CN116826737A - Photovoltaic power prediction method, device, storage medium and equipment - Google Patents
Photovoltaic power prediction method, device, storage medium and equipment Download PDFInfo
- Publication number
- CN116826737A CN116826737A CN202310831284.8A CN202310831284A CN116826737A CN 116826737 A CN116826737 A CN 116826737A CN 202310831284 A CN202310831284 A CN 202310831284A CN 116826737 A CN116826737 A CN 116826737A
- Authority
- CN
- China
- Prior art keywords
- data
- weather
- photovoltaic power
- prediction
- meteorological
- 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 68
- 238000013528 artificial neural network Methods 0.000 claims description 59
- 239000011159 matrix material Substances 0.000 claims description 40
- 238000012549 training Methods 0.000 claims description 22
- 230000006870 function Effects 0.000 claims description 21
- 125000004122 cyclic group Chemical group 0.000 claims description 19
- 238000012216 screening Methods 0.000 claims description 18
- 238000012545 processing Methods 0.000 claims description 16
- 238000004422 calculation algorithm Methods 0.000 claims description 13
- 238000010606 normalization Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 11
- 230000002159 abnormal effect Effects 0.000 claims description 9
- 230000005855 radiation Effects 0.000 claims description 9
- 125000003275 alpha amino acid group Chemical group 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 6
- 150000001875 compounds Chemical class 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000010248 power generation Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 5
- 238000005070 sampling Methods 0.000 description 3
- 238000007619 statistical method Methods 0.000 description 3
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 2
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000000053 physical method Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 238000004138 cluster model Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000001556 precipitation Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012887 quadratic function Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
- H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2337—Non-hierarchical techniques using fuzzy logic, i.e. fuzzy clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
- G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computing Systems (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Evolutionary Biology (AREA)
- Primary Health Care (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Power Engineering (AREA)
- Water Supply & Treatment (AREA)
- Medical Informatics (AREA)
- Automation & Control Theory (AREA)
- Fuzzy Systems (AREA)
- Public Health (AREA)
- Operations Research (AREA)
Abstract
The invention discloses a photovoltaic power prediction method, a device, a storage medium and equipment, wherein the method comprises the following steps: acquiring output data and meteorological data of a target photovoltaic power station in a prediction period; determining the weather type of a day to be predicted according to meteorological data of a target photovoltaic power station in a prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather; and based on the weather type of the day to be predicted, inputting the output data and the meteorological data of the target photovoltaic power station in the prediction period into a pre-trained prediction model corresponding to the weather type of the day to be predicted, and obtaining a prediction result of the solar photovoltaic power to be predicted. The method and the device can improve the photovoltaic prediction precision and reduce the uncertainty of the grid caused by large-scale grid connection of the photovoltaic.
Description
Technical Field
The invention relates to a photovoltaic power prediction method, a photovoltaic power prediction device, a storage medium and photovoltaic power prediction equipment, and belongs to the technical field of distributed photovoltaic power generation.
Background
In recent years, the photovoltaic industry is rapidly developed, the grid-connected scale is continuously expanded, and solar energy is one of the basic energy sources for social production and living. However, since the photovoltaic equipment is directly exposed in the field environment, meteorological factors such as solar radiation, ambient temperature, precipitation and the like can directly influence the photovoltaic output, the photovoltaic output power always shows stronger volatility and randomness, and if large-scale photovoltaic grid connection is directly carried out, huge impact is brought to a power system. Therefore, the method has important significance in short-term prediction of the photovoltaic output and ensuring the accuracy of the prediction of the photovoltaic output, and the uncertainty brought by the photovoltaic grid connection is reduced.
At present, a great deal of research on short-term prediction of photovoltaic output power at home and abroad is carried out, and according to research results, the main prediction method can be summarized into the following three types: physical methods, statistical methods, and machine learning methods. The physical method is to construct a photovoltaic power generation physical model, and obtain photovoltaic output through model simulation, and the physical formula adopted by the method has errors with actual conditions, and is very dependent on input power station geographic information and numerical weather forecast information, so that the model has poor anti-interference capability and low stability. The statistical method is to analyze a large amount of historical data, and to correlate the power to be predicted with the historical power data by adopting statistical methods such as time sequence decomposition and the like to obtain a photovoltaic probability prediction result. The machine learning method is characterized in that a machine learning method is adopted, data change characteristics are extracted from meteorological data and photovoltaic output historical data to predict, the machine learning method is a main stream method for predicting the photovoltaic output at present, and the accuracy of a prediction result is high. The current improvement of the prediction method based on machine learning focuses on the improvement of a prediction model algorithm, the influence of environmental factors on model training under different sample conditions is not fundamentally considered, and the model precision is improved only.
The existing photovoltaic prediction model is subjected to weather typing based on historical meteorological data, but the photovoltaic prediction accuracy still needs to be improved. On one hand, the method is not comprehensive in consideration of environmental meteorological factors, does not screen out influence factors with high correlation degree with photovoltaic output power, and causes inaccurate weather classification, and on the other hand, a single algorithm is adopted, so that the prediction accuracy is not high.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a photovoltaic power prediction method, device, medium and equipment, which can obviously improve the photovoltaic prediction precision and reduce the uncertainty of large-scale grid connection of photovoltaic to a power grid. In order to achieve the above purpose, the invention is realized by adopting the following technical scheme:
in a first aspect, the present invention provides a method for predicting photovoltaic power, including:
acquiring output data and meteorological data of a target photovoltaic power station in a prediction period;
determining the weather type of a day to be predicted according to meteorological data of a target photovoltaic power station in a prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather;
based on the weather type of the day to be predicted, outputting data and meteorological data of a target photovoltaic power station in a prediction period are input into a pre-trained prediction model corresponding to the weather type of the day to be predicted, and a prediction result of the solar photovoltaic power to be predicted is obtained; wherein the pre-trained predictive model comprises: the method comprises the steps of a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model, a pre-trained extreme gradient lifting-gating circulating neural network cloudy prediction model, a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model and a pre-trained extreme gradient lifting-gating circulating neural network extreme weather prediction model.
With reference to the first aspect, optionally, the pre-trained prediction model is trained by:
acquiring output data and meteorological data of a target photovoltaic power station in a history period;
preprocessing output data and meteorological data of a target photovoltaic power station in a history period to obtain a history period data set;
based on the historical period data set, screening key meteorological factors influencing the output of the target photovoltaic power station, and carrying out weather clustering by taking the key meteorological factors as clustering features to obtain weather types of all data in the historical period data set;
and inputting the historical period data under different weather types in the historical period data set into a pre-constructed extreme gradient lifting-gating cyclic neural network combined prediction model for model training to obtain pre-trained prediction models corresponding to different weather types.
With reference to the first aspect, optionally, the preprocessing the output data and the meteorological data of the target photovoltaic power station in the history period to obtain a history period data set includes:
taking the value and the negative value exceeding the rated power of the target photovoltaic power station in the output data of the target photovoltaic power station in the history period as an abnormal value, taking the average of the values before and after the abnormal value and filling the abnormal value; for the missing value in the output data of the target photovoltaic power station in the history period, taking the average of the values before and after the missing value to fill the missing value;
Normalizing the filled output data and meteorological data by the following formula:
in the formula (1), x' represents a value after normalization processing, x represents a value before normalization processing, min (x) represents a minimum value of the feature value, and max (x) represents a maximum value of the feature value;
and (3) arranging the weather data subjected to normalization processing into an m×144×7 matrix, arranging the output data subjected to normalization processing into an m×144×1 matrix, wherein m represents the number of weather features, and obtaining a historical period data set.
With reference to the first aspect, optionally, the screening of key meteorological factors affecting the output of the target photovoltaic power station includes:
taking the meteorological data in the historical period data set as data of meteorological influence factors, wherein the meteorological influence factors comprise: direct radiation, scattered radiation, wind speed, ambient temperature, relative humidity, rainfall, pressure and wind direction;
calculating DCC indexes between the output data of the photovoltaic power station and the data of the meteorological influence factors respectively, and calculating by the following formula:
in the formula (2), d Cor Representing the calculated DCC index, X represents the output data of the photovoltaic power station, Y represents the data of any meteorological influence factor, and d Cov (X, Y) represents the distance covariance of X and Y, specifically defined as:
in the formula (3), n represents the length of data, and k and l represent the positions of the data in the historical period data set;
and comparing the calculated DCC indexes, and screening out weather influence factors with the maximum DCC indexes, namely the weather influence factors with the maximum influence on the output power of the photovoltaic power station, wherein the weather influence factors are used as key weather factors.
With reference to the first aspect, optionally, the clustering of weather with the key meteorological factors as clustering features includes:
calculating a statistical index of output data in a photovoltaic output key meteorological factor and historical period data set;
the calculated statistical index is used as the clustering characteristic of fuzzy C-means classification, and the number C of clustering centers and the membership matrix U are initialized (0) And a termination threshold for iterative computation;
iterative membership matrix U and clustering center V= [ V ] 1 ,v 2 ,…,v c ]Calculating a loss function in a fuzzy C-means clustering algorithm;
when reaching a preset iteration calculation termination threshold, carrying out iteration termination to obtain an optimal membership matrix U and a clustering center V;
and determining the category to which the sample belongs according to the optimal membership matrix U, and completing weather clustering to obtain the weather type to which each sample belongs in the historical period data set.
With reference to the first aspect, optionally, the statistical index is a mean value, a standard deviation, a kurtosis and a skewness, and the calculation is performed by the following formula:
in the formula (4), the amino acid sequence of the compound,mean value of data of certain key meteorological factors affecting photovoltaic output, N represents the number of key meteorological factors affecting photovoltaic output and X i Data representing a certain key meteorological factor affecting photovoltaic output; sigma represents the standard deviation of data of a certain key meteorological factor affecting the photovoltaic output; k (k) ur Kurtosis, P, of data representing a critical meteorological factor affecting photovoltaic output i Representing the output power of station i +.>Representing the average output power s k And the deviation of data which represents a certain key meteorological factor affecting the photovoltaic output is shown.
With reference to the first aspect, optionally, the number c of cluster centers and the membership matrix U are initialized, so that the membership matrix meets the following constraint:
in the formula (5), u ij Indicating the membership degree of the ith sample to the jth class, and n indicating the number of samples.
With reference to the first aspect, optionally, the calculating a loss function in the fuzzy C-means clustering algorithm includes:
calculating a clustering center V according to the membership matrix:
in the formula (6), the amino acid sequence of the compound,represents the j-th class cluster center in the first iteration, i represents the iteration number and x i The position of the ith sample is represented, m represents a membership factor, and the value range of m is 1-m-infinity;
updating the membership matrix U and calculating the clustering loss function J (l) ：
In the formula (7), d ij Representing the Euclidean distance of the sample to the cluster center, d ij ＝||x i -v j ||，d ik ＝||x i -v k ||；
In formula (8), U (l) Representing the membership matrix at the first iteration, V (l) Represents the cluster center, J, at the first iteration (l) (U (l) ,V (l) ) Representing when the membership matrix is U (l) The clustering center is V (l) Clustering loss at that time.
With reference to the first aspect, optionally, when the preset termination threshold of iterative computation is reached, the iteration is terminated, so as to obtain an optimal membership matrix U and a cluster center V, including:
when (when)Or |J (l) -J (l-1) |<ε J When the iteration is terminated; wherein ε is u Epsilon is the membership termination threshold u >0；ε J Epsilon for the loss function termination threshold J >0。
With reference to the first aspect, optionally, determining the category to which the sample belongs according to the optimal membership matrix U, and completing weather clustering, where weather clustering results are classified into sunny days, cloudy days, overcast and rainy days and extreme weather according to weather types; when u is ij ＝max 1≤j≤c {u ij Sample x at the time of } i Is classified as class j.
With reference to the first aspect, optionally, inputting the historical period data under different weather types in the historical period data set into a pre-constructed extreme gradient lifting-gating cyclic neural network combined prediction model for model training, including:
According to meteorological data in the historical period data set, screening meteorological data with data characteristics conforming to a weather type of a sunny day from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network sunny day prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model;
according to meteorological data in the historical period data set, screening meteorological data with data characteristics conforming to the weather type as cloudiness from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network cloudiness prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network cloudiness prediction model;
according to the meteorological data in the historical period data set, screening the meteorological data with data characteristics conforming to the weather type as overcast and rainy from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network overcast and rainy prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model;
According to the meteorological data in the historical period data set, the meteorological data with the data characteristics conforming to the extreme weather of the weather type is screened from the meteorological data in the preset time period to be used as a sample, and a pre-built extreme gradient lifting-gating cyclic neural network extreme weather prediction model is input to carry out model training, so that the pre-trained extreme gradient lifting-gating cyclic neural network extreme weather prediction model is obtained.
With reference to the first aspect, optionally, the obtaining a prediction result of the solar photovoltaic power to be predicted includes:
selecting a pre-trained prediction model corresponding to a weather type based on the weather type of the day to be predicted;
inputting output data and meteorological data of a target photovoltaic power station in a prediction period into an extreme gradient lifting algorithm to predict, so as to obtain a first prediction result;
inputting the first prediction result as an input characteristic into a gating circulating neural network for prediction to obtain a second prediction result;
and carrying out weighted summation on the first prediction result and the second prediction result by adopting an error reciprocal method to obtain a prediction result of the solar photovoltaic power to be predicted.
In a second aspect, the present invention provides a photovoltaic power prediction apparatus, comprising:
The acquisition module is used for: the method comprises the steps of obtaining output data and meteorological data of a target photovoltaic power station in a prediction period;
weather type determination module: the weather type of the day to be predicted is determined according to the meteorological data of the target photovoltaic power station in the prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather;
and a prediction output module: the method comprises the steps of inputting output data and meteorological data of a target photovoltaic power station in a prediction period based on weather types of the days to be predicted, and inputting a pre-trained prediction model corresponding to the weather types of the days to be predicted to obtain a prediction result of solar photovoltaic power to be predicted; wherein the pre-trained predictive model comprises: the method comprises the following steps of a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model, a pre-trained extreme gradient lifting-gating circulating neural network cloudy prediction model, a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model and a pre-trained extreme gradient lifting-gating circulating neural network extreme weather prediction model.
In a third aspect, the present invention provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method of predicting photovoltaic power as described in the first aspect.
In a fourth aspect, the present invention provides an apparatus comprising:
a memory for storing instructions;
a processor configured to execute the instructions, cause the apparatus to perform operations implementing the method of predicting photovoltaic power as described in the first aspect.
Compared with the prior art, the photovoltaic power prediction method, device, medium and equipment provided by the embodiment of the invention have the beneficial effects that:
the method comprises the steps of obtaining output data and meteorological data of a target photovoltaic power station in a prediction period; determining the weather type of a day to be predicted according to meteorological data of a target photovoltaic power station in a prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather; based on the weather type of the day to be predicted, outputting data and meteorological data of a target photovoltaic power station in a prediction period are input into a pre-trained prediction model corresponding to the weather type of the day to be predicted, and a prediction result of the solar photovoltaic power to be predicted is obtained; the prediction method can obviously reduce the error of photovoltaic prediction, improve the accuracy of photovoltaic short-term prediction, and has effectiveness and scientificity in actual application scenes;
According to the method, the output data and the meteorological data of the target photovoltaic power station in the history period are preprocessed, the key meteorological factors influencing the output of the target photovoltaic power station are screened, the key meteorological factors are used as clustering features for weather clustering, the key meteorological factors with higher correlation degree with the output power of the photovoltaic can be screened, and the weather classification is accurate;
according to the method, an extreme gradient lifting-gating cyclic neural network combined prediction model is adopted, the extreme gradient lifting-gating cyclic neural network combined prediction model is established for different weather types, real data training models of the different weather types are utilized to obtain pre-trained prediction models corresponding to the different weather types, and effectiveness and accuracy of the pre-trained prediction models are guaranteed.
Drawings
Fig. 1 is a flowchart of a photovoltaic power prediction method according to an embodiment of the present invention;
fig. 2 is a graph of a sunny day output characteristic in a photovoltaic power prediction method according to a second embodiment of the present invention;
fig. 3 is a characteristic graph of cloudiness in a photovoltaic power prediction method according to a second embodiment of the present invention;
fig. 4 is a graph of a characteristic of a rainy day output in a method for predicting photovoltaic power according to a second embodiment of the present invention;
FIG. 5 is a graph showing extreme weather output characteristics in a photovoltaic power prediction method according to a second embodiment of the present invention;
FIG. 6 is a graph of output power prediction results versus output characteristics using different prediction models on a sunny day;
FIG. 7 is a graph of output power prediction results versus output power characteristics using different prediction models for clouding;
FIG. 8 is a graph of output power prediction results versus output power characteristics using different prediction models during overcast and rainy conditions;
FIG. 9 is a graph of output power prediction results versus output power characteristics using different prediction models during extreme weather.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for more clearly illustrating the technical aspects of the present invention, and are not intended to limit the scope of the present invention.
Embodiment one:
the embodiment of the invention provides a photovoltaic power prediction method, which comprises the following steps:
acquiring output data and meteorological data of a target photovoltaic power station in a prediction period;
determining the weather type of a day to be predicted according to meteorological data of a target photovoltaic power station in a prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather;
And based on the weather type of the day to be predicted, inputting the output data and the meteorological data of the target photovoltaic power station in the prediction period into a pre-trained prediction model corresponding to the weather type of the day to be predicted, and obtaining a prediction result of the solar photovoltaic power to be predicted.
As shown in fig. 1, in this embodiment, a prediction model is trained first, and prediction is performed using the trained prediction model.
The method comprises the following specific steps:
step 1: and acquiring output data and meteorological data of the target photovoltaic power station in the historical period and the prediction period.
Meteorological data of a target photovoltaic power station in a certain period are acquired, wherein the meteorological data comprise direct radiation, scattered radiation, wind speed, ambient temperature, relative humidity, rainfall, pressure intensity and wind direction, and the data sampling interval is 5 minutes.
And acquiring output data of the target photovoltaic power station in a certain period, wherein the data sampling interval is 5 minutes.
Step 2: and preprocessing output data and meteorological data of the target photovoltaic power station in the history period to obtain a history period data set.
Step 2.1: and taking the value exceeding the rated power of the target photovoltaic power station and the negative value in the output data of the target photovoltaic power station in the historical period as an abnormal value, taking the average of the values before and after the abnormal value and filling the abnormal value.
In this example, values other than the mean ± 3 x standard deviation are also outliers.
Step 2.2: and taking the average of values before and after the missing value to fill the missing value for the missing value in the output data of the target photovoltaic power station in the history period.
Step 2.3: normalizing the filled output data and meteorological data by the following formula:
in the formula (1), x' represents a value after normalization processing, x represents a value before normalization processing, min (x) represents a minimum value of the feature value, and max (x) represents a maximum value of the feature value.
Step 2.4: and (3) arranging the weather data subjected to normalization processing into an m×144×7 matrix, arranging the output data subjected to normalization processing into an m×144×1 matrix, wherein m represents the number of weather features, and obtaining a historical period data set.
Step 3: and screening key meteorological factors influencing the output of the target photovoltaic power station based on the historical period data set, and clustering weather by taking the key meteorological factors as clustering features to obtain weather types of all data in the historical period data set.
Step 3.1: taking the meteorological data in the historical period data set as data of meteorological influence factors, respectively calculating DCC indexes between the output data of the photovoltaic power station and the data of the meteorological influence factors, and calculating by the following formula:
In the formula (2), d Cor Representing the calculated DCC index, X represents the output data of the photovoltaic power station, Y represents the data of any meteorological influence factor, and d Cov (X, Y) represents the distance covariance of X and Y, specifically defined as:
in the formula (3), n represents the length of data, and k and l represent the positions of the data in the history period data set.
Step 3.2: and comparing the calculated DCC indexes, and screening out weather influence factors with the maximum DCC indexes, namely the weather influence factors with the maximum influence on the output power of the photovoltaic power station, wherein the weather influence factors are used as key weather factors.
And screening out indexes with larger influence on the output power of the photovoltaic power station as key meteorological factors, participating in subsequent analysis, and taking other meteorological factors into consideration.
Step 3.3: and calculating statistical indexes of the output data in the photovoltaic output key meteorological factors and the historical period data set.
In this embodiment, the key meteorological factors affecting the photovoltaic output are direct radiation, scattered radiation, wind speed and temperature.
The statistical indexes are mean value, standard deviation, kurtosis and skewness, and are calculated by the following formula:
in the formula (4), the amino acid sequence of the compound,mean value of data of certain key meteorological factors affecting photovoltaic output, N represents the number of key meteorological factors affecting photovoltaic output and X i Data representing a certain key meteorological factor affecting photovoltaic output; sigma represents the standard deviation of data of a certain key meteorological factor affecting the photovoltaic output; k (k) ur Kurtosis, P, of data representing a critical meteorological factor affecting photovoltaic output i Representing the output power of station i +.>Which represents the average output power of the device,s k and the deviation of data which represents a certain key meteorological factor affecting the photovoltaic output is shown.
Step 3.4: and taking the calculated statistical index as a clustering feature of fuzzy C-means classification, and classifying weather by using a fuzzy C-means classification method (FCM clustering algorithm).
Step 3.4.1: initializing the number c of clustering centers and a membership matrix U, so that the membership matrix meets the following constraint:
in the formula (5), u ij Indicating the membership degree of the ith sample to the jth class, and n indicating the number of samples.
Step 3.4.2: iterative membership matrix U and clustering center V= [ V ] 1 ,v 2 ,…,v c ]And calculating a loss function in the fuzzy C-means clustering algorithm.
Calculating a clustering center V according to the membership matrix:
in the formula (6), the amino acid sequence of the compound,represents the j-th class cluster center in the first iteration, i represents the iteration number and x i The position of the ith sample is represented, and m represents a membership factor (1.ltoreq.m.ltoreq.infinity).
Updating the membership matrix U and calculating the clustering loss function J (l) ：
In the formula (7), d ij Representing the Euclidean distance of the sample to the cluster center, d ij ＝||x i -v j ||，d ik ＝||x i -v k ||。
In formula (8), U (l) Representing the membership matrix at the first iteration, V (l) Represents the cluster center, J, at the first iteration (l) (U (l) ,V (l) ) Representing when the membership matrix is U (l) The clustering center is V (l) Clustering loss at that time.
Step 3.4.3: initializing a membership termination threshold epsilon u >0, loss function termination threshold ε J >0, whenOr |J (l) -J (l-1) |<ε J The iteration terminates, otherwise it returns to step 3.4.2. And when the iteration is terminated, obtaining an optimal membership matrix U and a clustering center V.
And determining the category to which the sample belongs according to the optimal membership matrix U, and completing weather clustering to obtain the weather type to which each sample belongs in the historical period data set.
When u is ij ＝max 1≤j≤c {u ij Sample x at the time of } i Is classified as class j.
The clustering results are classified into sunny days, cloudy days, overcast and rainy days and extreme weather according to weather types.
According to the method, the key meteorological factors are used as clustering features for weather clustering, so that the key meteorological factors with high correlation with the photovoltaic output power can be screened out, and the weather classification is accurate.
Step 4: and inputting the historical period data under different weather types in the historical period data set into a pre-constructed extreme gradient lifting-gating cyclic neural network combined prediction model for model training to obtain pre-trained prediction models corresponding to different weather types.
According to meteorological data in the historical period data set, screening meteorological data with data characteristics conforming to a weather type of a sunny day from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network sunny day prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model;
according to meteorological data in the historical period data set, screening meteorological data with data characteristics conforming to the weather type as cloudiness from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network cloudiness prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network cloudiness prediction model;
according to the meteorological data in the historical period data set, screening the meteorological data with data characteristics conforming to the weather type as overcast and rainy from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network overcast and rainy prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model;
According to the meteorological data in the historical period data set, the meteorological data with the data characteristics conforming to the extreme weather of the weather type is screened from the meteorological data in the preset time period to be used as a sample, and a pre-built extreme gradient lifting-gating cyclic neural network extreme weather prediction model is input to carry out model training, so that the pre-trained extreme gradient lifting-gating cyclic neural network extreme weather prediction model is obtained.
According to the method, an extreme gradient lifting-gating cyclic neural network combined prediction model is adopted, the extreme gradient lifting-gating cyclic neural network combined prediction model is established for different weather types, real data training models of the different weather types are utilized to obtain pre-trained prediction models corresponding to the different weather types, and effectiveness and accuracy of the pre-trained prediction models are guaranteed.
Step 5: determining the weather type of a day to be predicted according to meteorological data of a target photovoltaic power station in a prediction period; wherein the weather type includes: sunny days, cloudy days, overcast and rainy days and extreme weather.
Step 6: and based on the weather type of the day to be predicted, inputting the output data and the meteorological data of the target photovoltaic power station in the prediction period into a pre-trained prediction model corresponding to the weather type of the day to be predicted, and obtaining a prediction result of the solar photovoltaic power to be predicted.
Step 6.1: based on the weather type of the day to be predicted, a pre-trained prediction model corresponding to the weather type is selected.
Step 6.2: and inputting the output data and the meteorological data of the target photovoltaic power station in the prediction period into an extreme gradient lifting algorithm (XGBoost) for prediction to obtain a first prediction result.
The objective function of the extreme gradient lifting algorithm is:
in the formula (9), y i Representing the actual value of the model for the i-th sample,model predictive value representing the ith sample, < +.>Training error for measuring model predictive result and actual value, Ω (f) k ) A regularization term representing a kth predictive model.
The objective function can be converted into a unitary quadratic function, and the optimal omega and the objective function value are obtained by solving the following specific formulas:
in the formula (10):represents the most significant of ωFigure of merit, G j Representing the sum of the first derivatives +.>H j Representing the sum of the second derivatives>h i Representing the function l pair->Is a second derivative of (2); λ represents a parameter used to control leaf node score; x is X obj Let r denote the objective function value, r denote the parameters used to control the number of leaf nodes, and T denote the number of leaf nodes.
Step 6.3: and inputting the first prediction result as an input characteristic into a gate-controlled cyclic neural network (GRU) for prediction to obtain a second prediction result.
The calculation formula of the gating cyclic neural network is as follows:
in formula (11), z t Representing an update gate, sigma is a ReLU activation function, and the value range is [0,1 ]]，W z Represents the updated gate weight, h t-1 Represents the hidden layer output at time t-1, x t Data information indicating time t; r is (r) t Representing reset gate, W r Representing a reset gate weight;representing a candidate set, W representing a weight coefficient; h is a t Represents hidden layer output at time t, which represents the inner product of the vector.
Step 6.4: and carrying out weighted summation on the first prediction result and the second prediction result by adopting an error reciprocal method to obtain a prediction result of the solar photovoltaic power to be predicted.
The calculation formula of the error reciprocal method is as follows:
f PV ＝ω 1 f X +ω 2 f L
in the formula (12), f PV Representing the predicted result, ω, of the solar photovoltaic power to be predicted 1 Weight coefficient representing first prediction result, f X Omega as the first predictor 2 Weight coefficient representing second prediction result, f L Is the second predicted result; e, e 1 Representing a first predictor error, e 2 Representing a second predictor error.
The prediction method can obviously reduce the error of photovoltaic prediction, improve the accuracy of photovoltaic short-term prediction, and has effectiveness and scientificity in actual application scenes.
Embodiment two:
the embodiment of the invention adopts the public data set provided by DKASC (Desert Knowledge Australia Solar Center) to verify the photovoltaic power prediction method provided by the first embodiment.
DKASC is located in australia and a data set of a set of terrestrial photovoltaic power plants comprising 22 fast photovoltaic panels, each rated at 250W, with a total power of 5.5kW was selected for example analysis. The power station data set comprises photovoltaic output data, current day meteorological data and meteorological prediction data, and the data sampling interval is 5 minutes. And selecting 1 month to 1 month in one year as a history period, wherein data of the history period is used as a training sample, and 1 month to 9 months in the next year are used as a prediction period, and data of the prediction period are used as a test sample.
Ignoring data when power generation is not performed at night (power is 0), selecting 07:00-19:00 data for FCM cluster analysis, constructing an FCM cluster model by MATLAB software, wherein a matrix segmentation index is 2, the maximum number of iterations is 100, the number of clusters is 4, a fuzzy weighting parameter is 2, and a threshold is 10-5. Inputting the 365 samples subjected to pretreatment into an FCM clustering algorithm to obtain the following different weather types in 4: the output characteristic curves for the 4 weather types on sunny days, cloudy days, overcast days and extreme weather are shown in fig. 2-5. In sunny days, the overall photovoltaic output condition is stable, and the output curve has obvious time characteristics; when cloudy, the photovoltaic output curve has obvious fluctuation in the latter half section; in overcast and rainy days, the fluctuation is mainly concentrated in the middle section and the rear half section; in extreme weather, the photovoltaic output curve exhibits irregular, high-frequency, and severe fluctuations.
Based on FCM clustering results, the follow-up predictions are respectively developed for 4 weather types of sunny days, cloudy days, rainy days and extreme weather. The prediction of solar photovoltaic output power is performed by adopting an extreme gradient lifting-gating cyclic neural network combined prediction model, and the model provided by the embodiment I is verified to have effectiveness and accuracy by comparing the result of prediction by adopting a model of single extreme gradient lifting (XGBoost) and a single gating cyclic neural network (GRU) model under the condition of not performing FCM weather clustering and FCM weather clustering.
The calculation results are shown in Table 1 and FIGS. 6 to 9. In fig. 6 to 9, the abscissa indicates the predicted data time point, the interval between two adjacent points is 5 minutes, and the ordinate indicates the output power.
The average RMSE using the FCM-XGBoost-GRU prediction model provided in example one was 0.242kw and the average MAPE was 3.999%. Compared with XGBoost-GRU, FCM-XGBoost and FCM-GRU, the average RMSE is respectively reduced by 0.174, 0.092 and 0.179, and the average MAPE is respectively reduced by 2.004%, 2.580 and 4.577%, which shows that the anti-interference capability of the model can be improved by adding FCM for relieving, and the overall prediction stability is ensured.
In summary, the FCM-XGBoost-GRU photovoltaic prediction model provided in the photovoltaic power prediction method provided in the first embodiment can improve the overall accuracy of the photovoltaic short-term output prediction, has high robustness, and is beneficial to the stable operation of the power grid system.
Embodiment III:
the embodiment of the invention provides a photovoltaic power prediction device, which comprises:
the acquisition module is used for: the method comprises the steps of obtaining output data and meteorological data of a target photovoltaic power station in a prediction period;
weather type determination module: the weather type of the day to be predicted is determined according to the meteorological data of the target photovoltaic power station in the prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather;
and a prediction output module: the method comprises the steps of inputting output data and meteorological data of a target photovoltaic power station in a prediction period based on weather types of the days to be predicted, and inputting a pre-trained prediction model corresponding to the weather types of the days to be predicted to obtain a prediction result of solar photovoltaic power to be predicted; wherein the pre-trained predictive model comprises: the method comprises the following steps of a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model, a pre-trained extreme gradient lifting-gating circulating neural network cloudy prediction model, a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model and a pre-trained extreme gradient lifting-gating circulating neural network extreme weather prediction model.
Embodiment four:
An embodiment of the present application provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements a method for predicting photovoltaic power as described in embodiment one.
Fifth embodiment:
the embodiment of the application also provides equipment, which comprises:
a memory for storing instructions;
a processor configured to execute the instructions, cause the apparatus to perform operations implementing the method for predicting photovoltaic power as described in embodiment one.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (15)
1. A method for predicting photovoltaic power, comprising:
acquiring output data and meteorological data of a target photovoltaic power station in a prediction period;
determining the weather type of a day to be predicted according to meteorological data of a target photovoltaic power station in a prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather;
based on the weather type of the day to be predicted, outputting data and meteorological data of a target photovoltaic power station in a prediction period are input into a pre-trained prediction model corresponding to the weather type of the day to be predicted, and a prediction result of the solar photovoltaic power to be predicted is obtained; wherein the pre-trained predictive model comprises: the method comprises the steps of a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model, a pre-trained extreme gradient lifting-gating circulating neural network cloudy prediction model, a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model and a pre-trained extreme gradient lifting-gating circulating neural network extreme weather prediction model.
2. The method of claim 1, wherein the pre-trained predictive model is trained by:
Acquiring output data and meteorological data of a target photovoltaic power station in a history period;
preprocessing output data and meteorological data of a target photovoltaic power station in a history period to obtain a history period data set;
based on the historical period data set, screening key meteorological factors influencing the output of the target photovoltaic power station, and carrying out weather clustering by taking the key meteorological factors as clustering features to obtain weather types of all data in the historical period data set;
and inputting the historical period data under different weather types in the historical period data set into a pre-constructed extreme gradient lifting-gating cyclic neural network combined prediction model for model training to obtain pre-trained prediction models corresponding to different weather types.
3. The method for predicting photovoltaic power according to claim 2, wherein preprocessing the output data and the meteorological data of the target photovoltaic power station in the history period to obtain a history period data set comprises:
taking the value and the negative value exceeding the rated power of the target photovoltaic power station in the output data of the target photovoltaic power station in the history period as an abnormal value, taking the average of the values before and after the abnormal value and filling the abnormal value; for the missing value in the output data of the target photovoltaic power station in the history period, taking the average of the values before and after the missing value to fill the missing value;
Normalizing the filled output data and meteorological data by the following formula:
in the formula (1), x' represents a value after normalization processing, x represents a value before normalization processing, min (x) represents a minimum value of the feature value, and max (x) represents a maximum value of the feature value;
and (3) arranging the weather data subjected to normalization processing into an m×144×7 matrix, arranging the output data subjected to normalization processing into an m×144×1 matrix, wherein m represents the number of weather features, and obtaining a historical period data set.
4. The method of claim 2, wherein the screening for key meteorological factors affecting the output of the target photovoltaic power plant comprises:
taking the meteorological data in the historical period data set as data of meteorological influence factors, wherein the meteorological influence factors comprise: direct radiation, scattered radiation, wind speed, ambient temperature, relative humidity, rainfall, pressure and wind direction;
calculating DCC indexes between the output data of the photovoltaic power station and the data of the meteorological influence factors respectively, and calculating by the following formula:
in the formula (2), d Cor Representing the calculated DCC index, X represents the output data of the photovoltaic power station, Y represents the data of any meteorological influence factor, and d Cov (X, Y) represents the distance covariance of X and Y, specifically defined as:
in the formula (3), n represents the length of data, k and l represent the positions of the data in a historical period data set, the value range of k is 1-n, and the value range of l is 1-n;
and comparing the calculated DCC indexes, and screening out weather influence factors with the maximum DCC indexes, namely the weather influence factors with the maximum influence on the output power of the photovoltaic power station, wherein the weather influence factors are used as key weather factors.
5. The method for predicting photovoltaic power according to claim 2, wherein the clustering of weather with key meteorological factors as clustering features comprises:
calculating a statistical index of output data in a photovoltaic output key meteorological factor and historical period data set;
the calculated statistical indexes are used as clustering features of fuzzy C-means classification, and the number C of clustering centers, the membership matrix U and the termination threshold value of iterative computation are initialized;
iterative membership matrix U and clustering center V= [ V ] 1 ,v 2 ,…,v c ]Calculating a loss function in a fuzzy C-means clustering algorithm;
when reaching a preset iteration calculation termination threshold, carrying out iteration termination to obtain an optimal membership matrix U and a clustering center V;
and determining the category to which the sample belongs according to the optimal membership matrix U, and completing weather clustering to obtain the weather type to which each sample belongs in the historical period data set.
6. The method of claim 5, wherein the statistical indicators are mean, standard deviation, kurtosis, and skewness, and are calculated by:
in the formula (4), the amino acid sequence of the compound,mean value of data of certain key meteorological factors affecting photovoltaic output, N represents the number of key meteorological factors affecting photovoltaic output and X i Data representing a certain key meteorological factor affecting photovoltaic output; sigma represents the standard deviation of data of a certain key meteorological factor affecting the photovoltaic output; k (k) ur Kurtosis, P, of data representing a critical meteorological factor affecting photovoltaic output i Representing the output power of station i +.>Representing the average output power s k And the deviation of data which represents a certain key meteorological factor affecting the photovoltaic output is shown.
7. The method for predicting photovoltaic power according to claim 5, wherein the number of cluster centers c and the membership matrix U are initialized such that the membership matrix satisfies the following constraint:
in the formula (5), u ij Indicating the membership degree of the ith sample to the jth class, and n indicating the number of samples.
8. The method of claim 7, wherein calculating a loss function in a fuzzy C-means clustering algorithm comprises:
Calculating a clustering center V according to the membership matrix:
in the formula (6), the amino acid sequence of the compound,represents the j-th class cluster center in the first iteration, i represents the iteration number and x i The position of the ith sample is represented, m represents a membership factor, and the value range of m is 1-m-infinity;
updating the membership matrix U and calculating the clustering loss function J (l) ：
In the formula (7), d ij Representing the Euclidean distance of the sample to the cluster center, d ij ＝||x i -v j ||，d ik ＝||x i -v k ||；
In formula (8), U (l) Representing the membership matrix at the first iteration, V (l) Represents the cluster center, J, at the first iteration (l) (U (l) ,V (l) ) Representing when the membership matrix is U (l) The clustering center is V (l) Clustering loss at that time.
9. The method for predicting photovoltaic power according to claim 8, wherein when the preset termination threshold of iterative computation is reached, the iteration is terminated to obtain an optimal membership matrix U and a cluster center V, including:
when (when)Or |J (l) -J (l-1) |<ε J When the iteration is terminated; wherein ε is u Epsilon is the membership termination threshold u >0；ε J Epsilon for the loss function termination threshold J >0。
10. The method for predicting the photovoltaic power according to claim 9, wherein the determining the class to which the sample belongs according to the optimal membership matrix U completes the weather clustering, wherein the weather clustering result is classified into sunny days, cloudy days, overcast and rainy days and extreme weather according to the weather type; when u is ij ＝max 1≤j≤c {u ij Sample x at the time of } i Is classified as class j.
11. The method for predicting photovoltaic power according to claim 2, wherein the inputting the historical period data under different weather types in the historical period data set into the pre-constructed extreme gradient boost-gating cyclic neural network combined prediction model for model training comprises:
according to meteorological data in the historical period data set, screening meteorological data with data characteristics conforming to a weather type of a sunny day from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network sunny day prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model;
according to meteorological data in the historical period data set, screening meteorological data with data characteristics conforming to the weather type as cloudiness from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network cloudiness prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network cloudiness prediction model;
according to the meteorological data in the historical period data set, screening the meteorological data with data characteristics conforming to the weather type as overcast and rainy from the meteorological data in a preset time period as a sample, and inputting a pre-constructed extreme gradient lifting-gating circulating neural network overcast and rainy prediction model for model training to obtain a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model;
According to the meteorological data in the historical period data set, the meteorological data with the data characteristics conforming to the extreme weather of the weather type is screened from the meteorological data in the preset time period to be used as a sample, and a pre-built extreme gradient lifting-gating cyclic neural network extreme weather prediction model is input to carry out model training, so that the pre-trained extreme gradient lifting-gating cyclic neural network extreme weather prediction model is obtained.
12. The method for predicting photovoltaic power according to claim 1, wherein the obtaining the predicted result of the solar photovoltaic power to be predicted includes:
selecting a pre-trained prediction model corresponding to a weather type based on the weather type of the day to be predicted;
inputting output data and meteorological data of a target photovoltaic power station in a prediction period into an extreme gradient lifting algorithm to predict, so as to obtain a first prediction result;
inputting the first prediction result as an input characteristic into a gating circulating neural network for prediction to obtain a second prediction result;
and carrying out weighted summation on the first prediction result and the second prediction result by adopting an error reciprocal method to obtain a prediction result of the solar photovoltaic power to be predicted.
13. A photovoltaic power generation prediction apparatus, comprising:
The acquisition module is used for: the method comprises the steps of obtaining output data and meteorological data of a target photovoltaic power station in a prediction period;
weather type determination module: the weather type of the day to be predicted is determined according to the meteorological data of the target photovoltaic power station in the prediction period; wherein the weather type includes: sunny days, cloudy, overcast and rainy days and extreme weather;
and a prediction output module: the method comprises the steps of inputting output data and meteorological data of a target photovoltaic power station in a prediction period based on weather types of the days to be predicted, and inputting a pre-trained prediction model corresponding to the weather types of the days to be predicted to obtain a prediction result of solar photovoltaic power to be predicted; wherein the pre-trained predictive model comprises: the method comprises the following steps of a pre-trained extreme gradient lifting-gating circulating neural network sunny day prediction model, a pre-trained extreme gradient lifting-gating circulating neural network cloudy prediction model, a pre-trained extreme gradient lifting-gating circulating neural network overcast and rainy prediction model and a pre-trained extreme gradient lifting-gating circulating neural network extreme weather prediction model.
14. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the method of predicting photovoltaic power according to any one of claims 1-12.
15. An apparatus, comprising:
a memory for storing instructions;
a processor for executing the instructions to cause the apparatus to perform operations implementing the method of predicting photovoltaic power of any one of claims 1-12.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310831284.8A CN116826737A (en) | 2023-07-06 | 2023-07-06 | Photovoltaic power prediction method, device, storage medium and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310831284.8A CN116826737A (en) | 2023-07-06 | 2023-07-06 | Photovoltaic power prediction method, device, storage medium and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116826737A true CN116826737A (en) | 2023-09-29 |
Family
ID=88112519
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310831284.8A Pending CN116826737A (en) | 2023-07-06 | 2023-07-06 | Photovoltaic power prediction method, device, storage medium and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116826737A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117833237A (en) * | 2024-01-08 | 2024-04-05 | 广西北部湾陆海新能源股份有限公司 | Method and system for predicting output of photovoltaic power station based on model output prediction sequence |
| CN118017597A (en) * | 2024-02-07 | 2024-05-10 | 北京四方继保自动化股份有限公司 | Photovoltaic power station output prediction method, device, equipment and medium |
-
2023
- 2023-07-06 CN CN202310831284.8A patent/CN116826737A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117833237A (en) * | 2024-01-08 | 2024-04-05 | 广西北部湾陆海新能源股份有限公司 | Method and system for predicting output of photovoltaic power station based on model output prediction sequence |
| CN118017597A (en) * | 2024-02-07 | 2024-05-10 | 北京四方继保自动化股份有限公司 | Photovoltaic power station output prediction method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112949945B (en) | Wind power ultra-short-term prediction method for improving bidirectional long-term and short-term memory network | |
| CN106650767B (en) | Flood forecasting method based on cluster analysis and real-time correction | |
| CN107766990B (en) | Method for predicting power generation power of photovoltaic power station | |
| CN108921339B (en) | Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine | |
| CN112288164B (en) | Wind power combined prediction method considering spatial correlation and correcting numerical weather forecast | |
| CN111260126B (en) | Short-term photovoltaic power generation prediction method considering correlation degree of weather and meteorological factors | |
| CN113554466B (en) | Short-term electricity consumption prediction model construction method, prediction method and device | |
| CN105701572B (en) | Photovoltaic short-term output prediction method based on improved Gaussian process regression | |
| CN116826737A (en) | Photovoltaic power prediction method, device, storage medium and equipment | |
| CN115130741A (en) | Multi-model fusion based multi-factor power demand medium and short term prediction method | |
| CN113205226A (en) | Photovoltaic power prediction method combining attention mechanism and error correction | |
| Zhang et al. | Wind speed prediction research considering wind speed ramp and residual distribution | |
| CN112418346B (en) | Numerical weather forecast total radiation system error classification calculation method | |
| CN115115125B (en) | Photovoltaic power interval probability prediction method based on deep learning fusion model | |
| CN115796004A (en) | Photovoltaic power station ultra-short term power intelligent prediction method based on SLSTM and MLSTNet models | |
| CN114792156A (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
| CN114462718A (en) | CNN-GRU wind power prediction method based on time sliding window | |
| CN115860177A (en) | Photovoltaic power generation power prediction method based on combined machine learning model and application thereof | |
| CN116341613A (en) | Ultra-short-term photovoltaic power prediction method based on Informar encoder and LSTM | |
| CN114372631A (en) | Data-lacking area runoff prediction method based on small sample learning and LSTM | |
| CN115759389A (en) | Day-ahead photovoltaic power prediction method based on weather type similar day combination strategy | |
| CN115204444A (en) | Photovoltaic power prediction method based on improved cluster analysis and fusion integration algorithm | |
| CN114897204A (en) | Method and device for predicting short-term wind speed of offshore wind farm | |
| CN118428754A (en) | New energy power generation input power prediction method and system | |
| CN113095547A (en) | Short-term wind power prediction method based on GRA-LSTM-ICE model |
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 |