical effects of the invention are as follows:
according to the method provided by the invention, through the VMD-SVM-WSA-GM combined model, the accurate prediction of the short-term output power of the photovoltaic power station is realized, the scientific scheduling decision of the power system after the photovoltaic grid connection is facilitated, and the operation stability of the power system is improved.
Drawings
Fig. 1 is a general block diagram of a photovoltaic power station output power prediction method based on a VMD-SVM-WSA-GM combined model according to an embodiment of the present invention.
Fig. 2 is a flowchart of optimizing parameters of a support vector machine SVM prediction model by using a whale-group optimization algorithm WSA according to a second embodiment of the present invention.
Fig. 3 is a comparison chart of regression prediction data obtained by predicting photovoltaic power by applying the combination model according to the third embodiment of the present invention and original data.
Detailed Description
In order to better understand the aspects of the present invention, the present invention will be described in further detail with reference to the accompanying drawings and detailed description.
Example 1
Fig. 1 is a general block diagram of a photovoltaic power station output power prediction method based on a VMD-SVM-WSA-GM combined model. It should be noted that the method is a time series prediction method, which is essentially different from a method for predicting output power of a photovoltaic system based on influencing factors (such as solar irradiation intensity, ambient temperature, wind speed, etc.). The prediction method based on the influence factors is to take a plurality of influence factor data sequences as multidimensional input quantities, and obtain corresponding photovoltaic power station output power predicted values through a prediction model. Setting a time period window with the length of m, taking m output power values before a predicted time point as m-dimensional input values, and obtaining a current photovoltaic power station output power predicted value through a prediction model; before the next time point is predicted, correspondingly sliding a time unit to the right by a time window, taking m output power values in the time window as input quantities, and iteratively performing the process until the prediction is finished. Preferably, the length of the time period window is 6-10.
The photovoltaic power station output power prediction method based on the VMD-SVM-WSA-GM combined model comprises the following specific steps:
step 1, acquiring historical output power data of a target photovoltaic power station, wherein the historical output power data is arranged into a time sequence according to sampling time by taking 1 hour (1 h) as a unit. According to the actual situation of the output power prediction problem of the target photovoltaic power station, the historical output power data can be less than 1 hour or a plurality of hours as the sampling frequency.
Step 2, preprocessing the historical output power data sequence, wherein preprocessing measures comprise equidistant generation and first-order accumulation generation, and specifically comprise the following steps:
the equidistant data used for complementing the missing data of a certain output power sampling point is obtained by taking the average value of the data in a given step length before and after the sampling point, and preferably, the step length is 3-6.
Generating a first-order accumulation for eliminating the influence of random factors in a historical output power data sequence of the photovoltaic power station, wherein the historical output power data sequence is X (0) The first-order accumulation generation sequence is X (1) The following are provided:
X (0) ＝(x (0) (1),x (0) (2),…,x (0) (n))
X (1) ＝(x (1) (1),x (1) (2),…,x (1) (n))
wherein x is (0) (j) For the element in the historical output power data sequence of the photovoltaic power station, j is the sequence number of the element in the historical output power data sequence, j is [1, n ]]；x (1) (k) For the first-order accumulation generating sequence, k is the sequence number of the element in the first-order accumulation generating sequence, and k is E [1, n]。
And 3, decomposing the first-order accumulation generation sequence of the historical output power data into subsequences with limited bandwidth by adopting a variation mode decomposition method VMD, wherein each subsequence corresponds to different modes, namely a mode 1, a mode 2, a mode … and a mode q.
Preferably, the number q of decomposition layers of the VMD needs to be set externally, and the invention adopts a frequency spectrum analysis method to determine the number q of decomposition layers of the VMD in a mode of generating the number of main frequencies in a sequence frequency spectrum chart by first-order accumulation of the historical output power data.
And 4, respectively establishing a Support Vector Machine (SVM) prediction model for each subsequence decomposed by the VMD in the step 3.
Preferably, the method comprises the steps of,kernel function K (x) of support vector machine i X) type is determined as radial basis function RBF as follows:
wherein x is i And x represents any point in space and a certain center, respectively, and sigma is the width parameter of the radial basis function.
And 5, optimizing parameters of each support vector machine SVM prediction model by using a whale swarm optimization algorithm WSA, wherein the SVM parameters to be optimized are a penalty parameter C, a sensitivity loss parameter epsilon and a parameter sigma of a radial basis function.
Step 6, carrying out summation reconstruction on the prediction results of each SVM model to obtain a first-order accumulation result prediction value sequence of the output power of the photovoltaic power station, and marking the first-order accumulation result prediction value sequence as
The following are provided:
step 7, calculating a first-order accumulation generation sequence X of the historical output power data (1) And photovoltaic power station output power first-order accumulation result predicted value sequence
The difference between the two is noted as error sequence E as follows:
E＝(e(1),e(2),…,e(n))
step 8, predicting the error sequence E by using a GM (1, N) model to obtain a predicted value sequence of the error, which is recorded as
The following are provided:
step 9, calculating the first-order accumulation result of the predicted output power
And error prediction value sequence +.>
And carrying out subtraction reduction on the difference value of the two to obtain a predicted value sequence Y of the output power of the target photovoltaic power station, wherein the predicted value sequence Y is as follows:
Y＝(y(1),y(2),…,y(n))
wherein y (1) is the 1 st element in the predicted value sequence of the output power of the target photovoltaic power station,
and->
The 1 st element in the predicted value sequence of the first-order accumulated result and the predicted value sequence of the error of the output power of the photovoltaic power station are respectively; y (k) is the kth element in the predicted value sequence of the output power of the target photovoltaic power station, k is the serial number, ">
And->
K and k-1 elements in a predicted value sequence of a first-order accumulated result of output power of the photovoltaic power station are respectively +.>
And->
K and k-1 elements in the predicted value sequence of the error, respectively.
And step 10, calculating a relative error, and checking the prediction accuracy of the photovoltaic power station output power prediction method based on the VMD-SVM-WSA-GM combined model.
It should be noted that, the invention divides the obtained photovoltaic power station output power history data into a training set and a testing set, wherein: the training set is used for establishing the VMD-SVM-WSA-GM based combined model, the testing set is used for testing and checking the prediction precision of the combined model, and the combined model can be used for carrying out prediction work after the training process and the testing process are completed. The foregoing steps 1 to 10 are the complete flow of the training set and the training process, and for the test process and the application process, only the m modal components corresponding to the output power at different moments are input according to the established model set including q support vector machine SVM prediction models, so as to obtain q prediction value subsequences of different modalities, then the q prediction value subsequences are summed and reconstructed, and then the error prediction value obtained through the gray model GM (1, 1) is added, so as to finally obtain the prediction value sequence of the output power of the photovoltaic power station.
Example two
As shown in fig. 2, a flowchart of optimizing parameters of a support vector machine SVM prediction model by using whale swarm optimization algorithm WSA in the present invention is specifically described as follows:
step 501, establishing a mathematical expression of a support vector machine SVM prediction model parameter optimization problem, as follows:
s.t.C min ≤C p ≤C max
ε min ≤ε p ≤ε max
σ min ≤σ p ≤σ max
1≤p≤q
wherein C is p 、ε p Sum sigma p Penalty parameter, sensitivity loss parameter and parameter of radial basis function of p-th SVM predictive model respectively, C min And C max 、ε min And epsilon max 、σ min Sum sigma max The penalty parameter C, the sensitivity loss parameter epsilon and the width parameter sigma of the radial basis function are respectively lower/upper valued limits.
The optimization objective of the parameter optimization problem is to minimize the sum of error sequences E, i.e. the first-order accumulation of historical output power data generates a sequence X (1) And photovoltaic power station output power first-order accumulation result predicted value sequence
The sum of the difference values of the corresponding components is the smallest; the constraint condition of the parameter optimization problem is the value range constraint which is satisfied by the parameters of the SVM prediction model; the optimization variables of the parameter optimization problem are row vectors formed by parameter arrangement of q SVM prediction models, and the row vectors are as follows:
C 1 ε 1 σ 1 C 2 ε 2 σ 2 …C p ε p σ p …C q ε q σ q
step 502, initializing whale shoal omega in whale shoal algorithm, taking the population size pop as 100, and taking the maximum evolutionary algebra maxgen as 10000.
Step 503, initializing whale positions in a whale population.
Step 504, each whale individual is first evaluated and its fitness value is calculated.
Step 505, omega for the current individual in whale population i Searching for "preferred and most recent" whale W, if W is present, Ω i Move to W and calculate omega after the move i Is used for the adaptation value of the (c).
Step 506, determining whether all Ω -fish in the current whale are traversed i If yes, continuing the next step; no, go to step 505.
Step 507, judging whether a preset maximum evolution algebra maxgen is reached, if yes, continuing the next step; if not, the evolution algebra self-increases by 1 and jumps to step 505.
And step 508, after the optimization calculation is finished, outputting the optimization parameters of each SVM prediction model.
Example III
Fig. 3 is a comparison chart of regression prediction data obtained by predicting photovoltaic power by applying the combination model according to the third embodiment of the present invention and original data.
In order to verify the effectiveness of the model and the method provided by the invention, the model and the method are applied to the problem of predicting the output power of a certain practical photovoltaic power station. 480 groups of output power data of the photovoltaic power station are used as a data set. Each group of data corresponds to the output power sampling value of the photovoltaic power station in every 30 minutes (min), and the output power of the photovoltaic power station at night is ignored, so that the data corresponding to the night time period in 24 hours each day need to be removed, and only the output power of 12 hours each day is taken, so that the data of each day are 24 groups. The 480 sets of data are further divided into training and testing sets, containing 360 and 120 sets of data, respectively. The specific application process is as follows: the models presented herein are first trained with a training set, then tested and validated on a test set.
As can be seen from fig. 3, in 5 days (120 sets of data) corresponding to the test set, the regression prediction data obtained according to the model and method of the present invention maintains high consistency compared with the original data, and the regression prediction data well reproduces the fluctuation rule of the original data with little deviation. The offset values for the first 20 sets of data are truncated as shown in table 1 below.
TABLE 1 deviation of regression prediction data from raw data (including absolute and relative errors)
As can be seen from the above table, the prediction accuracy of the model and method of the present invention is very high, and the relative error of all the data except the first three groups among the above 20 groups of data is less than 3.67%, and the relative error of 10 groups of data is less than 1%. This demonstrates the effectiveness and superiority of the proposed model and method.
In summary, the method disclosed by the invention aims at the characteristics of strong randomness and volatility of the output power of the photovoltaic system, provides a prediction method based on a VMD-SVM-WSA-GM combined model, realizes accurate prediction of the short-term output power of the photovoltaic power station, is beneficial to scientific scheduling decision of the power system after photovoltaic grid connection, and improves the operation stability of the power system.
It will be further appreciated by those of ordinary skill in the art that implementing all or part of the steps in the methods of the above embodiments may be accomplished by program instructions in hardware associated with the program instructions, where the program instructions may be stored on a computer readable storage medium, where the storage medium includes ROM/RAM, magnetic disk, optical disk, etc.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.
Claims (2)
1. The photovoltaic power station output power prediction method based on the combined model is characterized in that the combined model is a VMD-SVM-WSA-GM combined model, and the power generation value of the photovoltaic power station in the future period is predicted according to historical output power data of the photovoltaic power station, and the method comprises the following steps:
step 1, acquiring historical output power data of a target photovoltaic power station, wherein the historical output power data are arranged into a time sequence according to sampling time by taking set time as a unit;
step 2, preprocessing the historical output power data sequence, wherein preprocessing measures comprise equidistant generation and first-order accumulation generation, and specifically comprise the following steps:
the method comprises the steps of equally spacing data for supplementing missing data of a certain output power sampling point, taking an average value of the data in a given step length before and after the sampling point, wherein the step length is taken as 3-6;
generating a first-order accumulation for eliminating the influence of random factors in a historical output power data sequence of the photovoltaic power station, wherein the historical output power data sequence is X (0) The first-order accumulation generation sequence is X (1) The following are provided:
X (0) ＝(x (0) (1),x (0) (2),…,x (0) (n))
X (1) ＝(x (1) (1),x (1) (2),…,x (1) (n))
wherein x is (0) (j) For the element in the historical output power data sequence of the photovoltaic power station, j is the sequence number of the element in the historical output power data sequence, j is [1, n ]]；x (1) (k) For the first-order accumulation generating sequence, k is the sequence number of the element in the first-order accumulation generating sequence, and k is E [1, n]；
Step 3, decomposing the first-order accumulation generation sequence of the historical output power data into subsequences with limited bandwidth by adopting a variation mode decomposition method VMD, wherein each subsequence corresponds to different modes, namely a mode 1, a mode 2, a mode … and a mode q;
the number q of the decomposition layers of the VMD is required to be set externally, and the number q of the decomposition layers of the VMD is determined by adopting a spectrum analysis method in a mode of generating the number of main frequencies in a sequence spectrogram through first-order accumulation of the historical output power data;
step 4, respectively establishing a support vector machine SVM prediction model for each subsequence decomposed by the VMD in step 3, and supporting a kernel function K (x i X) type is determined as radial basis function RBF as follows:
wherein x is i And x represents any point and a certain center in space, and sigma is the width parameter of a radial basis function;
step 5, optimizing parameters of each support vector machine SVM prediction model by using whale swarm optimization algorithm WSA, wherein SVM parameters to be optimized are penalty parameters C, sensitivity loss parameters epsilon and width parameters sigma of radial basis functions;
step 6, carrying out summation reconstruction on the prediction results of each SVM model to obtain a first-order accumulation result prediction value sequence of the output power of the photovoltaic power station, and marking the first-order accumulation result prediction value sequence as
The following are provided:
step 7, calculating a first-order accumulation generation sequence X of the historical output power data (1) And photovoltaic power station output power first-order accumulation result predicted value sequence
The difference between the two is noted as error sequence E as follows:
E＝(e(1),e(2),…,e(n))
step 8, predicting the error sequence E by using a GM (1, N) gray model to obtain a predicted value sequence of the error, which is recorded as
The following are provided:
step 9, calculating the first-order accumulation result of the predicted output power
And error prediction value sequence +.>
And carrying out subtraction reduction on the difference value of the two to obtain a predicted value sequence Y of the output power of the target photovoltaic power station, wherein the predicted value sequence Y is as follows:
Y＝(y(1),y(2),…,y(n))
wherein y (1) is the 1 st element in the predicted value sequence of the output power of the target photovoltaic power station,
and->
The 1 st element in the predicted value sequence of the first-order accumulated result and the predicted value sequence of the error of the output power of the photovoltaic power station are respectively; y (k) is the kth element in the predicted value sequence of the output power of the target photovoltaic power station, k is the serial number, ">
And->
K and k-1 elements in a predicted value sequence of a first-order accumulated result of output power of the photovoltaic power station are respectively +.>
And->
K and k-1 elements in the predicted value sequence of the error respectively;
and step 10, calculating a relative error, and checking the prediction accuracy of the photovoltaic power station output power prediction method based on the VMD-SVM-WSA-GM combined model.
2. The method for predicting the output power of the photovoltaic power station based on the combined model according to claim 1, wherein the detailed process of optimizing the parameters of the support vector machine SVM prediction model by applying whale swarm optimization algorithm WSA in the step 5 is as follows:
step 501, establishing a mathematical expression of a support vector machine SVM prediction model parameter optimization problem, as follows:
s.t.C min ≤C p ≤C max
ε min ≤ε p ≤ε max
σ min ≤σ p ≤σ max
1≤p≤q
wherein C is p 、ε p Sum sigma p Penalty parameter, sensitivity loss parameter and parameter of radial basis function of p-th SVM predictive model respectively, C min And C max 、ε min And epsilon max 、σ min Sum sigma max The lower limit/upper limit of the values of the punishment parameter C, the sensitivity loss parameter epsilon and the width parameter sigma of the radial basis function are respectively adopted;
the optimization objective of the parameter optimization problem is to minimize the sum of error sequences E, i.e. the first-order accumulation of historical output power data generates a sequence X (1) And photovoltaic power station output power first-order accumulation result predicted value sequence
The sum of the difference values of the corresponding components is the smallest; the constraint condition of the parameter optimization problem is the value range constraint which is satisfied by the parameters of the SVM prediction model; the optimization variables of the parameter optimization problem are row vectors formed by parameter arrangement of q SVM prediction models, and the row vectors are as follows:
C 1 ε 1 σ 1 C 2 ε 2 σ 2 …C p ε p σ p …C q ε q σ q
step 502, initializing whale shoal omega in a whale shoal algorithm, taking the population scale pop as 100, and taking the maximum evolutionary algebra maxgen as 10000;
step 503, initializing whale positions in whale groups;
step 504, evaluating each whale individual for the first time, and calculating the fitness value of each whale individual;
step 505, omega for the current individual in whale population i Searching for "preferred and most recent" whale W, if W is present, Ω i Move to W and calculate omega after the move i Is a fitness value of (a);
step 506, determining whether all Ω -fish in the current whale are traversed i If yes, continuing the next step; if not, jumping to step 505;
step 507, judging whether a preset maximum evolution algebra maxgen is reached, if yes, continuing the next step; if not, the evolution algebra self-increases by 1, and jumps to step 505;
and step 508, after the optimization calculation is finished, outputting the optimization 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 CN110059891A (en) | 2019-07-26 |
| CN110059891B true 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) |
Families Citing this family (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 |
| 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 |
| CN112613650B (en) * | 2020-12-14 | 2022-06-28 | 燕山大学 | Novel three-stage short-term power load prediction method |
| CN113052330B (en) * | 2021-03-18 | 2022-08-02 | 淮北师范大学 | Ox penis effect weakening method based on VMD-SVM algorithm |
| 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 |
| CN113688869B (en) * | 2021-07-21 | 2022-05-27 | 广东工业大学 | Photovoltaic data missing reconstruction method based on generation countermeasure network |
| 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 |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106485368A (en) * | 2016-11-04 | 2017-03-08 | 江南大学 | A kind of short term prediction method for energy resource system power output |
| 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
Also Published As
| Publication number | Publication date |
|---|---|
| CN110059891A (en) | 2019-07-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110059891B (en) | Photovoltaic power station output power prediction method based on VMD-SVM-WSA-GM combined model | |
| Zhang et al. | Short-term wind speed prediction model based on GA-ANN improved by VMD | |
| CN111008728B (en) | Prediction method for short-term output of distributed photovoltaic power generation system | |
| CN106529814B (en) | Distributed photovoltaic ultra-short term prediction method based on Adaboost clustering and Markov chain | |
| CN110309603B (en) | Short-term wind speed prediction method and system based on wind speed characteristics | |
| CN110110912B (en) | Photovoltaic power multi-model interval prediction method | |
| CN110929953A (en) | Photovoltaic power station ultra-short term output prediction method based on cluster analysis | |
| CN108053048A (en) | A kind of gradual photovoltaic plant ultra-short term power forecasting method of single step and system | |
| CN110601250B (en) | Method for evaluating stability of static voltage of access point based on new energy power station | |
| Zhang et al. | Interval prediction of ultra-short-term photovoltaic power based on a hybrid model | |
| CN111242353A (en) | Wind power combined prediction modeling and prediction method | |
| CN115275991A (en) | Active power distribution network operation situation prediction method based on IEMD-TA-LSTM model | |
| CN115759415A (en) | Power consumption demand prediction method based on LSTM-SVR | |
| CN117239722A (en) | System wind load short-term prediction method considering multi-element load influence | |
| He et al. | Forecasting power grid states for regional energy markets with deep neural networks | |
| CN114154684A (en) | Short-term photovoltaic power prediction method based on data mining and multi-core support vector machine | |
| CN114298132A (en) | Wind power prediction method and device and electronic equipment | |
| Sun et al. | Short-term power load prediction based on VMD-SG-LSTM | |
| CN116341728A (en) | Ultra-short-term photovoltaic output power prediction method based on data driving | |
| CN115796327A (en) | Wind power interval prediction method based on VMD (vertical vector decomposition) and IWOA-F-GRU (empirical mode decomposition) -based models | |
| CN116128211A (en) | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene | |
| Shen et al. | An interval analysis scheme based on empirical error and mcmc to quantify uncertainty of wind speed | |
| Tian et al. | Fusion modeling for wind power forecasting based on redundant method elimination | |
| Cheng et al. | Temperature & humidity-weighted-modified GRNN based prediction model of photovoltaic power generation | |
| Liu et al. | Intelligent voltage prediction of active distribution network with high proportion of distributed photovoltaics |
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 |