CN116956753B - Distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution - Google Patents
Distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution Download PDFInfo
- Publication number
- CN116956753B CN116956753B CN202311218640.5A CN202311218640A CN116956753B CN 116956753 B CN116956753 B CN 116956753B CN 202311218640 A CN202311218640 A CN 202311218640A CN 116956753 B CN116956753 B CN 116956753B
- Authority
- CN
- China
- Prior art keywords
- distributed photovoltaic
- clustering
- model
- data
- photovoltaic
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 125000004122 cyclic group Chemical group 0.000 title claims abstract description 39
- 238000000034 method Methods 0.000 title claims abstract description 31
- 238000002922 simulated annealing Methods 0.000 title claims abstract description 26
- 238000012417 linear regression Methods 0.000 claims abstract description 20
- 238000012937 correction Methods 0.000 claims abstract description 19
- 238000005259 measurement Methods 0.000 claims abstract description 11
- 238000013528 artificial neural network Methods 0.000 claims abstract description 5
- 238000013527 convolutional neural network Methods 0.000 claims description 16
- 238000005457 optimization Methods 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 11
- 230000001419 dependent effect Effects 0.000 claims description 6
- 238000003064 k means clustering Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000010248 power generation Methods 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2111/00—Details relating to CAD techniques
- G06F2111/06—Multi-objective optimisation, e.g. Pareto optimisation using simulated annealing [SA], ant colony algorithms or genetic algorithms [GA]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/04—Power grid distribution networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Probability & Statistics with Applications (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Photovoltaic Devices (AREA)
Abstract
The invention provides a distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution, comprising the following steps: collecting actual measurement meteorological data and actual photovoltaic output data of each distributed photovoltaic station at fixed interval time length; acquiring predicted meteorological data of a corresponding time period; establishing a predicted weather correction model, and performing model correction on predicted weather data of each distributed photovoltaic site; clustering the distributed photovoltaic sites by different methods; based on the corrected predicted meteorological data, the corrected actual photovoltaic output data and the clustering result, a model of the predicted meteorological-photovoltaic output is built in each class by using a cyclic convolution neural network algorithm, and a final predicted result is obtained by using a multiple linear regression algorithm. The invention improves the photovoltaic output prediction speed and prediction precision of the distributed photovoltaic.
Description
Technical Field
The invention belongs to the technical field of new energy power, and particularly relates to a distributed photovoltaic output prediction method and device based on simulated annealing and a cyclic convolution neural network.
Background
As an important component of the novel power system, distributed photovoltaic is an essential research direction aiming at improving the accuracy of output prediction of the distributed photovoltaic.
At present, the installed capacity of the distributed photovoltaic is gradually increased, and some provinces of the installed capacity of the distributed photovoltaic even exceed more than 50% of the total installed capacity of the photovoltaic, so that the prediction accuracy of the distributed photovoltaic has great influence on the balance of a power grid, a scheduling plan, the absorption of the noon peak and the like.
Because the distributed photovoltaic distribution range is wider, a certain degree of difficulty is caused to data acquisition and transmission, and therefore, the distributed prediction is often performed in a template station conversion mode. Specifically, a site with convenient data acquisition is selected, power prediction modeling of a single site is carried out on the site, and power prediction conversion in the whole area is carried out according to the installed corresponding relation between the site and the whole area or the power generation corresponding relation and the like.
The photovoltaic power generation capacity is greatly influenced by meteorological elements such as cloud movement, thickness of high-middle-low layer cloud, irradiance, humidity and temperature, the model establishment aiming at the template station can only be characterized aiming at the meteorological influence of a single point of the template station, and the distance between the template stations is generally long, so that the modeling mode of template station conversion is difficult to describe the influence of the meteorological elements such as cloud movement on the distributed photovoltaic between the template stations, and the photovoltaic prediction accuracy of the whole area is negatively influenced.
Disclosure of Invention
The invention provides a distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution, which adopt a regional prediction scheme to integrally predict a distributed photovoltaic region so as to achieve the aim of improving the accuracy of the distributed photovoltaic prediction.
In order to achieve the above purpose, the technical scheme of the invention is realized as follows:
a distributed photovoltaic prediction method based on simulated annealing and cyclic convolution, comprising:
s1, collecting actual measurement meteorological data and actual photovoltaic output data of each distributed photovoltaic station at fixed interval time; acquiring predicted meteorological data of a corresponding time period;
s2, establishing a predicted weather correction model, and performing model correction on predicted weather data of each distributed photovoltaic site to obtain corrected predicted weather data;
s3, clustering the distributed photovoltaic stations by adopting a K-MEANS algorithm based on the coordinates of the distributed photovoltaic stations; based on the corrected predicted meteorological data of each distributed photovoltaic station, clustering the predicted meteorological data by adopting a hierarchical clustering algorithm; clustering the actual measurement meteorological data of each distributed photovoltaic station by adopting a KNN clustering algorithm;
s4, based on the corrected predicted meteorological data, the corrected actual photovoltaic output data and the clustering result obtained in the S3, a model of the predicted meteorological-photovoltaic output is built in each type by using a cyclic convolution neural network algorithm, when model training is carried out, super-parameter optimizing is carried out by adopting a simulated annealing algorithm, and finally, final prediction is carried out on the prediction result of each type of model by adopting a multiple linear regression algorithm, so that a final prediction result is obtained.
Further, in step S2, the establishing of the predicted weather correction model includes: and taking the actually measured meteorological data of the distributed photovoltaic stations as variables to be fitted, taking the predicted meteorological data as input variables, performing feature optimization and super-parameter optimization based on the data, and establishing a predicted meteorological correction model.
Further, in step S3, the specific step of performing K-MEANS clustering based on the coordinates of each distributed photovoltaic site includes: and (3) adopting K-MEANS clustering to calculate the distances between the coordinates of different distributed photovoltaic sites, and gathering the coordinates with the short distances into one type according to the distances between the coordinates.
Further, in step S3, the specific step of hierarchical clustering based on the predicted meteorological data of each distributed photovoltaic site includes: and (3) hierarchical clustering is adopted, similarity among predicted meteorological data sequences of different distributed photovoltaic sites is calculated, and sequences with larger similarity are clustered into one class according to the similarity among different data sequences.
Further, in step S3, the specific step of performing KNN clustering based on the measured meteorological data of each distributed photovoltaic site includes: and (3) calculating the similarity between the actually measured meteorological data sequences of different distributed photovoltaic sites by adopting KNN clustering, and gathering the sequences with larger similarity into one type according to the similarity between different data sequences.
Further, in step S4, the model training specifically includes:
s401, taking the corrected predicted meteorological data as input of a cyclic convolutional neural network model, and taking the actual photovoltaic output data as a target to be approximated of the cyclic convolutional neural network model;
s402, performing super-parameter optimization on the cyclic convolutional neural network model by using a simulated annealing algorithm;
and S403, when the output of the cyclic convolution neural network model is close to the actual photovoltaic output data to be approximated, the super-parameter optimizing is completed, and the model training is completed.
Further, in step S4, the process of using the multiple linear regression algorithm for the prediction result of each model class includes: the prediction result of each model is set as an independent variable and expressed as x 0 ，x 1 ，... x n The method comprises the steps of carrying out a first treatment on the surface of the n is the total number of prediction results of each model, and the final prediction result is set as dependent variable y, a 0 x 0 +a 1 x 1 ...+a n x n +a n+1 =y, a 0 To a n+1 Calculating a as a parameter using least squares 0 To a n+1 And (5) modeling of the multiple linear regression algorithm is completed.
The invention also provides a distributed photovoltaic prediction device based on simulated annealing and cyclic convolution, which comprises:
and the acquisition module is used for: collecting actual measurement meteorological data and actual photovoltaic output data of each distributed photovoltaic station at fixed interval time length; acquiring predicted meteorological data of a corresponding time period;
and (3) a correction module: establishing a predicted weather correction model, and performing model correction on predicted weather data of each distributed photovoltaic site to obtain corrected predicted weather data;
and a clustering module: clustering the distributed photovoltaic stations by adopting a K-MEANS algorithm based on the coordinates of the distributed photovoltaic stations; based on the corrected predicted meteorological data of each distributed photovoltaic station, clustering the predicted meteorological data by adopting a hierarchical clustering algorithm; clustering the actual measurement meteorological data of each distributed photovoltaic station by adopting a KNN clustering algorithm;
and a prediction module: based on the corrected predicted meteorological data, the actual photovoltaic output data and the clustering result obtained by the clustering module, a model for predicting meteorological-photovoltaic output is built in each type by using a cyclic convolutional neural network algorithm, when model training is carried out, super-parameter optimizing is carried out by using a simulated annealing algorithm, and finally, the final prediction result of each type of model is carried out by using a multiple linear regression algorithm, so that the final prediction result is obtained.
Further, the clustering module includes:
the K-MEANS clustering unit is used for clustering the K-MEANS, calculating the distances between the coordinates of different distributed photovoltaic sites, and gathering the coordinates with the short distances into a class according to the distances between the coordinates;
the hierarchical clustering unit is used for calculating the similarity between the predicted meteorological data sequences of different distributed photovoltaic sites by adopting hierarchical clustering, and gathering the sequences with larger similarity into a class according to the similarity between different data sequences;
and the KNN clustering unit is used for calculating the similarity between actually measured meteorological data sequences of different distributed photovoltaic sites by adopting KNN clustering, and gathering the sequences with larger similarity into one type according to the similarity between different data sequences.
Further, the prediction module includes a linear regression unit, and the process of adopting a multiple linear regression algorithm to the prediction result of each type of model includes: the prediction result of each model is set as an independent variable and expressed as x 0 ，x 1 ，... x n The method comprises the steps of carrying out a first treatment on the surface of the n is the total number of prediction results of each model, and the final prediction result is set as dependent variable y, a 0 x 0 +a 1 x 1 ...+a n x n +a n+1 =y, a 0 To a n+1 Calculating a as a parameter using least squares 0 To a n+1 And (5) modeling of the multiple linear regression algorithm is completed.
Compared with the prior art, the invention has the following beneficial effects:
1. according to the invention, region clustering is carried out through clustering in different modes, photovoltaic output prediction is carried out according to each clustering mode, and finally multiple linear regression is carried out on the prediction result of each clustering mode, so that the photovoltaic output prediction of the whole region is realized, the problem that the prediction result obtained by adopting a template method conversion mode cannot describe regional cloud layer movement and the like is avoided, and the generalization capability of the model is improved by adopting multiple clustering modes for modeling;
2. according to the invention, the weight of the cyclic convolutional neural network is optimized by adopting the simulated annealing algorithm, so that the prediction speed of photovoltaic output prediction is improved, and a certain prediction precision is improved.
Drawings
FIG. 1 is a schematic flow chart of an embodiment of the present invention;
FIG. 2 is a graph comparing predicted results of an embodiment of the present invention.
Detailed Description
It should be noted that, without conflict, the embodiments of the present invention and features of the embodiments may be combined with each other.
For the purpose of making the objects and features of the present invention more comprehensible, embodiments accompanied with figures are described in detail below.
As shown in fig. 1, the method provided by the invention specifically includes:
1. collection of historical data: collecting actual measurement meteorological data and actual photovoltaic output data of each distributed photovoltaic site comprising a distributed photovoltaic area at fixed interval time length; the method comprises the steps of obtaining forecast meteorological data of each distributed photovoltaic station in a corresponding time period; the forecast weather data can be obtained from forecast data of weather centers in all the world, and global grid numerical weather forecast data produced autonomously can also be used;
2. model correction is carried out on the predicted meteorological data of each distributed photovoltaic station, so that errors caused by the predicted meteorological data are reduced; the correction method comprises the following steps: using actual measurement meteorological data of historical data of a distributed photovoltaic station as a variable to be fitted, predicting meteorological data as an input variable, and performing characteristic optimization and super-parameter optimization of a model based on the data to establish a predicted meteorological correction model;
3. clustering is carried out, and the clustering process comprises the following steps:
(1) Based on the coordinates of each distributed photovoltaic site, the distributed photovoltaic sites are clustered by adopting a K-MEANS algorithm, wherein the K-MEANS algorithm is an unsupervised clustering algorithm, and for a given sample set, the samples are divided into K clusters according to the distance between the samples, so that the points in the clusters are tightly connected together as much as possible, and the distance between the clusters is as large as possible.
Specifically, in the step, the distances between the coordinates of different distributed photovoltaic stations are calculated through a K-MEANS algorithm, and the coordinates with the short distances are gathered into one type according to the distances between the coordinates;
(2) Based on the predicted meteorological data of each distributed photovoltaic site, clustering the predicted meteorological data by adopting a hierarchical clustering algorithm, wherein the hierarchical clustering algorithm can be used for defining the similarity among nodes, and then forming the nodes into a hierarchical structure by adopting single-connection hierarchical clustering or full-connection hierarchical clustering;
specifically, in the step, the similarity between predicted meteorological data sequences of different distributed photovoltaic sites is calculated through a hierarchical clustering algorithm, and sequences with larger similarity are gathered into one type according to the similarity between different data sequences;
(3) Based on the actually measured meteorological data of each distributed photovoltaic site, clustering the meteorological data by adopting a KNN clustering algorithm, wherein the KNN clustering algorithm is a clustering algorithm with supervised learning, and classification is also carried out according to distance;
in this step, actually measured meteorological data of different distributed photovoltaic sites are clustered, and the actually measured meteorological data are data sequences, so that sequence similarity is used as the distance between actually measured meteorological data, and clustering is performed according to the similarity between different data sequences through a KNN clustering algorithm.
4. Based on the corrected predicted meteorological data and clustering results of each distributed photovoltaic station in the area, a model establishment of predicted meteorological-photovoltaic output is carried out in each type by using a cyclic convolution neural network algorithm based on a simulated annealing algorithm. And finally, obtaining a final prediction result by adopting a multiple linear regression algorithm.
Firstly, building a cyclic convolution neural network model specifically comprises the following steps: taking the corrected predicted meteorological data as the input of a cyclic convolutional neural network model, and taking the actual photovoltaic output data as a target to be approximated of the cyclic convolutional neural network model; performing super-parameter optimization on the cyclic convolutional neural network model by using a simulated annealing algorithm; when the output of the cyclic convolution neural network model is close to the actual output data of the photovoltaic to be approximated, super-parameter optimizing is completed, and model training is completed; the sufficient proximity may be a set threshold within which the sufficient proximity may be considered.
The specific process of using the simulated annealing algorithm is as follows: initializing parameters of the cyclic convolutional neural network, and randomly generating a group of initial parameters; setting an initial temperature and a termination temperature for a simulated annealing algorithm; determining whether to accept the neighborhood solution according to the objective function value of the neighborhood solution and the objective function value of the current solution; if the objective function value of the neighborhood solution is better, accepting the neighborhood solution; otherwise, accepting the differential solution with a certain probability; the algorithm is terminated when a termination temperature or other termination condition is reached, returning to the currently found optimal parameters.
Second, the multiple linear regression algorithm includes: establishing a relation between a plurality of independent variables x and dependent variables y as a 0 x 0 +a 1 x 1 ...+a n x n +a n+1 =y, calculate a using least squares 0 To a n+1 And (3) completing modeling of multi-source linear regression. In the invention, the prediction result of each model obtained by clustering three clustering methods is combined with independent variable, and x is the sum of the prediction results of the models of each model 0 ，x 1 ，... x n Respectively representing the predicted results of a certain class, wherein n represents the total number of the predicted results of the model of each class, namely the total number of independent variables; the dependent variable y represents the final prediction result, a 0 x 0 +a 1 x 1 ...+a n x n +a n+1 In the expression of =y, a 0 To a n+1 For the parameters of various prediction results, using least square method to calculate all parameters a 0 To a n+1 Modeling of the multiple linear regression algorithm is completed, so that the method is used for predicting the distributed photovoltaic output.
As shown in FIG. 2, the comparison result of the photovoltaic output prediction of different methods of a certain distributed photovoltaic project is shown, the prediction precision of the method (the algorithm of the invention in the figure) provided by the invention is 95.54%, the precision of the original algorithm is 86.06%, and the precision is improved by 9.48%.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, alternatives, and improvements that fall within the spirit and scope of the invention.
Claims (7)
1. A distributed photovoltaic prediction method based on simulated annealing and cyclic convolution, comprising:
s1, collecting actual measurement meteorological data and actual photovoltaic output data of each distributed photovoltaic station at fixed interval time; acquiring predicted meteorological data of a corresponding time period;
s2, establishing a predicted weather correction model, and performing model correction on predicted weather data of each distributed photovoltaic site to obtain corrected predicted weather data;
s3, clustering the distributed photovoltaic stations by adopting a K-MEANS algorithm based on the coordinates of the distributed photovoltaic stations; based on the corrected predicted meteorological data of each distributed photovoltaic station, clustering the predicted meteorological data by adopting a hierarchical clustering algorithm; clustering the actual measurement meteorological data of each distributed photovoltaic station by adopting a KNN clustering algorithm;
s4, based on the corrected predicted meteorological data, the corrected actual photovoltaic output data and the clustering result obtained in the S3, building a model for predicting meteorological-photovoltaic output by using a cyclic convolution neural network algorithm in each type, performing super-parameter optimization by using a simulated annealing algorithm when performing model training, and finally performing final prediction on the prediction result of the model in each type by using a multiple linear regression algorithm to obtain a final prediction result;
in step S4, the model training specifically includes:
s401, taking the corrected predicted meteorological data as input of a cyclic convolutional neural network model, and taking the actual photovoltaic output data as a target to be approximated of the cyclic convolutional neural network model;
s402, performing super-parameter optimization on the cyclic convolutional neural network model by using a simulated annealing algorithm;
s403, when the output of the cyclic convolution neural network model is close to the actual output data of the photovoltaic to be approximated, super-parameter optimizing is completed, and model training is completed;
in step S4, the process of using the multiple linear regression algorithm for the prediction result of each model class includes: the prediction result of each model is set as an independent variable and expressed as x 0 ，x 1 ，... x n The method comprises the steps of carrying out a first treatment on the surface of the n is the total number of prediction results of each model, and the final prediction result is set as dependent variable y, a 0 x 0 +a 1 x 1 ...+a n x n +a n+1 =y, a 0 To a n+1 Calculating a as a parameter using least squares 0 To a n+1 And (5) modeling of the multiple linear regression algorithm is completed.
2. The method for distributed photovoltaic prediction based on simulated annealing and cyclic convolution according to claim 1, wherein in step S2, the establishment of the predicted weather correction model comprises: and taking the actually measured meteorological data of the distributed photovoltaic stations as variables to be fitted, taking the predicted meteorological data as input variables, performing feature optimization and super-parameter optimization based on the data, and establishing a predicted meteorological correction model.
3. The distributed photovoltaic prediction method based on simulated annealing and cyclic convolution according to claim 1, wherein in step S3, the specific step of performing K-MEANS clustering based on coordinates of each distributed photovoltaic site comprises: and (3) adopting K-MEANS clustering to calculate the distances between the coordinates of different distributed photovoltaic sites, and gathering the coordinates with the short distances into one type according to the distances between the coordinates.
4. The distributed photovoltaic prediction method based on simulated annealing and cyclic convolution according to claim 1, wherein in step S3, the specific step of performing hierarchical clustering based on the predicted meteorological data of each distributed photovoltaic site comprises: and (3) hierarchical clustering is adopted, similarity among predicted meteorological data sequences of different distributed photovoltaic sites is calculated, and sequences with larger similarity are clustered into one class according to the similarity among different data sequences.
5. The distributed photovoltaic prediction method based on simulated annealing and cyclic convolution according to claim 1, wherein in step S3, the specific step of performing KNN clustering based on measured meteorological data of each distributed photovoltaic site comprises: and (3) calculating the similarity between the actually measured meteorological data sequences of different distributed photovoltaic sites by adopting KNN clustering, and gathering the sequences with larger similarity into one type according to the similarity between different data sequences.
6. A distributed photovoltaic prediction apparatus based on simulated annealing and cyclic convolution, comprising:
and the acquisition module is used for: collecting actual measurement meteorological data and actual photovoltaic output data of each distributed photovoltaic station at fixed interval time length; acquiring predicted meteorological data of a corresponding time period;
and (3) a correction module: establishing a predicted weather correction model, and performing model correction on predicted weather data of each distributed photovoltaic site to obtain corrected predicted weather data;
and a clustering module: clustering the distributed photovoltaic stations by adopting a K-MEANS algorithm based on the coordinates of the distributed photovoltaic stations; based on the corrected predicted meteorological data of each distributed photovoltaic station, clustering the predicted meteorological data by adopting a hierarchical clustering algorithm; clustering the actual measurement meteorological data of each distributed photovoltaic station by adopting a KNN clustering algorithm;
and a prediction module: based on the corrected predicted meteorological data, the actual photovoltaic output data and the clustering result obtained by the clustering module, a model for predicting meteorological-photovoltaic output is built in each type by using a cyclic convolutional neural network algorithm, when model training is carried out, a simulated annealing algorithm is adopted to carry out super-parameter optimizing, and finally, a final prediction result is carried out on the prediction result of each type of model by adopting a multiple linear regression algorithm, so that a final prediction result is obtained;
the model training comprises: taking the corrected predicted meteorological data as the input of the cyclic convolutional neural network model, and taking the actual photovoltaic output data as a target to be approximated of the cyclic convolutional neural network model; performing super-parameter optimization on the cyclic convolutional neural network model by using a simulated annealing algorithm; when the output of the cyclic convolution neural network model is close to the actual output data of the photovoltaic to be approximated, super-parameter optimizing is completed, and model training is completed;
the prediction module comprises a linear regression unit, and the process of adopting a multiple linear regression algorithm to the prediction result of each type of model comprises the following steps: the prediction result of each model is set as an independent variable and expressed as x 0 ，x 1 ，... x n The method comprises the steps of carrying out a first treatment on the surface of the n is the total number of prediction results of each model, and the final prediction result is set as dependent variable y, a 0 x 0 +a 1 x 1 ...+a n x n +a n+1 =y, a 0 To a n+1 Calculating a as a parameter using least squares 0 To a n+1 And (5) modeling of the multiple linear regression algorithm is completed.
7. The simulated annealing and cyclic convolution based distributed photovoltaic prediction apparatus of claim 6, wherein said clustering module comprises:
the K-MEANS clustering unit is used for clustering the K-MEANS, calculating the distances between the coordinates of different distributed photovoltaic sites, and gathering the coordinates with the short distances into a class according to the distances between the coordinates;
the hierarchical clustering unit is used for calculating the similarity between the predicted meteorological data sequences of different distributed photovoltaic sites by adopting hierarchical clustering, and gathering the sequences with larger similarity into a class according to the similarity between different data sequences;
and the KNN clustering unit is used for calculating the similarity between actually measured meteorological data sequences of different distributed photovoltaic sites by adopting KNN clustering, and gathering the sequences with larger similarity into one type according to the similarity between different data sequences.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311218640.5A CN116956753B (en) | 2023-09-21 | 2023-09-21 | Distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311218640.5A CN116956753B (en) | 2023-09-21 | 2023-09-21 | Distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116956753A CN116956753A (en) | 2023-10-27 |
| CN116956753B true CN116956753B (en) | 2023-12-08 |
Family
ID=88449628
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311218640.5A Active CN116956753B (en) | 2023-09-21 | 2023-09-21 | Distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116956753B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086928A (en) * | 2018-07-27 | 2018-12-25 | 福州大学 | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model |
| CN112633572A (en) * | 2020-12-21 | 2021-04-09 | 国网甘肃省电力公司电力科学研究院 | Day classification dual-model photovoltaic power generation combined prediction method based on prediction meteorological data |
| CN115049163A (en) * | 2022-08-15 | 2022-09-13 | 国能日新科技股份有限公司 | Photovoltaic region prediction method and device based on tabu search and deep stable learning |
| CN116542396A (en) * | 2023-06-15 | 2023-08-04 | 深圳市国电科技通信有限公司 | Distributed photovoltaic output prediction method and device, storage medium and electronic equipment |
| CN116702937A (en) * | 2022-12-14 | 2023-09-05 | 国网湖北省电力有限公司荆门供电公司 | Photovoltaic output day-ahead prediction method based on K-means mean value clustering and BP neural network optimization |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113128793A (en) * | 2021-05-19 | 2021-07-16 | 中国南方电网有限责任公司 | Photovoltaic power combination prediction method and system based on multi-source data fusion |
-
2023
- 2023-09-21 CN CN202311218640.5A patent/CN116956753B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109086928A (en) * | 2018-07-27 | 2018-12-25 | 福州大学 | Photovoltaic plant realtime power prediction technique based on SAGA-FCM-LSSVM model |
| CN112633572A (en) * | 2020-12-21 | 2021-04-09 | 国网甘肃省电力公司电力科学研究院 | Day classification dual-model photovoltaic power generation combined prediction method based on prediction meteorological data |
| CN115049163A (en) * | 2022-08-15 | 2022-09-13 | 国能日新科技股份有限公司 | Photovoltaic region prediction method and device based on tabu search and deep stable learning |
| CN116702937A (en) * | 2022-12-14 | 2023-09-05 | 国网湖北省电力有限公司荆门供电公司 | Photovoltaic output day-ahead prediction method based on K-means mean value clustering and BP neural network optimization |
| CN116542396A (en) * | 2023-06-15 | 2023-08-04 | 深圳市国电科技通信有限公司 | Distributed photovoltaic output prediction method and device, storage medium and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116956753A (en) | 2023-10-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113128793A (en) | Photovoltaic power combination prediction method and system based on multi-source data fusion | |
| CN109002915B (en) | Photovoltaic power station short-term power prediction method based on Kmeans-GRA-Elman model | |
| CN108921339B (en) | Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine | |
| CN110942205B (en) | Short-term photovoltaic power generation power prediction method based on HIMVO-SVM | |
| CN107766986A (en) | Leak integral form echo state network on-line study photovoltaic power Forecasting Methodology | |
| CN112418346B (en) | Numerical weather forecast total radiation system error classification calculation method | |
| CN115374995A (en) | Distributed photovoltaic and small wind power station power prediction method | |
| CN109242136A (en) | A kind of micro-capacitance sensor wind power Chaos-Genetic-BP neural network prediction technique | |
| CN110163444A (en) | A kind of water demand prediction method based on GASA-SVR | |
| CN111242355A (en) | Photovoltaic probability prediction method and system based on Bayesian neural network | |
| CN116167531A (en) | Photovoltaic power generation prediction method based on digital twin | |
| CN115115125A (en) | Photovoltaic power interval probability prediction method based on deep learning fusion model | |
| CN116523142A (en) | Photovoltaic cluster interval prediction method considering space-time characteristic clustering | |
| CN117498400A (en) | Distributed photovoltaic and energy storage data processing method and system | |
| CN105956708A (en) | Grey correlation time sequence based short-term wind speed forecasting method | |
| CN115995810A (en) | Wind power prediction method and system considering weather fluctuation self-adaptive matching | |
| CN112633565B (en) | Photovoltaic power set interval prediction method | |
| CN116956753B (en) | Distributed photovoltaic prediction method and device based on simulated annealing and cyclic convolution | |
| CN115907131B (en) | Method and system for constructing electric heating load prediction model in northern area | |
| CN115544895B (en) | Photovoltaic power station annual output guarantee rate model optimization method | |
| CN116167465A (en) | Solar irradiance prediction method based on multivariate time series ensemble learning | |
| CN114626622A (en) | Photovoltaic power prediction method, system, device and storage medium | |
| CN116128211A (en) | Wind-light-water combined short-term optimization scheduling method based on wind-light uncertainty prediction scene | |
| CN115764855A (en) | Real-time adjustable capacity and available electric quantity prediction method for electric vehicle quick charging station | |
| CN114676866A (en) | Method, device and storage medium for correcting load prediction before day based on error correction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |