CN116187498A - Photovoltaic power generation power prediction method based on frequency domain decomposition - Google Patents
Photovoltaic power generation power prediction method based on frequency domain decomposition Download PDFInfo
- Publication number
- CN116187498A CN116187498A CN202211493788.5A CN202211493788A CN116187498A CN 116187498 A CN116187498 A CN 116187498A CN 202211493788 A CN202211493788 A CN 202211493788A CN 116187498 A CN116187498 A CN 116187498A
- Authority
- CN
- China
- Prior art keywords
- module
- trend
- period
- decomposition
- data
- 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.)
- Withdrawn
Links
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 74
- 238000010248 power generation Methods 0.000 title claims abstract description 50
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000004927 fusion Effects 0.000 claims abstract description 26
- 238000005286 illumination Methods 0.000 claims abstract description 22
- 238000003062 neural network model Methods 0.000 claims abstract description 15
- 238000012545 processing Methods 0.000 claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 7
- 238000004140 cleaning Methods 0.000 claims abstract description 6
- 230000000737 periodic effect Effects 0.000 claims description 9
- 230000009466 transformation Effects 0.000 claims description 5
- 230000008569 process Effects 0.000 claims description 3
- 238000010586 diagram Methods 0.000 description 5
- 238000007619 statistical method Methods 0.000 description 5
- 238000000053 physical method Methods 0.000 description 4
- 238000013135 deep learning Methods 0.000 description 3
- 230000005855 radiation Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000004146 energy storage Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Educational Administration (AREA)
- Health & Medical Sciences (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a photovoltaic power generation power prediction method based on frequency domain decomposition, which relates to the technical field of photovoltaic power generation, and comprises the following steps: collecting historical data of a photovoltaic power plant, constructing a multi-dimensional time sequence data sample set, cleaning the multi-dimensional time sequence data sample set by using a quarter-point intra-distance algorithm, and separating the multi-dimensional time sequence data sample set into a sunlight intensity data set and a power sequence data set; constructing a neural network model based on the frequency domain decomposition, the neural network model comprising: the system comprises an illumination prediction module, a power prediction module and a fusion module; using a sunlight intensity data set and a power sequence data set as a training neural network model, wherein the sunlight intensity data set is input into a light prediction module, the power sequence data set is input into a power prediction module, and output results of the light prediction module and the power prediction module are input into a fusion module together for processing; and the data is processed by the fusion module and then the result is output as the predicted photovoltaic power generation power of the photovoltaic power plant.
Description
Technical Field
The invention relates to the technical field of photovoltaic power generation, in particular to a photovoltaic power generation power prediction method based on frequency domain decomposition.
Background
Photovoltaic power generation is a power generation method for receiving solar radiation by using a solar panel and converting the solar radiation into electric energy, and because the solar radiation is required to obtain energy, the power generation amount is influenced by factors such as quarterly, cloudy, day and night, and the like, and the output power of a photovoltaic power generation system shows randomness and intermittence. In addition, because of the difficulty in energy storage, when most of the electric energy in the power grid comes from photovoltaic power generation, the stability of the power system is affected, and the development of the photovoltaic industry is limited. Accordingly, predicting photovoltaic power generation power is a concern.
The prediction of photovoltaic power generation data mainly has two main ideas: statistical methods and physical methods are used. The statistical method is to perform statistical analysis on historical data, find out the internal rule of the historical data and predict the internal rule; the physical method is to take the meteorological parameters as input values and solve the predicted power through a physical model. Both have respective limitations: the statistical method only emphasizes the regularity of the data, and can not respond in time to sudden weather changes; however, the physical method cannot give consideration to the period and the rule, and the prediction result is often inaccurate.
With the rapid development of artificial intelligence methods, deep learning-based models have been developed and applied to many fields. Deep learning is a new branch of machine learning methods, and a method for predicting photovoltaic power generation power by using a model based on deep learning, such as RNN or LSTM, is paid attention to. Compared with the traditional physical and statistical methods, the deep learning model can mine deep features from the photovoltaic power sequences and obtain more accurate prediction results. However, the RNN network has defects that it cannot process long-time/power sequences, and LSTM can only predict the history of power, and cannot consider sudden weather conditions.
Disclosure of Invention
The invention aims at: the photovoltaic power generation power prediction method capable of considering historical data and weather conditions is provided.
The technical scheme of the invention is as follows: the utility model provides a photovoltaic power generation power prediction method based on frequency domain decomposition, which comprises the following steps:
s1, collecting historical data of a photovoltaic power plant to construct a multi-dimensional time sequence data sample set, cleaning the multi-dimensional time sequence data sample set by using a quarter-point intra-distance algorithm, and separating the multi-dimensional time sequence data sample set into a sunlight intensity data set and a power sequence data set;
s2, constructing a neural network model based on frequency domain decomposition, wherein the neural network model comprises: the system comprises an illumination prediction module, a power prediction module and a fusion module; using a sunlight intensity data set and a power sequence data set as a training neural network model, wherein the sunlight intensity data set is input into a light prediction module, the power sequence data set is input into a power prediction module, and output results of the light prediction module and the power prediction module are input into a fusion module together for processing;
s3, outputting a result as predicted photovoltaic power generation power of the photovoltaic power plant after the data are processed by the fusion module;
the illumination prediction module and the power prediction module in the step S2 have the same structure, namely a two-layer encoder and a one-layer decoder, data are input to the decoder after two-round encoding, and an initialization sequence input to the decoder is processed into a final result and output to the fusion module.
In any of the foregoing solutions, further, the encoder includes: the device comprises a frequency learning module, a period-trend decomposition module and a forward propagation module, wherein residual errors among the modules are connected, and data are input into an encoder and then are processed by the modules in sequence and then output.
In any of the above solutions, further, the encoder processing procedure includes:
the data firstly enter a frequency learning module, time sequence data is obtained through processing, the time sequence data and unprocessed data are connected in a residual way, the time sequence data and unprocessed data are sent to a period-trend decomposition module together, the period-trend decomposition module decomposes the sent data to obtain a period component and a trend component on a time sequence, the period component is output after the trend component is discarded, the period component is also connected with a signal input into the module in a residual way, and the signal is sent to a forward propagation module; and after the data output by the forward propagation module is fused with the data input by the forward propagation module, the data enters a period-trend decomposition module, and after the trend component is discarded, the period component is finally output as the result of the encoder.
In any of the above technical solutions, further, the two-layer encoder is divided into a first-layer encoder and a second-layer encoder, and the two encoders have identical structures, and the result processed by the two-layer encoder is decomposed into a value and a key input decoder.
In any of the foregoing solutions, further, the decoder includes: the device comprises a frequency learning module, a period-trend decomposition module, a frequency domain attention module and a forward propagation module, wherein residual errors among the modules are connected, and data are input into a decoder and then are processed by the modules in sequence and then are output.
In any of the above solutions, further, the decoder processing includes:
firstly, initializing a periodic component and a trend component, inputting the initialized periodic component into a frequency learning module, making residual links of results before and after transformation, then sending the connection results into a period-trend decomposition module, making residual links of the trend component output by the period-trend decomposition module and the initialized trend, and marking the result as A; the periodic component output by the periodic-trend decomposition module is sent to the frequency domain attention module together with the sequence output by the encoder;
the frequency domain attention module links the processed result with the output result of the period-trend decomposition module as residual error and sends the residual error to the next period-trend decomposition module; the period-trend decomposition module decomposes the period-trend into period quantity and trend quantity, the period quantity enters the forward propagation module, the trend quantity is connected with the result A in a residual way, and the result is marked as B;
the input and output results of the forward propagation module are transmitted to the next period-trend decomposition module after being subjected to residual error connection, the period-trend decomposition module decomposes the result into period quantity and trend quantity, the trend quantity and B are subjected to residual error connection, and the result is marked as C;
and finally, the period quantity obtained by the decomposition of the period-trend decomposition module is fused with the trend quantity C, and the result is output to the fusion module.
In any of the above technical solutions, further, the gaussian window processing specifically includes:
where n is the number of discrete sequence data points and M is the width of the window.
In any of the above solutions, further, the weighted average formula is:
wherein L (n) is the output value, i.e. the value put into the fully connected layer, and L (n) is the original sequence data point.
The beneficial effects of the invention are as follows:
aiming at the periodicity and the instability of the photovoltaic power generation power, two identical network modules are designed to respectively predict sunlight and power, and then the prediction results are fused and integrated to obtain the photovoltaic power generation power prediction result;
the frequency domain attention module is introduced into the decoder to convert the time domain data into the frequency domain, and the frequency domain of the time domain data with obvious periodicity such as illumination information and photovoltaic power generation power often contains more visual information, so that the result can be predicted more accurately by the model for learning the frequency domain information.
Drawings
The advantages of the foregoing and additional aspects of the invention will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a schematic flow chart of a method of photovoltaic power generation power prediction based on frequency domain decomposition according to one embodiment of the invention;
FIG. 2 is a block diagram of an illumination prediction module or power prediction module of a photovoltaic power generation power prediction method based on frequency domain decomposition according to one embodiment of the present invention;
FIG. 3 is a block diagram of an encoder of a photovoltaic power generation power prediction method based on frequency domain decomposition according to one embodiment of the present invention;
FIG. 4 is a block diagram of a frequency learning module of an encoder of a photovoltaic power generation power prediction method based on frequency domain decomposition according to an embodiment of the present invention;
FIG. 5 is a decoder block diagram of a photovoltaic power generation power prediction method based on frequency domain decomposition according to one embodiment of the present invention;
FIG. 6 is a frequency domain attention module block diagram of a decoder of a photovoltaic power generation power prediction method based on frequency domain decomposition according to one embodiment of the present invention;
FIG. 7 is a flow chart of a fusion module process of a photovoltaic power generation power prediction method based on frequency domain decomposition according to one embodiment of the present invention;
fig. 8 is a graph of predicted power generation versus actual power generation for a photovoltaic power generation power prediction method based on frequency domain decomposition according to one embodiment of the present invention.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings and appended detailed description. It should be noted that, without conflict, embodiments of the present invention and features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those described herein, and the scope of the invention is therefore not limited to the specific embodiments disclosed below.
As shown in fig. 1, the present embodiment provides a method for predicting photovoltaic power generation power based on frequency domain decomposition, which includes:
s1, collecting historical data of a photovoltaic power plant to construct a multi-dimensional time sequence data sample set, cleaning the multi-dimensional time sequence data sample set by using a quarter-point intra-distance algorithm, and separating the multi-dimensional time sequence data sample set into a sunlight intensity data set and a power sequence data set.
Specifically, the collected historical data of the photovoltaic power plant includes time, illumination amplitude, temperature, air pressure and actual power, the data at the same moment is stored as a multi-dimensional time sequence data sample to be a local file, in this embodiment, 5000 multi-dimensional time sequence data samples are accumulated and acquired at a frequency of 10 minutes and one sample, and photovoltaic power generation power data affected by various different scene factors is covered, so that the photovoltaic power generation system is more comprehensive.
And detecting abnormal data by using a quarter-point intra-distance algorithm, reconstructing and standardizing the data after cleaning the data set, and separating the data into a sunlight intensity data set and a power sequence data set according to types.
S2, constructing a neural network model based on frequency domain decomposition, wherein the neural network model comprises: the system comprises an illumination prediction module, a power prediction module and a fusion module; and using the sunlight intensity data set and the power sequence data set as training sets to train the neural network model, wherein the sunlight intensity data set is input into the illumination prediction module, the power sequence data set is input into the power prediction module, and output results of the illumination prediction module and the power prediction module are input into the fusion module together for processing.
As shown in fig. 2, the illumination prediction module and the power prediction module have the same structure, namely a two-layer encoder and a one-layer decoder, data are input to the decoder after two-round encoding, and an initialization sequence in the decoder is processed into a final result and output to the fusion module.
As shown in fig. 3, the encoder includes: the device comprises a frequency learning module, a period-trend decomposition module and a forward propagation module, wherein residual connection among the modules prevents gradient explosion or disappearance, and data are input into an encoder and then are processed by the modules in sequence and then are output.
As shown in fig. 4, after the data enters the frequency learning module, the data is subjected to linear transformation, the result is denoted as Q, the Q is subjected to discrete fourier transformation, the result is denoted as Q, and the result obtained after downsampling the Q is denoted as Q
Then, the result obtained by performing linear operation on the parameter matrix R is defined as +.>
Then pair->
After the frequency is completed, performing inverse Fourier transform to obtain processed time sequence data y and outputting the processed time sequence data y.
And then, carrying out residual connection on the data processed by the frequency learning module and the unprocessed data, and sending the data into a period-trend decomposition module, wherein the period-trend decomposition module decomposes the sent data to obtain a period component and a trend component on a time sequence, outputs the period component after discarding the trend component, and also carries out residual connection with a signal input into the module and sends the signal to a forward propagation module.
The forward propagation module is a fully connected network, and the parameters are used as training parameters. The data output by the forward propagation module is fused with the data input by the forward propagation module, then enters a period-trend decomposition module, discards the trend component, finally inputs the period component as a result to a second layer encoder, the second layer encoder has the same structure as the first layer encoder, and the result processed by the two layers of encoders is decomposed into a value and a key input decoder.
As shown in fig. 5, the decoder includes: a frequency learning module, a period-trend decomposition module, a frequency domain attention module and a forward propagation module.
In the decoder, firstly, initializing a periodic component and a trend component, inputting the initialized periodic component into a frequency learning module, making residual links of results before and after transformation, then sending the connection results into a period-trend decomposition module, making residual links of the trend component output by the period-trend decomposition module and the initialized trend, and marking the result as A; the periodic component output by the period-trend decomposition module is sent to the frequency domain attention module along with the sequence output by the encoder.
The frequency domain attention module links the processed result with the output result of the period-trend decomposition module as residual error and sends the residual error to the next period-trend decomposition module; the period-trend decomposition module decomposes the period-trend into period quantity and trend quantity, the period quantity enters the forward propagation module, the trend quantity is connected with the result A in a residual way, and the result is marked as B.
The forward propagation module adopts a full connection layer as in the encoder, the input and output results are connected in a residual way and then sent to the next period-trend decomposition module, the period-trend decomposition module decomposes the result into period quantity and trend quantity, the trend quantity is connected in a residual way with the B, and the result is marked as C. And finally, the period quantity obtained by the decomposition of the period-trend decomposition module is fused with the trend quantity C, and the result is output to the fusion module.
As shown in fig. 6, the value v and the key k output by the encoder are input first in the frequency domain attention module, specifically, the input of the encoder to the frequency domain attention module of the decoder is divided into two paths, and the value v and the key k are input respectively; and secondly, marking the result of the network layer in front of the decoder as q input, performing linear operation on q and the key k, performing linear operation on the result and the value v to obtain an attention result, and performing frequency complementation and inverse Fourier transform on the attention result to obtain time sequence data, wherein the result is transmitted to the next layer for learning.
And S3, outputting the result as the predicted photovoltaic power generation power of the photovoltaic power plant after the data are processed by the fusion module.
As shown in fig. 7, the data output by the illumination prediction module and the power prediction module are weighted and averaged after being processed by gaussian windows with step length of 1 and beta value of 1.2, the results output by the illumination prediction module and the power prediction module are input into a forward propagation module in the fusion module together, and the result data is finally output after being processed by a full connection layer.
Wherein the gaussian window formula is:
where n is the number of discrete sequence data points and M is the width of the window.
The weighted average formula is:
wherein L (n) is the output value, i.e. the value put into the fully connected layer, and L (n) is the original sequence data point.
In another embodiment of the invention, the photovoltaic power generation power trend of a village is predicted, as shown in fig. 8, the result predicted by the photovoltaic power generation power prediction method provided by the invention is well fitted with the actual power generation result, and has higher accuracy.
In summary, the invention provides a photovoltaic power generation power prediction method based on frequency domain decomposition, which comprises the following steps:
s1, collecting historical data of a photovoltaic power plant to construct a multi-dimensional time sequence data sample set, cleaning the multi-dimensional time sequence data sample set by using a quarter-point intra-distance algorithm, and separating the multi-dimensional time sequence data sample set into a sunlight intensity data set and a power sequence data set.
S2, constructing a neural network model based on frequency domain decomposition, wherein the neural network model comprises: the system comprises an illumination prediction module, a power prediction module and a fusion module; and using the sunlight intensity data set and the power sequence data set as training sets to train the neural network model, wherein the sunlight intensity data set is input into the illumination prediction module, the power sequence data set is input into the power prediction module, and output results of the illumination prediction module and the power prediction module are input into the fusion module together for processing.
And S3, outputting a result as the predicted photovoltaic power generation power of the photovoltaic power plant after the data are processed by the fusion module.
The illumination prediction module and the power prediction module in the step S2 have the same structure, namely a two-layer encoder and a one-layer decoder, data are input to the decoder after two-round encoding, and an initialization sequence input to the decoder is processed into a final result and output to the fusion module.
The steps in the invention can be sequentially adjusted, combined and deleted according to actual requirements.
Although the invention has been disclosed in detail with reference to the accompanying drawings, it is to be understood that such description is merely illustrative and is not intended to limit the application of the invention. The scope of the invention is defined by the appended claims and may include various modifications, alterations and equivalents of the invention without departing from the scope and spirit of the invention.
Claims (9)
1. A photovoltaic power generation power prediction method based on frequency domain decomposition, the method comprising:
s1, collecting historical data of a photovoltaic power plant to construct a multi-dimensional time sequence data sample set, cleaning the multi-dimensional time sequence data sample set by using a quarter-point intra-distance algorithm, and separating the multi-dimensional time sequence data sample set into a sunlight intensity data set and a power sequence data set;
s2, constructing a neural network model based on frequency domain decomposition, wherein the neural network model comprises the following components: the system comprises an illumination prediction module, a power prediction module and a fusion module; training the neural network model by using the sunlight intensity data set and the power sequence data set as training sets, wherein the sunlight intensity data set is input into the illumination prediction module, the power sequence data set is input into the power prediction module, and output results of the illumination prediction module and the power prediction module are input into the fusion module together for processing;
s3, outputting a result as predicted photovoltaic power generation power of the photovoltaic power plant after the data are processed by the fusion module;
the illumination prediction module and the power prediction module in step S2 have the same structure, and are two layers of encoders and one layer of decoder, data are input to the decoder after two rounds of encoding, and an initialization sequence in the decoder is additionally input to be processed into a final result and output to the fusion module.
2. The frequency domain decomposition based photovoltaic power generation power prediction method according to claim 1, wherein said encoder comprises: the device comprises a frequency learning module, a period-trend decomposition module and a forward propagation module, wherein residual errors among the modules are connected, and data are input into an encoder and then are processed by the modules in sequence and then output.
3. The frequency domain decomposition based photovoltaic power generation power prediction method according to claim 2, wherein said encoder process comprises:
the data firstly enter a frequency learning module and are processed to obtain time sequence data, the time sequence data and unprocessed data are connected in a residual way and are sent to a period-trend decomposition module, the period-trend decomposition module decomposes the sent data to obtain a period component and a trend component on a time sequence, the period component is output after the trend component is discarded, the period component is also connected with a signal input into the module in a residual way, and the signal is sent to a forward propagation module; and after the data output by the forward propagation module is fused with the data input by the forward propagation module, the data enters a period-trend decomposition module, and after the trend component is discarded, the period component is finally output as the result of the encoder.
4. The method for predicting the power of photovoltaic power generation based on frequency domain decomposition according to claim 1, wherein the two-layer encoder is divided into a first-layer encoder and a second-layer encoder, which are identical in structure, and the result processed by the two-layer encoder is decomposed into a value and a key to be input into the decoder.
5. The frequency domain decomposition based photovoltaic power generation power prediction method according to claim 1, wherein said decoder comprises: the device comprises a frequency learning module, a period-trend decomposition module, a frequency domain attention module and a forward propagation module, wherein residual errors among the modules are connected, and data are input into a decoder and then are processed by the modules in sequence and then are output.
6. The frequency domain decomposition based photovoltaic power generation power prediction method according to claim 5, wherein said decoder processing comprises:
firstly, initializing a periodic component and a trend component, inputting the initialized periodic component into the frequency learning module, making residual error links between the results before and after transformation, and then sending the connection results into the period-trend decomposition module, making residual error links between the trend component output by the period-trend decomposition module and the initialized trend, and marking the result as A; the periodic component output by the periodic-trend decomposition module is sent to the frequency domain attention module together with the sequence output by the encoder;
the frequency domain attention module links the processed result with the output result of the period-trend decomposition module as residual error and sends the residual error to the next period-trend decomposition module; the period-trend decomposition module decomposes the period-trend into period quantity and trend quantity, the period quantity enters the forward propagation module, the trend quantity is connected with the result A in a residual way, and the result is marked as B;
the input and output results of the forward propagation module are transmitted to the next period-trend decomposition module after being subjected to residual error connection, the period-trend decomposition module decomposes the result into period quantity and trend quantity, the trend quantity and B are subjected to residual error connection, and the result is marked as C;
and finally, the period quantity obtained by the decomposition of the period-trend decomposition module is fused with the trend quantity C, and the result is output to the fusion module.
7. The method for predicting the power of photovoltaic power generation based on frequency domain decomposition according to claim 1, wherein the results obtained by the illumination prediction module and the power prediction module are input into the fusion module, the illumination prediction data and the power prediction data are respectively processed by a gaussian window and then weighted average, the results are input into a forward propagation module positioned in the fusion module together, and the result data are finally output after being processed by a full-connection layer.
8. The method for predicting the power of a photovoltaic power generation based on frequency domain decomposition according to claim 7, wherein said gaussian window processing is specifically:
where n is the number of discrete sequence data points and M is the width of the window.
9. The method for predicting the power of a photovoltaic power generation based on frequency domain decomposition of claim 7, wherein said weighted average formula is:
wherein L (n) is the output value, i.e. the value put into the fully connected layer, and L (n) is the original sequence data point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211493788.5A CN116187498A (en) | 2022-11-25 | 2022-11-25 | Photovoltaic power generation power prediction method based on frequency domain decomposition |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211493788.5A CN116187498A (en) | 2022-11-25 | 2022-11-25 | Photovoltaic power generation power prediction method based on frequency domain decomposition |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116187498A true CN116187498A (en) | 2023-05-30 |
Family
ID=86431506
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211493788.5A Withdrawn CN116187498A (en) | 2022-11-25 | 2022-11-25 | Photovoltaic power generation power prediction method based on frequency domain decomposition |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116187498A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116415744A (en) * | 2023-06-12 | 2023-07-11 | 深圳大学 | Power prediction method and device based on deep learning and storage medium |
| CN117131790A (en) * | 2023-10-27 | 2023-11-28 | 西南石油大学 | Photovoltaic module cleaning period prediction method under probability coding and decoding framework |
-
2022
- 2022-11-25 CN CN202211493788.5A patent/CN116187498A/en not_active Withdrawn
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116415744A (en) * | 2023-06-12 | 2023-07-11 | 深圳大学 | Power prediction method and device based on deep learning and storage medium |
| CN116415744B (en) * | 2023-06-12 | 2023-09-19 | 深圳大学 | Power prediction method and device based on deep learning and storage medium |
| CN117131790A (en) * | 2023-10-27 | 2023-11-28 | 西南石油大学 | Photovoltaic module cleaning period prediction method under probability coding and decoding framework |
| CN117131790B (en) * | 2023-10-27 | 2024-01-23 | 西南石油大学 | Photovoltaic module cleaning period prediction method under probability coding and decoding framework |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116187498A (en) | Photovoltaic power generation power prediction method based on frequency domain decomposition | |
| CN109214575B (en) | Ultrashort-term wind power prediction method based on small-wavelength short-term memory network | |
| CN109635928A (en) | A kind of voltage sag reason recognition methods based on deep learning Model Fusion | |
| CN112990553B (en) | Wind power ultra-short-term power prediction method using self-attention mechanism and bilinear fusion | |
| CN111242377A (en) | Short-term wind speed prediction method integrating deep learning and data denoising | |
| CN113688869B (en) | Photovoltaic data missing reconstruction method based on generation countermeasure network | |
| CN116307291B (en) | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition | |
| CN116911419A (en) | Long time sequence prediction method based on trend correlation feature learning | |
| Li et al. | Temporal attention based tcn-bigru model for energy time series forecasting | |
| CN117190078B (en) | Abnormality detection method and system for monitoring data of hydrogen transportation pipe network | |
| CN112862209B (en) | Industrial equipment monitoring data prediction method | |
| CN115619999A (en) | Real-time monitoring method and device for power equipment, electronic equipment and readable medium | |
| CN112613494B (en) | Power line monitoring abnormality identification method and system based on deep countermeasure network | |
| CN113361782A (en) | Photovoltaic power generation power short-term rolling prediction method based on improved MKPLS | |
| CN117374913A (en) | Wave energy prediction method and device based on STL decomposition and multilayer seq2seq model | |
| CN115713044B (en) | Method and device for analyzing residual life of electromechanical equipment under multi-condition switching | |
| CN117113054A (en) | Multi-element time sequence prediction method based on graph neural network and transducer | |
| CN116070768A (en) | Short-term wind power prediction method based on data reconstruction and TCN-BiLSTM | |
| CN115661532A (en) | Operating state evaluation method and system of switch equipment | |
| CN116008747A (en) | Yogi-mLSTM cable partial discharge identification method and diagnosis system based on wavelet threshold denoising | |
| CN112257938B (en) | Photovoltaic power generation power prediction method and device | |
| Kang et al. | Research on forecasting method for effluent ammonia nitrogen concentration based on GRA-TCN | |
| Jin et al. | A hybrid prediction framework based on deep learning for wind power | |
| CN118472943B (en) | Short-term wind power prediction method based on multi-feature fusion period enhancement | |
| Xia et al. | Research on Solar Radiation Estimation based on Singular Spectrum Analysis-Deep Belief Network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20230530 |