CN108876013A - One kind being based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique - Google Patents
One kind being based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique Download PDFInfo
- Publication number
- CN108876013A CN108876013A CN201810526061.XA CN201810526061A CN108876013A CN 108876013 A CN108876013 A CN 108876013A CN 201810526061 A CN201810526061 A CN 201810526061A CN 108876013 A CN108876013 A CN 108876013A
- Authority
- CN
- China
- Prior art keywords
- day
- parameter sample
- kth
- sample
- combination
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000004927 fusion Effects 0.000 title claims abstract description 13
- 230000001537 neural effect Effects 0.000 title claims abstract description 13
- 238000013528 artificial neural network Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims abstract description 16
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 8
- 238000010219 correlation analysis Methods 0.000 claims abstract description 5
- 238000005286 illumination Methods 0.000 claims abstract description 4
- 241000196324 Embryophyta Species 0.000 claims description 23
- 230000005855 radiation Effects 0.000 claims description 16
- 238000009792 diffusion process Methods 0.000 claims description 9
- 238000010606 normalization Methods 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 5
- 241000208340 Araliaceae Species 0.000 claims description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 claims description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 claims description 3
- 235000008434 ginseng Nutrition 0.000 claims description 3
- 210000005036 nerve Anatomy 0.000 claims 1
- 230000002354 daily effect Effects 0.000 description 10
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 8
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 8
- 238000010248 power generation Methods 0.000 description 7
- 238000011160 research Methods 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000006835 compression Effects 0.000 description 2
- 238000007906 compression Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000000611 regression analysis Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003749 cleanliness Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- 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
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to one kind to be based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique, it is input with the meteorologic parameters such as the generated output at daily each moment, daily illumination, environment temperature, humidity and wind speed and second day meteorologic parameter, the generated output at second day each moment is to export prediction model of the training based on Elman neural network.The prediction of photovoltaic generation power is carried out with this model.Then the best similar day of day to be predicted is determined by grey correlation analysis algorithm using daily meteorologic parameter as Meteorological Characteristics value, by the generated output at each moment of best similar day, the input variable of meteorologic parameter and the meteorologic parameter of day to be predicted as model predicts the generated output at each moment of day to be predicted.The present invention can quick and precisely predict the generated output of photovoltaic plant.
Description
Technical field
The present invention relates to photovoltaic plant short term power electric powder prediction, it is especially a kind of based on best similar day and
Elman neural fusion photovoltaic plant short term power prediction technique.
Background technique
In recent years, with the rapid development of social economy, Devoting Major Efforts To Developing green novel energy source, which has become, solves the energy and ring
A kind of strong means of border problem.The photovoltaic power generation unique charm exhausted with its cleanliness without any pollution and never has rapidly become generation
The hot spot of boundary concern and research, becomes the best substitute of traditional fossil energy.However the unstability of environment makes photovoltaic
The output of power generation have very strong fluctuation and intermittence, reduce it is grid-connected after stability, substantially increase photovoltaic hair
The difficulty of TV university sizable application.Therefore, the generated output of photovoltaic plant is predicted not only mitigate photovoltaic generating system
Grid-connected adverse effect, and photovoltaic power generation is combined with dispatching of power netwoks, electric load distribution etc., it can also plan
With the entire electric system of operation, the stability and utilization rate of system are substantially increased, is had to photovoltaic power generation factory and electric system
Important economic significance, have very high realistic meaning and learning value.
Currently, the power forecasting method of photovoltaic plant mainly has forecasting by regression analysis, Grey Theory Forecast method and artificial
Intelligent predicting method etc..Forecasting by regression analysis determines power and other parameters by the statistical analysis to power and other parameters
Between correlativity be usually used in establish regression equation, then the prediction that regression equation is realized as prediction model
Short-term forecast.Grey Theory Forecast method carries out analysis and pre- by generating ordered series of numbers modeling come digging system internal information
It surveys.Since photovoltaic power generation has great randomness and uncertainty, and influence factor is numerous, belongs to gray system, therefore ash
Color prediction theory can be applied to photovoltaic power forecasting research very well.The advantages of prediction technique is that method is simple, is easy to calculate,
And short-term forecast precision is high, is easily verified.
But in recent years, as artificial intelligence is widely applied in photovoltaic power generation power prediction, linear regression
The use of method and Grey Theory Forecast method greatly reduces, therewith the artificial intelligence prediction based on neural network, support vector machine
Algorithm is the Predicting Technique being most widely used in recent years.Wherein the prediction algorithm based on support vector machines can be solved preferably
Certainly Small Sample Size, precision is higher, but when carrying out parameter optimization to it using optimization algorithm, training time for needing to grow very much.
And although neural network easily falls into local minimum, due to it is higher fitting with generalization ability and the training time it is shorter, compare
For, it is better than support vector machines in terms of estimated performance, also has been achieved for more successfully applying at present.
Wherein, Elman neural network is compared with traditional BP neural network, and more one receive feedback letter from hidden layer
Number, for remember hidden layer neuron previous moment output valve undertaking layer so that network to historical data have sensibility,
Increase the ability of network itself processing multidate information.Therefore Elman neural network more better than BP neural network is selected herein
As prediction model.In order to more accurately predict the generated output of day to be predicted, it is closest to need to find meteorological condition therewith
Input of the history day as model.
Currently, there is not yet the Elman neural network based on best similar day is calculated in the document and patent published
Method is applied to the research of photovoltaic plant short term power prediction.
Summary of the invention
In view of this, the purpose of the present invention is to propose to one kind to be based on best similar day and Elman neural fusion photovoltaic
Power station short term power prediction technique can quick and precisely predict the generated output of photovoltaic plant.
The present invention is realized using following scheme:One kind being based on best similar day and Elman neural fusion photovoltaic plant
Short term power prediction technique, specifically includes following steps:
Step S1:The meteorologic parameter on photovoltaic plant generated output over the years and weather station is acquired, daily parameter is obtained
Sample combination;
Step S2:Abnormal data is removed, the combination of parameter sample is normalized;
Step S3:Determine input layer number m, the node in hidden layer n and output layer number of nodes p of Elman neural network
And the parameters of initialization Elman neural network;
Step S4:The Elman obtained using the parameter sample combination training step S3 on the day before first day to day to be predicted
Neural network is constantly modified the hidden layer number of the Elman neural network by trial and error procedure, obtains training pattern;
Step S5:The best of day to be predicted is determined according to the Meteorological Characteristics value combination grey correlation analysis algorithm of day to be predicted
Similar day;
Step S6:By the parameter sample combination of best similar day and the instruction of the meteorologic parameter input step S4 of day to be predicted
Practice and the generated output of day to be predicted is predicted in model, obtains day to be predicted per output power value every other hour.
Further, in step S1, the parameter sample combination includes photovoltaic plant generated output over the years and weather station
On illumination, environment temperature, humidity, wind speed information;
The parameter sample combination is denoted as (Pki, Gki, Dki, Tki, Wki, Hki), wherein k is the sequence on the date of sample collection
Number, indicate that number of days, k are 1 integer for arriving N;At the time of i is sample collection in one day, indicate that moment number, i are 1 integer for arriving Nt;
PkiFor the power parameter sample at i-th of moment in the combination of kth day parameter sample, GkiFor the in the combination of kth day parameter sample
The global horizontal radiation parameter sample at i moment, DkiFor the diffusion levels spoke at i-th of moment in the combination of kth day parameter sample
Penetrate parameter sample, TkiFor the environment temperature parameter sample at i-th of moment in the combination of kth day parameter sample, WkiFor kth day ginseng
The wind speed parameter sample at i-th of moment in numerical example combination, HkiFor the phase at i-th of moment in the combination of kth day parameter sample
To humidity parameter sample.
Further, in step S2, the combination of parameter sample is normalized specially:It will using scale compression method
The multi-group data at the same moment of same parameter sample is mapped in section [0,1], and mapping equation is:
In formula, y' indicates the data obtained after normalization, AimaxIndicate the maximum value in i-th of moment of data group A,
AiminIndicate the minimum value in i-th of moment of data group A, AkiIndicate certain of i-th of moment in the combination of kth day parameter sample
A parameter sample.
Wherein, with power sample P=(P1i,P2i... Pki... PNi) for, specific mapping equation is:
In formula, PimaxIndicate the maximum value in i-th of moment of data group P, PiminIn i-th of moment for indicating data group P
Minimum value.
Further, in step S3, the Elman neural network is specifically configured to:Input layer number m is 23, hidden
The n of number containing node layer is 10, and output layer number of nodes p is 11, and the number of iterations 2000, other parameters use default setting;Wherein, hidden
The formula of the number containing layer is as follows:
In formula, a is system parameter, and the value range of a is 1-10.
Further, in step S4, training step S3 is combined using the parameter sample on the day before first day to day to be predicted
Obtained Elman neural network is specially:According to date day to be predicted of selection, by first day in history day to day to be predicted
Parameter sample combination in one day and the meteorologic parameter one day after of this day are as input, the function at the correspondence moment one day after
Rate value obtains an input and output combination, to training Elman neural network as output;
The input and output combination is denoted as:
In formula, k is the serial number on the date of sample collection, indicates that number of days, k are 1 integer for arriving N;Preceding 13 variables indicate defeated
Enter the input variable in output combination, the last one variable indicates output variable;The wherein P in input variablekFor each of kth day
The power parameter sample at a moment,WithFor the average daily global horizontal radiation parameter sample of kth day and kth+1 day,WithFor kth day and kth+1 day average daily diffusion levels radiation parameter sample, TkmaxAnd T(k+1)maxFor kth day and kth+1 day
Maximum environmental temperature parameter sample, TkminAnd T(k+1)minFor kth day and kth+1 day minimum environment temperature parameter sample,WithFor kth day and kth+1 day average daily wind speed parameter sample,WithFor the average daily relatively wet of kth day and kth+1 day
Spend parameter sample, output variable Pk+1For the power parameter sample at kth+1 day each moment.
Further, step S5 specifically includes following steps:
Step S51:The Meteorological Characteristics value is denoted as:
(Gkmax, Gkmin, Dkmax,Dkmin,Tkmax,Tkmin,Wkmax,Wkmin,Hkmax,Hkmin),
In formula, k is the serial number on the date of sample collection, indicates that number of days, k are 1 integer for arriving N;GkmaxAnd GkminTo be maximum and
Minimum whole world horizontal radiation parameter sample, DkmaxAnd DkminFor minimum and maximum diffusion levels radiation parameter sample, TkmaxAnd Tkmin
For minimum and maximum environment temperature parameter sample, WkmaxAnd WkminFor minimum and maximum wind speed parameter sample, HkmaxAnd HkminFor most
Big and minimum relative humidity pa sample;
Step S52:The degree of association for calculating preset date and its a few days ago to be predicted, the degree of association maximum that day is determined as
Best similar day;Wherein, the preset date is first 10 days.Wherein the calculating of the degree of association uses following formula:
In formula, riIndicate the degree of association of history day i and day to be predicted, k indicates characteristic value number, ξiIndicate history day i
With the incidence coefficient of day to be measured;Wherein, ξi(k) calculating uses following formula:
In formula, y (k) indicates the Meteorological Characteristics value after day normalization to be predicted, xi(k) after indicating history day i normalization
Meteorological Characteristics value, ρ indicate that resolution ratio, ρ take 0.5, k to indicate characteristic value number.
Preferably, needing the combination of parameter sample and the meteorologic parameter of day to be predicted by best similar day in step S6
As input, trained model predicting per generated output every other hour to day to be predicted is utilized.The input
Sample setting is identical as step S4 setting.
Compared with prior art, the invention has the following beneficial effects:
1, the present invention is capable of the generated output at following one day each moment of look-ahead photovoltaic plant.
2, the present invention determines best similar day with grey correlation analysis algorithm, carries out in conjunction with Elman neural network model pre-
It surveys, effectively improves the accuracy of photovoltaic power station power generation power prediction.By taking on March 30th, 2018 as an example, root mean square of the invention is missed
Difference is 3.40KW, mean absolute percentage error 3.23%, the coefficient of determination 0.995.
Detailed description of the invention
Fig. 1 is the method flow schematic diagram of the embodiment of the present invention.
Fig. 2 is the power prediction result schematic diagram of the embodiment of the present invention.
Fig. 3 is to predict error curve at each moment of the embodiment of the present invention.
Specific embodiment
The present invention will be further described with reference to the accompanying drawings and embodiments.
It is noted that following detailed description is all illustrative, it is intended to provide further instruction to the application.Unless another
It indicates, all technical and scientific terms used herein has usual with the application person of an ordinary skill in the technical field
The identical meanings of understanding.
It should be noted that term used herein above is merely to describe specific embodiment, and be not intended to restricted root
According to the illustrative embodiments of the application.As used herein, unless the context clearly indicates otherwise, otherwise singular
Also it is intended to include plural form, additionally, it should be understood that, when in the present specification using term "comprising" and/or " packet
Include " when, indicate existing characteristics, step, operation, device, component and/or their combination.
As shown in Figure 1, present embodiments providing a kind of based on best similar day and Elman neural fusion photovoltaic plant
Short term power prediction technique, specifically includes following steps:
Step S1:The meteorologic parameter on photovoltaic plant generated output over the years and weather station is acquired, daily parameter is obtained
Sample combination;
Step S2:Abnormal data is removed, the combination of parameter sample is normalized;
Step S3:Determine input layer number m, the node in hidden layer n and output layer number of nodes p of Elman neural network
And the parameters of initialization Elman neural network;
Step S4:The Elman obtained using the parameter sample combination training step S3 on the day before first day to day to be predicted
Neural network is constantly modified the hidden layer number of the Elman neural network by trial and error procedure, obtains training pattern;
Step S5:The best of day to be predicted is determined according to the Meteorological Characteristics value combination grey correlation analysis algorithm of day to be predicted
Similar day;
Step S6:By the parameter sample combination of best similar day and the instruction of the meteorologic parameter input step S4 of day to be predicted
Practice and the generated output of day to be predicted is predicted in model, obtains day to be predicted per output power value every other hour.
Preferably, acquiring the alice springs light that photovoltaic plant used by data is Australia in the present embodiment
Overhead utility, power station are made of the photovoltaic panel that 22 rated values are 250W, and array rated value is 5.5KW, is carried out by inverter
It generates electricity by way of merging two or more grid systems.
In the present embodiment, in step S1, the parameter sample combination includes photovoltaic plant generated output over the years and gas
As the illumination on station, environment temperature, humidity, wind speed information;
The parameter sample combination is denoted as (Pki, Gki, Dki, Tki, Wki, Hki), wherein k is the sequence on the date of sample collection
Number, indicate that number of days, k are 1 integer for arriving N;At the time of i is sample collection in one day, indicate that moment number, i are 1 integer for arriving Nt;
PkiFor the power parameter sample at i-th of moment in the combination of kth day parameter sample, GkiFor the in the combination of kth day parameter sample
The global horizontal radiation parameter sample at i moment, DkiFor the diffusion levels spoke at i-th of moment in the combination of kth day parameter sample
Penetrate parameter sample, TkiFor the environment temperature parameter sample at i-th of moment in the combination of kth day parameter sample, WkiFor kth day ginseng
The wind speed parameter sample at i-th of moment in numerical example combination, HkiFor the phase at i-th of moment in the combination of kth day parameter sample
To humidity parameter sample.
In the present embodiment, in step S2, the combination of parameter sample is normalized specially:Using scale compression
The multi-group data at the same moment of same parameter sample is mapped in section [0,1] by method, and mapping equation is:
In formula, y' indicates the data obtained after normalization, AimaxIndicate the maximum value in i-th of moment of data group A,
AiminIndicate the minimum value in i-th of moment of data group A, AkiIndicate certain of i-th of moment in the combination of kth day parameter sample
A parameter sample.
Wherein, with power sample P=(P1i,P2i... Pki... PNi) for, specific mapping equation is:
In formula, PimaxIndicate the maximum value in i-th of moment of data group P, PiminIn i-th of moment for indicating data group P
Minimum value.
In the present embodiment, in step S3, the Elman neural network is specifically configured to:Input layer number m is
23, node in hidden layer n are 10, and output layer number of nodes p is 11, and the number of iterations 2000, other parameters use default setting;Its
In, the formula of hidden layer number is as follows:
In formula, a is system parameter, and the value range of a is 1-10.
In the present embodiment, it in step S4, is walked using the parameter sample combined training on the day before first day to day to be predicted
The Elman neural network that rapid S3 is obtained is specially:According to date day to be predicted of selection, first day in history day is arrived to be predicted
One day parameter sample combination of day and the meteorologic parameter one day after of this day are as input, the correspondence moment one day after
Performance number as output, input and output combination is obtained, to training Elman neural network;
The input and output combination is denoted as:
In formula, k is the serial number on the date of sample collection, indicates that number of days, k are 1 integer for arriving N;Preceding 13 variables indicate defeated
Enter the input variable in output combination, the last one variable indicates output variable;The wherein P in input variablekFor each of kth day
The power parameter sample at a moment,WithFor the average daily global horizontal radiation parameter sample of kth day and kth+1 day,WithFor kth day and kth+1 day average daily diffusion levels radiation parameter sample, TkmaxAnd T(k+1)maxFor kth day and kth+1 day
Maximum environmental temperature parameter sample, TkminAnd T(k+1)minFor kth day and kth+1 day minimum environment temperature parameter sample,WithFor kth day and kth+1 day average daily wind speed parameter sample,WithFor the average daily relatively wet of kth day and kth+1 day
Spend parameter sample, output variable Pk+1For the power parameter sample at kth+1 day each moment.
In the present embodiment, step S5 specifically includes following steps:
Step S51:The Meteorological Characteristics value is denoted as:
(Gkmax, Gkmin, Dkmax,Dkmin,Tkmax,Tkmin,Wkmax,Wkmin,Hkmax,Hkmin),
In formula, k is the serial number on the date of sample collection, indicates that number of days, k are 1 integer for arriving N;GkmaxAnd GkminTo be maximum and
Minimum whole world horizontal radiation parameter sample, DkmaxAnd DkminFor minimum and maximum diffusion levels radiation parameter sample, TkmaxAnd Tkmin
For minimum and maximum environment temperature parameter sample, WkmaxAnd WkminFor minimum and maximum wind speed parameter sample, HkmaxAnd HkminFor most
Big and minimum relative humidity pa sample;
Step S52:The degree of association for calculating preset date and its a few days ago to be predicted, the degree of association maximum that day is determined as
Best similar day;Wherein, the preset date is first 10 days.Wherein the calculating of the degree of association uses following formula:
In formula, riIndicate the degree of association of history day i and day to be predicted, k indicates characteristic value number, ξiIndicate history day i
With the incidence coefficient of day to be measured;Wherein, ξi(k) calculating uses following formula:
In formula, y (k) indicates the Meteorological Characteristics value after day normalization to be predicted, xi(k) after indicating history day i normalization
Meteorological Characteristics value, ρ indicate that resolution ratio, ρ take 0.5, k to indicate characteristic value number.
Particularly, in the present embodiment, for predicting on March 30th, 2018, in March, 2018 is obtained according to weather forecast
Meteorologic parameter on the 30th is as shown in table 1.The 10 days a few days ago to be predicted degrees of association with it are calculated according to table 1, the degree of association is maximum
That day is determined as best similar day.The calculation method of the degree of association is calculated using above formula.Preceding calculation of relationship degree result on the 10th such as 2 institute of table
Show.As shown in Table 2, in the present embodiment, the best similar day in March 30 2018 day to be predicted is on March 23rd, 2018, most
Good association angle value is 0.85.
The meteorologic parameter on March 30 2018 day to be predicted of table 1
The degree of association on the 10th before table 2 March 30 2018 day to be predicted
Date | 3-29 | 3-28 | 3-27 | 3-26 | 3-25 | 3-24 | 3-23 | 3-22 | 3-21 | 3-20 |
The degree of association | 0.82 | 0.83 | 0.72 | 0.79 | 0.74 | 0.75 | 0.85 | 0.76 | 0.73 | 0.46 |
Preferably, in the present embodiment, in step S6, needing to combine the parameter sample of best similar day and to be predicted
The meteorologic parameter of day utilizes trained model carrying out in advance per generated output every other hour to day to be predicted as input
It surveys.The input sample setting is identical as step S4 setting.
Preferably, the prediction result of the present embodiment and prediction of each moment error curve are as shown in Figures 2 and 3, model is missed
Poor index value is as shown in table 3.The root-mean-square error RMSE in March 30 2018 day to be predicted is 3.40KW, average absolute percentage
Ratio error MAPE is 3.23%, R2It is 0.995, MAPE error within 5%, R2Reach 0.99 or more.
3 model error index of table
RMSE(KW) | MAPE (%) | R2 |
3.40 | 3.23 | 0.995 |
It is corresponding, using day to be predicted proxima luce (prox. luc), that is, adjacent day data sample as model input as control
Group, using best similar day data sample as model input regard experimental group, select different months totally 10 days as to pre-
Survey is tested day, and each operation 10 times, experimental result is as shown in table 4.From the point of view of average value, the MAPE average value of experimental group is
9.65%, the MAPE average value 12.32% than control group improves about 3 percentage points, and the RMSE average value of experimental group is 9.14KW,
RMSE average value 12.44KW than control group reduces 3.3KW, the R of experimental group2Average value is 0.98, than the R of control group2Average value
0.97 improves 0.01.From the point of view of mean square deviation, experimental group MAPE mean square deviation is 5.41%, smaller than control group MAPE mean square deviation 5.63%
0.22 percentage point, experimental group RMSE mean square deviation is 3.01KW, experimental group 1.03KW smaller than control group RMSE mean square deviation 4.04KW
R2Mean square deviation is 0.02, than control group R2Mean square deviation 0.03 small 0.01.The MAPE average value of experimental group, RMSE average value and 3 are
Variance yields is respectively less than control group, and R2Average value is greater than control group.It is only several for the experimental group and control group of every day
The experimental result of its control group can be better than experimental group, this is because the best similar day of day to be predicted and adjacent day are same in the past few days
One day, two group models were equivalent at this time, and neural network each run the result is that random, so the result for control group occur is excellent
It is entirely sensible in experimental group.Therefore, the Elman neural network based on best similar day and the Elman mind based on adjacent day
Higher compared to precision through network, prediction effect is more preferable.
4 two groups of experiment prediction result comparisons of table
The foregoing is merely presently preferred embodiments of the present invention, all equivalent changes done according to scope of the present invention patent with
Modification, is all covered by the present invention.
Claims (6)
1. one kind is existed based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique, feature
In:Include the following steps:
Step S1:The meteorologic parameter on photovoltaic plant generated output over the years and weather station is acquired, daily parameter sample is obtained
Combination;
Step S2:Abnormal data is removed, the combination of parameter sample is normalized;
Step S3:Determine Elman neural network input layer number m, node in hidden layer n and output layer number of nodes p and
Initialize the parameters of Elman neural network;
Step S4:The Elman nerve obtained using the parameter sample combination training step S3 on the day before first day to day to be predicted
Network is constantly modified the hidden layer number of the Elman neural network by trial and error procedure, obtains training pattern;
Step S5:The best similar of day to be predicted is determined according to the Meteorological Characteristics value combination grey correlation analysis algorithm of day to be predicted
Day;
Step S6:By the parameter sample combination of best similar day and the training mould of the meteorologic parameter input step S4 of day to be predicted
The generated output of day to be predicted is predicted in type, obtains day to be predicted per output power value every other hour.
2. according to claim 1 a kind of based on best similar day and the short-term function of Elman neural fusion photovoltaic plant
Rate prediction technique, it is characterised in that:In step S1, the parameter sample combination includes photovoltaic plant generated output over the years and gas
As the illumination on station, environment temperature, humidity, wind speed information;
The parameter sample combination is denoted as (Pki, Gki, Dki, Tki, Wki, Hki), wherein k is the serial number on the date of sample collection, table
Show that number of days, k are 1 integer for arriving N;At the time of i is sample collection in one day, indicate that moment number, i are 1 integer for arriving Nt;PkiIt is
The power parameter sample at i-th of moment in parameter sample combination in k days, GkiFor i-th of moment in the combination of kth day parameter sample
Global horizontal radiation parameter sample, DkiFor the diffusion levels radiation parameter sample at i-th of moment in the combination of kth day parameter sample
This, TkiFor the environment temperature parameter sample at i-th of moment in the combination of kth day parameter sample, WkiFor kth day parameter sample group
The wind speed parameter sample at i-th of moment in conjunction, HkiFor the relative humidity ginseng at i-th of moment in the combination of kth day parameter sample
Numerical example.
3. according to claim 1 a kind of based on best similar day and the short-term function of Elman neural fusion photovoltaic plant
Rate prediction technique, it is characterised in that:In step S2, the combination of parameter sample is normalized specially:Using ratio pressure
The multi-group data at the same moment of same parameter sample is mapped in section [0,1] by contracting method, and mapping equation is:
In formula, y' indicates the data obtained after normalization, AimaxIndicate the maximum value in i-th of moment of data group A, AiminTable
Show the minimum value in i-th of moment of data group A, AkiIndicate some parameter at i-th of moment in the combination of kth day parameter sample
Sample.
4. according to claim 1 a kind of based on best similar day and the short-term function of Elman neural fusion photovoltaic plant
Rate prediction technique, it is characterised in that:In step S3, the Elman neural network is specifically configured to:Input layer number m is
23, node in hidden layer n are 10, and output layer number of nodes p is 11, and the number of iterations 2000, other parameters use default setting;Its
In, the formula of hidden layer number is as follows:
In formula, a is system parameter, and the value range of a is 1-10.
5. according to claim 1 a kind of based on best similar day and the short-term function of Elman neural fusion photovoltaic plant
Rate prediction technique, it is characterised in that:In step S4, walked using the parameter sample combined training on the day before first day to day to be predicted
The Elman neural network that rapid S3 is obtained is specially:According to date day to be predicted of selection, first day in history day is arrived to be predicted
One day parameter sample combination of day and the meteorologic parameter one day after of this day are as input, the correspondence moment one day after
Performance number as output, input and output combination is obtained, to training Elman neural network;
The input and output combination is denoted as:
In formula, k is the serial number on the date of sample collection, indicates that number of days, k are 1 integer for arriving N;Preceding 13 variables indicate that input is defeated
Input variable in combination out, the last one variable indicate output variable;The wherein P in input variablekFor kth day it is each when
The power parameter sample at quarter,WithFor the average daily global horizontal radiation parameter sample of kth day and kth+1 day,WithFor kth day and kth+1 day average daily diffusion levels radiation parameter sample, TkmaxAnd T(k+1)maxFor kth day and kth+1 day
Maximum environmental temperature parameter sample, TkminAnd T(k+1)minFor kth day and kth+1 day minimum environment temperature parameter sample,WithFor kth day and kth+1 day average daily wind speed parameter sample,WithFor the average daily relatively wet of kth day and kth+1 day
Spend parameter sample, output variable Pk+1For the power parameter sample at kth+1 day each moment.
6. according to claim 1 a kind of based on best similar day and the short-term function of Elman neural fusion photovoltaic plant
Rate prediction technique, it is characterised in that:Step S5 specifically includes following steps:
Step S51:The Meteorological Characteristics value is denoted as:
(Gkmax, Gkmin, Dkmax,Dkmin,Tkmax,Tkmin,Wkmax,Wkmin,Hkmax,Hkmin),
In formula, k is the serial number on the date of sample collection, indicates that number of days, k are 1 integer for arriving N;GkmaxAnd GkminIt is minimum and maximum
Global horizontal radiation parameter sample, DkmaxAnd DkminFor minimum and maximum diffusion levels radiation parameter sample, TkmaxAnd TkminFor most
Big and minimum environment temperature parameter sample, WkmaxAnd WkminFor minimum and maximum wind speed parameter sample, HkmaxAnd HkminTo be maximum and
Minimum relative humidity pa sample;
Step S52:The degree of association for calculating preset date and its a few days ago to be predicted, the degree of association maximum that day is determined as most preferably
Similar day;Wherein the calculating of the degree of association uses following formula:
In formula, riIndicate the degree of association of history day i and day to be predicted, k indicates characteristic value number, ξiIndicate indicate history day i with to
Survey the incidence coefficient of day;Wherein, ξi(k) calculating uses following formula:
In formula, y (k) indicates the Meteorological Characteristics value after day normalization to be predicted, xi(k) meteorology after indicating history day i normalization is special
Value indicative, ρ indicate that resolution ratio, ρ take 0.5, k to indicate characteristic value number.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810526061.XA CN108876013A (en) | 2018-05-29 | 2018-05-29 | One kind being based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810526061.XA CN108876013A (en) | 2018-05-29 | 2018-05-29 | One kind being based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108876013A true CN108876013A (en) | 2018-11-23 |
Family
ID=64335245
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810526061.XA Pending CN108876013A (en) | 2018-05-29 | 2018-05-29 | One kind being based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108876013A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110148068A (en) * | 2019-05-23 | 2019-08-20 | 福州大学 | One kind being based on meteorological data similarity analysis and LSTM neural fusion photovoltaic plant ultra-short term power forecasting method |
CN112508255A (en) * | 2020-12-01 | 2021-03-16 | 北京科技大学 | Photovoltaic output ultra-short-term prediction method and system based on multi-source heterogeneous data |
CN112884238A (en) * | 2021-03-12 | 2021-06-01 | 国网冀北电力有限公司电力科学研究院 | Photovoltaic power generation power prediction method and device |
CN112906987A (en) * | 2021-03-29 | 2021-06-04 | 福州大学 | Photovoltaic power prediction method based on convolutional neural network and two-dimensional meteorological matrix |
CN112927097A (en) * | 2021-01-29 | 2021-06-08 | 国网辽宁省电力有限公司阜新供电公司 | Photovoltaic power generation short-term prediction method based on GRA-ABC-Elman model |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104268638A (en) * | 2014-09-11 | 2015-01-07 | 广州市香港科大霍英东研究院 | Photovoltaic power generation system power predicting method of elman-based neural network |
CN104463349A (en) * | 2014-11-11 | 2015-03-25 | 河海大学 | Photovoltaic generated power prediction method based on multi-period comprehensive similar days |
CN105631558A (en) * | 2016-03-22 | 2016-06-01 | 国家电网公司 | BP neural network photovoltaic power generation system power prediction method based on similar day |
-
2018
- 2018-05-29 CN CN201810526061.XA patent/CN108876013A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104268638A (en) * | 2014-09-11 | 2015-01-07 | 广州市香港科大霍英东研究院 | Photovoltaic power generation system power predicting method of elman-based neural network |
CN104463349A (en) * | 2014-11-11 | 2015-03-25 | 河海大学 | Photovoltaic generated power prediction method based on multi-period comprehensive similar days |
CN105631558A (en) * | 2016-03-22 | 2016-06-01 | 国家电网公司 | BP neural network photovoltaic power generation system power prediction method based on similar day |
Non-Patent Citations (2)
Title |
---|
王晓兰等: "基于相似日和径向基函数神经网络的光伏阵列输出功率预测", 《电力自动化设备》 * |
钟春霞: "基于相似日选择算法和Elman神经网络的光伏输出功率预测", <南京工程学院学报:自然科学版> * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110148068A (en) * | 2019-05-23 | 2019-08-20 | 福州大学 | One kind being based on meteorological data similarity analysis and LSTM neural fusion photovoltaic plant ultra-short term power forecasting method |
CN112508255A (en) * | 2020-12-01 | 2021-03-16 | 北京科技大学 | Photovoltaic output ultra-short-term prediction method and system based on multi-source heterogeneous data |
CN112508255B (en) * | 2020-12-01 | 2021-09-07 | 北京科技大学 | Photovoltaic output ultra-short-term prediction method and system based on multi-source heterogeneous data |
CN112927097A (en) * | 2021-01-29 | 2021-06-08 | 国网辽宁省电力有限公司阜新供电公司 | Photovoltaic power generation short-term prediction method based on GRA-ABC-Elman model |
CN112884238A (en) * | 2021-03-12 | 2021-06-01 | 国网冀北电力有限公司电力科学研究院 | Photovoltaic power generation power prediction method and device |
CN112906987A (en) * | 2021-03-29 | 2021-06-04 | 福州大学 | Photovoltaic power prediction method based on convolutional neural network and two-dimensional meteorological matrix |
CN112906987B (en) * | 2021-03-29 | 2023-02-21 | 福州大学 | Photovoltaic power prediction method based on convolutional neural network and two-dimensional meteorological matrix |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Gao et al. | Day-ahead power forecasting in a large-scale photovoltaic plant based on weather classification using LSTM | |
CN108876013A (en) | One kind being based on best similar day and Elman neural fusion photovoltaic plant short term power prediction technique | |
Liu et al. | Prediction of short-term PV power output and uncertainty analysis | |
Li et al. | Renewable energy prediction: A novel short-term prediction model of photovoltaic output power | |
CN111626506B (en) | Regional photovoltaic power probability prediction method based on federal learning and cooperative regulation and control system thereof | |
CN106936147B (en) | A kind of optimization operation management method based on the micro-capacitance sensor of dual-layer optimization towards electric heat storage boiler | |
CN109002915A (en) | Photovoltaic plant short term power prediction technique based on Kmeans-GRA-Elman model | |
CN106446494B (en) | Honourable power forecasting method based on wavelet packet-neural network | |
CN102930358B (en) | A kind of neural net prediction method of photovoltaic power station power generation power | |
CN109858673A (en) | A kind of photovoltaic generating system power forecasting method | |
CN109685257A (en) | A kind of photovoltaic power generation power prediction method based on Support vector regression | |
CN110942205B (en) | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM | |
CN109376951B (en) | Photovoltaic probability prediction method | |
CN108205717A (en) | A kind of photovoltaic generation power Multiple Time Scales Forecasting Methodology | |
CN109978284B (en) | Photovoltaic power generation power time-sharing prediction method based on hybrid neural network model | |
CN109086928A (en) | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model | |
CN108280545A (en) | A kind of photovoltaic power prediction technique based on K mean cluster neural network | |
CN109934395B (en) | Multi-hydropower-region short-term power load prediction method based on time-sharing and regional meteorological data | |
CN105447509A (en) | Short-term power prediction method for photovoltaic power generation system | |
CN105160423A (en) | Photovoltaic power generation prediction method based on Markov residual error correction | |
Li et al. | Day-ahead hourly photovoltaic generation forecasting using extreme learning machine | |
Chen et al. | Research on wind power prediction method based on convolutional neural network and genetic algorithm | |
Tao et al. | On comparing six optimization algorithms for network-based wind speed forecasting | |
Zhong et al. | PV power short-term forecasting model based on the data gathered from monitoring network | |
CN117875752A (en) | Power system flexible operation domain assessment method based on self-organizing map decision tree |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181123 |