CN117277304A - Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time - Google Patents

Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time Download PDF

Info

Publication number
CN117277304A
CN117277304A CN202311255468.0A CN202311255468A CN117277304A CN 117277304 A CN117277304 A CN 117277304A CN 202311255468 A CN202311255468 A CN 202311255468A CN 117277304 A CN117277304 A CN 117277304A
Authority
CN
China
Prior art keywords
power generation
photovoltaic power
short
sunrise
prediction
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
Application number
CN202311255468.0A
Other languages
Chinese (zh)
Inventor
孙钰淑
任鑫
王一妹
周利
祝金涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huaneng Clean Energy Research Institute
Huaneng Lancang River Hydropower Co Ltd
Original Assignee
Huaneng Clean Energy Research Institute
Huaneng Lancang River Hydropower Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Huaneng Clean Energy Research Institute, Huaneng Lancang River Hydropower Co Ltd filed Critical Huaneng Clean Energy Research Institute
Priority to CN202311255468.0A priority Critical patent/CN117277304A/en
Publication of CN117277304A publication Critical patent/CN117277304A/en
Pending legal-status Critical Current

Links

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
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/217—Validation; Performance evaluation; Active pattern learning techniques
    • 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
    • G06N3/045—Combinations of networks
    • 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
    • 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
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/381—Dispersed generators
    • 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
    • H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
    • H02J2300/20—The dispersed energy generation being of renewable origin
    • H02J2300/22—The renewable source being solar energy

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Software Systems (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Mathematical Physics (AREA)
  • Business, Economics & Management (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Computing Systems (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Power Engineering (AREA)
  • Economics (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time. Belongs to the technical field of photovoltaic power generation, and comprises the following steps: collecting original data, cleaning the original data to obtain a data set, and dividing the data set into a training set and a testing set; constructing a photovoltaic power generation ultra-short-term power prediction characteristic, and establishing a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and an Attention mechanism; and training by using the training set to obtain a photovoltaic power generation ultra-short term power prediction result, and verifying the photovoltaic power generation ultra-short term power prediction result by using the testing set. The method improves the accuracy and the practicability of prediction, considers the dynamic changes of sunrise and sunset time more accurately, and can adapt to the changes of different areas and seasons, thereby providing better support for the operation and management of the photovoltaic power generation system.

Description

Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time
Technical Field
The invention belongs to the field of photovoltaic power generation, and particularly relates to a photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time.
Background
Photovoltaic power generation is a clean energy source, but its power generation is affected by various factors including sunrise time and sunset time of the sun. The traditional photovoltaic power generation power prediction method mainly depends on meteorological data such as solar radiation, temperature, cloud coverage and the like to estimate the power generation capacity of the photovoltaic battery pack. However, these methods have limitations in considering dynamic changes in sunrise time and sunset time, because conventional photovoltaic power generation power prediction methods often fail to capture the exact moment of sunlight. This deficiency can lead to inaccuracy in the power predictions, particularly at and around sunrise times.
Existing photovoltaic power generation power prediction methods which attempt to consider dynamic changes of sunrise and sunset time generally only perform simple statistical analysis based on historical data, or set a power prediction value to zero when a predicted period is less than sunrise time or greater than sunset time, and still have limitations. The Chinese patent publication No. CN103390902A published 11/13 in 2013 discloses a photovoltaic power station ultra-short term power prediction method based on a least square method, and provides that when the predicted period is smaller than sunrise time or larger than sunset time, the power prediction value is set to zero, and the photovoltaic battery pack still has certain power generation capacity before sunrise or after sunset, and the power prediction value is set to zero, so that energy waste can be caused. In addition, the prior art has the defect of being not suitable for the change of different areas and seasons and being not flexible enough.
Therefore, the current photovoltaic power generation power prediction method needs to solve the problem that the dynamic changes of sunrise and sunset time cannot be accurately considered.
Disclosure of Invention
The invention aims to solve the problem that the existing photovoltaic power generation power prediction method cannot accurately consider dynamic changes of sunrise and sunset time, and provides a photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time, which provide better support for operation and management of a photovoltaic power generation system.
In order to achieve the above purpose, the invention adopts the following technical scheme:
a photovoltaic power generation ultra-short-term power prediction method considering sunrise and sunset time comprises the following steps:
collecting original data, wherein the original data comprises actual measurement photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
preprocessing the original data to obtain a training set and a testing set;
constructing a photovoltaic power generation ultra-short-term power prediction characteristic by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic power generation ultra-short-term power prediction characteristic comprises a sunrise and sunset characteristic code, and the sunrise and sunset characteristic code judges whether the prediction time is close to sunrise and sunset;
establishing a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts photovoltaic power generation power of four hours in the future at a prediction time point;
training a photovoltaic power generation ultra-short term power prediction model by using a training set to obtain a photovoltaic power generation ultra-short term power prediction result;
and verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
Further, the photovoltaic ultra-short term power prediction features also include original features, statistical features and time codes;
the original features are specifically actual measurement photovoltaic power generation power of a predicted day in a sliding time window, the statistical features are specifically features extracted from the original features by using a statistical method, and the statistical method comprises an average value, a median and a standard deviation;
the time code comprises minutes, hours, weeks, months and years corresponding to the predicted time point;
the sunrise and sunset characteristic codes use cosine codes or time distances, the cosine codes specifically code each day in one year into a value from 0 to 1, a cosine function is used for obtaining periodic codes, and sunrise and sunset time shows periodic change in one year; the time distance is the time difference between the current time and the sunrise and sunset time.
Further, the photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and Attention mechanism established by utilizing the photovoltaic power generation ultra-short-term power prediction characteristics specifically comprises the following steps: firstly, inputting original data in a historical time window into a BiLSTM layer, capturing time sequence information and sequence relation of the original data in the historical time window by the BiLSTM layer, and then focusing Attention on the time sequence information and sequence relation related to prediction by a photovoltaic power generation ultra-short-term power prediction model through an Attention mechanism, and learning fluctuation and trend of photovoltaic power generation in different time periods.
Further, the Attention mechanism identifies key features and secondary features from data within a preselected historical time window and extracts key features and assigns weights to the key features and secondary features, the key features being assigned a higher weight than the secondary features.
Further, the photovoltaic power generation ultra-short term power prediction result of the photovoltaic power generation ultra-short term power prediction model is evaluated after the photovoltaic power generation ultra-short term power prediction result is verified by the verification set, the evaluation comprises calculation and analysis of performance indexes, the performance indexes comprise average absolute errors, and the formula is as follows:
wherein MAE is the average absolute error, m is the number of samples, y i To be a true value of the value,is a predicted value of the model;
the value range of the MAE is [0, + ], when the photovoltaic power generation ultra-short-term power prediction model predicts completely accurately, the calculated MAE is 0, which means that the accuracy of the prediction result of the photovoltaic power generation ultra-short-term power prediction model reaches 100%, and the photovoltaic power generation ultra-short-term power prediction model is a perfect model.
Further, the month accuracy and the day accuracy of the photovoltaic power generation ultra-short term power prediction result are checked by using a photovoltaic power station ultra-short term power prediction check index accuracy calculation formula, the day accuracy of the photovoltaic power generation ultra-short term power prediction method in the fourth hour is greater than or equal to 90%, and when the day accuracy of the photovoltaic power generation ultra-short term power prediction method in the fourth hour is less than 90%, the following formula is used for checking:
wherein: p (P) Mi Is the real time of iPower of the power plane, P Pi For the i-th time predicted value of ultra-short-term power prediction, cap is the available capacity of a photovoltaic power station, n is the number of samples in the photovoltaic power generation period, and P N The installed capacity of the photovoltaic power station;
the month accuracy of the photovoltaic power generation ultra-short period power prediction method is an arithmetic average value of the day accuracy of the photovoltaic power generation ultra-short period power prediction method.
Further, the preprocessing specifically includes cleaning original data to obtain a data set, and dividing the data set into a training set and a testing set; the cleaning is specifically to delete or repair any missing, abnormal or inaccurate data points in the original data.
The ultra-short-term power prediction system for photovoltaic power generation considering sunrise and sunset time comprises an acquisition data module, a data processing module and a data processing module, wherein the acquisition data module is used for acquiring original data, and the original data comprise actual measurement photovoltaic power generation of a photovoltaic station for predicting the sunrise and sunset time of a day and the predicted sunrise and sunset time of the day;
the preprocessing module is used for preprocessing the original data to obtain a training set and a testing set;
the characteristic construction module is used for constructing photovoltaic power generation ultra-short-term power prediction characteristics by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic ultra-short-term power prediction characteristics comprise sunrise and sunset characteristic codes, and the sunrise and sunset characteristic codes judge whether the prediction time is close to sunrise and sunset;
the model building module is used for building a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing the photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts photovoltaic power generation power of four hours in the future at a prediction time point;
the model training module is used for training a photovoltaic power generation ultra-short-term power prediction model by using the training set to obtain a photovoltaic power generation ultra-short-term power prediction result;
and the result verification module is used for verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, the processor implementing a photovoltaic power generation ultrashort-term power prediction method that takes into account sunrise and sunset time when executing the computer program. The photovoltaic power generation ultra-short-term power prediction method considering sunrise and sunset time comprises the following steps:
collecting original data, wherein the original data comprises actual measurement photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
preprocessing the original data to obtain a training set and a testing set;
constructing a photovoltaic power generation ultra-short-term power prediction characteristic by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic power generation ultra-short-term power prediction characteristic comprises a sunrise and sunset characteristic code, and the sunrise and sunset characteristic code judges whether the prediction time is close to sunrise and sunset;
establishing a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts photovoltaic power generation power of four hours in the future at a prediction time point;
training a photovoltaic power generation ultra-short term power prediction model by using a training set to obtain a photovoltaic power generation ultra-short term power prediction result;
and verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
A computer readable storage medium storing a computer program which when executed by a processor implements a photovoltaic power generation ultra-short term power prediction method that takes into account sunrise and sunset time. The photovoltaic power generation ultra-short-term power prediction method considering sunrise and sunset time comprises the following steps:
collecting original data, wherein the original data comprises actual measurement photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
preprocessing the original data to obtain a training set and a testing set;
constructing a photovoltaic power generation ultra-short-term power prediction characteristic by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic power generation ultra-short-term power prediction characteristic comprises a sunrise and sunset characteristic code, and the sunrise and sunset characteristic code judges whether the prediction time is close to sunrise and sunset;
establishing a photovoltaic power generation ultra-short-term power prediction model considering sunrise and sunset time by utilizing the photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts photovoltaic power generation power of four hours in the future at a prediction time point;
training a photovoltaic power generation ultra-short term power prediction model by using a training set to obtain a photovoltaic power generation ultra-short term power prediction result;
and verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
Compared with the prior art, the invention has the following beneficial technical effects:
according to the photovoltaic power generation ultra-short-term power prediction method considering sunrise and sunset time, accuracy and practicality of prediction are improved by using sunrise and sunset time codes, dynamic changes of sunrise and sunset time are considered more accurately, a bidirectional long and short time memory network (BiLSTM) is introduced, and the model is allowed to capture dependency relations and modes in time sequence data more comprehensively compared with a traditional unidirectional LSTM model. BiLSTM considers both past and future information, thereby significantly improving the accuracy of photovoltaic ultra-short term power predictions. By adopting the Attention mechanism, key information can be dynamically extracted from data in a plurality of different history time windows. This mechanism helps to reduce co-linearity problems in the data, improving the stability and robustness of the model. By automatically focusing on the information most useful for prediction, the accuracy of photovoltaic power prediction is significantly improved. Meanwhile, the sensitivity of the model to irrelevant variables and noise is reduced, and the prediction reliability is further improved, so that better support is provided for the operation and management of the photovoltaic power generation system.
According to the invention, the sunrise and sunset code is utilized in the prediction process, and the illumination change near sunrise and sunset time is particularly concerned, because the time usually has obvious influence on the photovoltaic power, the influence of the sun position change on the power can be better captured by incorporating the information into a model, so that the prediction accuracy is improved, important support is provided for the reliable operation of the photovoltaic power generation system, and the real-time scheduling requirement of the power system is met especially near sunrise and sunset time. The energy waste is reduced, and the potential power generation capacity of the photovoltaic battery pack is effectively utilized.
The method and the system consider the characteristics of the time point prediction such as the number of minutes of hours, the number of hours in a day, the day of the week, the date in a month, the date in a year and the like, describe the influence of time more comprehensively, enhance the flexibility, enable the method and the system to adapt to the illumination conditions and time changes of different regions and seasons, and provide better support for the operation and management of the new energy power station.
The invention provides a more advanced, more reliable and more practical power prediction method for the photovoltaic power generation field so as to promote the sustainable development of clean energy.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the invention.
FIG. 1 is a schematic flow chart of a photovoltaic power generation ultra-short term power prediction method taking sunrise and sunset time into consideration;
fig. 2 is a schematic structural diagram of a photovoltaic power generation ultra-short term power prediction system taking sunrise and sunset time into consideration;
FIG. 3 is a schematic diagram of an electronic device according to the present invention;
FIG. 4 is a flowchart of a photovoltaic power generation ultra-short term power prediction method considering sunrise and sunset time in an embodiment of the invention;
FIG. 5 is a schematic diagram of a BiLSTM neural network in accordance with an embodiment of the present invention;
FIG. 6 is a schematic diagram of an Attention unit structure in an embodiment of the present invention;
FIG. 7 shows the month accuracy of the fourth hour sunrise and sunset interval prediction of the test set according to the embodiment of the present invention.
Detailed Description
The invention will be described in detail below with reference to the drawings in connection with embodiments. It should be noted that, in the case of no conflict, the embodiments and features in the embodiments may be combined with each other.
The following detailed description is exemplary and is intended to provide further details of the invention. Unless defined otherwise, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments in accordance with the invention.
Referring to fig. 1, a photovoltaic power generation ultra-short term power prediction method considering sunrise and sunset time includes the following steps:
collecting original data, wherein the original data comprises actual measurement photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
preprocessing the original data to obtain a training set and a testing set;
constructing a photovoltaic power generation ultra-short-term power prediction characteristic by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic power generation ultra-short-term power prediction characteristic comprises sunrise and sunset characteristic codes, and the sunrise and sunset characteristic codes judge whether the prediction time is close to sunrise and sunset;
establishing a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts the photovoltaic power generation power of four hours in the future at a prediction time point;
training a photovoltaic power generation ultra-short term power prediction model by using a training set to obtain a photovoltaic power generation ultra-short term power prediction result;
and verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
The photovoltaic ultra-short term power prediction features also include original features, statistical features and time codes;
the original features are specifically actual measurement photovoltaic power generation power of a predicted day in a sliding time window, the statistical features are specifically features extracted from the original features by using a statistical method, and the statistical method comprises average value, median and standard deviation;
the sunrise and sunset characteristic codes utilize cosine codes or time distances, the cosine codes specifically code each day of the year into a value from 0 to 1, a cosine function is used for obtaining periodic codes, and sunrise and sunset time shows periodic change in the year; the time distance is the time difference between the current time and the sunrise and sunset time;
the time code includes minutes, hours, weeks, months, and years corresponding to the predicted time point.
The photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and Attention mechanism established by utilizing the photovoltaic power generation ultra-short-term power prediction characteristics specifically comprises the following steps: firstly, inputting original data in a historical time window into a BiLSTM layer, capturing time sequence information and sequence relation of the original data in the historical time window by the BiLSTM layer, and then focusing Attention on the time sequence information and sequence relation related to prediction by a photovoltaic power generation ultra-short-term power prediction model through an Attention mechanism, and learning fluctuation and trend of photovoltaic power generation in different time periods.
The Attention mechanism identifies and extracts key features from data within a preselected historical time window and assigns weights to the key features and the secondary features, the key features being assigned a higher weight than the secondary features.
And evaluating the photovoltaic power generation ultra-short-term power prediction result of the photovoltaic power generation ultra-short-term power prediction model after verifying the photovoltaic power generation ultra-short-term power prediction result by using a verification set, wherein the evaluation comprises calculation and analysis of performance indexes, wherein the performance indexes comprise average absolute errors, and the formula is as follows:
wherein MAE is the average absolute error, m is the number of samples, y i To be a true value of the value,is a predicted value of the model;
the value range of the MAE is [0, + ], when the photovoltaic power generation ultra-short-term power prediction model predicts completely accurately, the calculated MAE is 0, which means that the accuracy of the prediction result of the photovoltaic power generation ultra-short-term power prediction model reaches 100%, and the photovoltaic power generation ultra-short-term power prediction model is a perfect model.
The month accuracy and the day accuracy of the photovoltaic power generation ultra-short term power prediction result are checked by using a calculation formula of the photovoltaic power station ultra-short term power prediction check index accuracy, the day accuracy of the photovoltaic power generation ultra-short term power prediction method in the fourth hour is greater than or equal to 90%, and when the day accuracy of the photovoltaic power generation ultra-short term power prediction method in the fourth hour is less than 90%, the following formula is checked:
wherein: p (P) Mi For the actual power at time i, P Pi For the i-th time predicted value of ultra-short-term power prediction, cap is the available capacity of a photovoltaic power station, n is the number of samples in the photovoltaic power generation period, and P N The installed capacity of the photovoltaic power station;
the month accuracy of the photovoltaic power generation ultra-short period power prediction method is an arithmetic average value of the day accuracy of the photovoltaic power generation ultra-short period power prediction method.
The preprocessing specifically comprises the steps of cleaning original data to obtain a data set, and dividing the data set into a training set and a testing set; cleaning is specifically deleting or repairing any missing, abnormal or inaccurate data points in the original data.
Referring to fig. 2, a photovoltaic power generation ultra-short term power prediction system considering sunrise and sunset time comprises an acquisition data module, a data processing module and a data processing module, wherein the acquisition data module is used for acquiring original data, and the original data comprise actual measured photovoltaic power generation power of a photovoltaic station for predicting a day and sunrise and sunset time of the predicted day;
the preprocessing module is used for preprocessing the original data to obtain a training set and a testing set;
the characteristic construction module is used for constructing photovoltaic power generation ultra-short-term power prediction characteristics by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic ultra-short-term power prediction characteristics comprise sunrise and sunset characteristic codes, and the sunrise and sunset characteristic codes judge whether the prediction time is close to sunrise and sunset;
the model building module is used for building a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing the photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts the photovoltaic power generation power of four hours in the future at a prediction time point;
the model training module is used for training a photovoltaic power generation ultra-short-term power prediction model by using the training set to obtain a photovoltaic power generation ultra-short-term power prediction result;
and the result verification module is used for verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
Referring to fig. 3, an electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, which when executed implements a photovoltaic power generation ultrashort-term power prediction method that takes into account sunrise and sunset time. A photovoltaic power generation ultra-short-term power prediction method considering sunrise and sunset time comprises the following steps: collecting original data, wherein the original data comprises actual measurement photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
preprocessing the original data to obtain a training set and a testing set;
constructing a photovoltaic power generation ultra-short-term power prediction characteristic by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic power generation ultra-short-term power prediction characteristic comprises sunrise and sunset characteristic codes, and the sunrise and sunset characteristic codes judge whether the prediction time is close to sunrise and sunset;
establishing a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts the photovoltaic power generation power of four hours in the future at a prediction time point;
training a photovoltaic power generation ultra-short term power prediction model by using a training set to obtain a photovoltaic power generation ultra-short term power prediction result;
and verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
A computer readable storage medium storing a computer program which when executed by a processor implements a photovoltaic power generation ultra-short term power prediction method that takes into account sunrise and sunset time. A photovoltaic power generation ultra-short-term power prediction method considering sunrise and sunset time comprises the following steps: collecting original data, wherein the original data comprises actual measurement photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
preprocessing the original data to obtain a training set and a testing set;
constructing a photovoltaic power generation ultra-short-term power prediction characteristic by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic power generation ultra-short-term power prediction characteristic comprises sunrise and sunset characteristic codes, and the sunrise and sunset characteristic codes judge whether the prediction time is close to sunrise and sunset;
establishing a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts the photovoltaic power generation power of four hours in the future at a prediction time point;
training a photovoltaic power generation ultra-short term power prediction model by using a training set to obtain a photovoltaic power generation ultra-short term power prediction result;
and verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
The invention is explained in further detail below with reference to examples:
a photovoltaic power generation ultra-short-term power prediction method considering sunrise and sunset time is shown in a flow chart of fig. 4, and a photovoltaic station with a loading capacity of 50MW is taken as an example.
Data preparation
(1) Raw data acquisition: collecting actual power generation data of the photovoltaic power plant from 2023, 7, 1 to 2023, 8, 14, covering a period of about 1 half month; the actual power generation power data of the photovoltaic power station is the final verification basis of the prediction accuracy;
in order to better consider the dynamic change of the photovoltaic power, according to the longitude and latitude and altitude information of the photovoltaic power station, the sunrise and sunset time of the day of the predicted day is obtained;
PyEphem is a Python library for astronomical calculations, behind which complex astronomical algorithms and formulas are used to perform various calculations, including sunrise and sunset calculations. The invention calculates the function sun_rise_set_transit_ephem of sunrise, sunset and solar transit time by using PyEphem, and the function returns sunrise, sunset and solar transit time according to preset time, longitude and latitude, altitude, air pressure and temperature. (2) data cleaning: the original data is cleaned, and any missing, abnormal or inaccurate data points are deleted or repaired to form a data set.
(3) Data segmentation: for model training and validation, the entire dataset was divided into training and test sets, in a 7:3 ratio. I.e. 70% of the data were used to train (2023, 7, 1, to 2023, 7, 31, days) the model, while the remaining 30% were used to test (2023, 8, 1, to 2023, 8, 14 days) the predictive effect of the model.
(II) model feature construction
Characteristic engineering: and constructing characteristics related to photovoltaic ultra-short-term power prediction.
The following are the details of the photovoltaic power prediction-related feature engineering:
(1) Historical power data
Historical power data over different time windows of the past 1 hour, 2 hours, 3 hours, 4 hours, or 6 hours, etc., is selected for use as desired. The flexibility of the sliding time window enables the model to be adjusted according to actual conditions and data properties so as to achieve the optimal prediction effect. For example, if data is collected every 15 minutes, then the power value for the first four hours (i.e., the first 16 data points) may be used as a feature.
Calculating statistical features, such as mean, median, standard deviation, etc., using historical power data can help the model capture long-term trends and short-term fluctuations.
To more fully understand the impact of historical data over different time spans on the predicted performance, attempts were made to predict the photovoltaic power for the next 4 hours using data over historical time windows of the past 1 hour, 2 hours, 3 hours, 4 hours, and 6 hours, and to construct corresponding training data sets and prediction models (simply "1_to_4", "2_to_4", "3_to_4", "4_to_4", and "6_to_4" prediction models).
The following are specific examples of training data and corresponding labels when predicted with time window data for the past 4 hours:
the time window of the training data 'train_x' is: 2023-05-07:15:00 to 2023-05-0708:00:00, contained 16 data points altogether (15 minutes time interval). These data points (i.e., samples) include characteristics of measured power, hours, and minutes.
The time window of the training label 'train_label' is: 2023-05-07:08:15:00 to 2023-05-0712:00:00 also contained 16 data points. These data points represent photovoltaic power over the next 4 hours.
(2) Sunrise and sunset code
Photovoltaic power generation is significantly affected by sunlight, so sunrise and sunset times become key factors for prediction. And these times vary due to seasonal and geographic variations.
Since sunrise and sunset times exhibit significant periodic changes over the year, cosine codes can be used to capture this periodicity. Specifically, each day of the year may be encoded as a value from 0 to 1, and then the periodic encoding is obtained using a cosine function.
In addition to directly using sunrise and sunset times, the time difference between the current time and sunrise and sunset can be calculated, which helps the model more accurately predict power changes before and after sunrise and before and after sunset.
(3) Time coding
The time impact is more fully described taking into account the number of minutes of hours, hours in a day, days of the week, dates in a month, dates in a year, etc. characteristics of the predicted time point.
In conclusion, the characteristic engineering can remarkably improve the photovoltaic power prediction performance. By combining the historical data and sunrise and sunset information, the model can better understand and capture the change rule of the power, so that more accurate prediction can be made.
(4) Selecting a model based on BiLSTM and attribute mechanism model training time sequence prediction, training the model by using training set data, and continuously adjusting model parameters until the prediction effect of the model on the training set reaches a satisfactory level.
The data in the past history time window is sent to the BiLSTM layer, and the advantage is that the time sequence information and the sequence relation of the data can be captured, which is very critical for the prediction of the photovoltaic power generation power. Through BiLSTM, the model can learn the hidden complex time sequence mode in the data, including fluctuation and trend of the photovoltaic power in different time periods.
The Attention mechanism is a technique for dynamically extracting key information from multiple data sources. In photovoltaic power generation power prediction, the function is to automatically pay attention to information most useful for prediction, thereby reducing the influence of uncorrelated variables and noise on prediction. Through the Attention mechanism, the model can weight information according to the importance of the data, so that the model is more focused on the characteristics with obvious influence on the power prediction, and the prediction accuracy is improved.
In the invention, a BiLSTM and an Attention mechanism are adopted, the BiLSTM is an optimized improvement of the traditional unidirectional LSTM, and the BiLSTM combines a forward LSTM layer and a backward LSTM layer, and both layers can influence output. The single LSTM can fully utilize the data history information, avoid the generation of long-distance dependence, and the BiLSTM is beneficial to the input of forward sequence information and backward sequence information, fully considers the past and future information, and is beneficial to further improving the accuracy of model prediction. The BiLSTM structure is shown in FIG. 5. Wherein x is 1 ，x 2 ，x 3 ，...，x t Representing t 1 ～t i (i∈[1～t]) Input data corresponding to each moment, A 1 ，A 2 ，A 3 ，...，A t ，B 1 ，B 2 ，B 3 ，...，B t Representing the LSTM hidden states of the respective forward and backward iterations, Y 1 ，Y 2 ，Y 3 ，...，Y t Representing the corresponding output data, ω 1 ，ω 2 ，ω 3 ，...，ω 6 Representing the corresponding weights of the layers.
The hidden layer update state of the forward LSTM and the backward LSTM and the BiLSTM final output process are shown in the following formulas:
A i ＝f 1 (ω 1 x i +ω 2 A i-1 )
B j ＝f 2 (ω 3 x i +ω 5 B i+1 )
Y i ＝f 3 (ω 4 A i +ω 6 B i )
the Attention originates from the simulation of the human brain Attention characteristics, and the method is firstly applied to the field of image processing. In the field of deep learning, the Attention mechanism distributes larger weights to key contents and smaller weights to other contents according to different characteristics, and the information processing efficiency can be improved through differentiated weight distribution. The Attention cell structure is shown in fig. 6.
The Attention state transition procedure is shown in the following formula.
S ti ＝Vtan(Wh t +Uh i +b)，i＝1，2，3...t-1
h t′ ＝f(F，h t ，y t )
Wherein: a, a ti Output value h for BiLSTM hidden layer i The currently input Attention weight value; y is 1 ，y 2 ，y 3 ，...，y t Is an input sequence; h is a 1 ，h 2 ，h 3 ，...，h t To correspond to the input sequence y 1 ，y 2 ，y 3 ，...，y t Hidden layer state value of (h), i.e. h t For y corresponding to input t Hiding the layer state value; h is a t′ Is the final feature vector; v, W, U, b are learning parameters of the model, which are updated continuously with the model training process.
Alternatively, photovoltaic power predictions are made using conventional time series models, such as ARIMA (autoregressive integral moving average) and the like.
Alternatively, prediction is performed using only the BiLSTM and Attention mechanisms, irrespective of sunrise and sunset coding and time coding.
(III) model verification and evaluation
(1) Model verification: the predicted effect of the model is verified using the separated test set data.
(2) Performance evaluation: based on the prediction results of the model on the test set, various performance indicators are calculated and analyzed,
the MAE value range is [0, + ], when the model prediction is completely accurate, the calculated MAE is 0, which represents that the model prediction accuracy reaches 100%, and the model is a perfect model. In the formula, m is the number of samples, y i To be a true value of the value,is the predicted value of the model.
(IV) results of experiments
The month accuracy and the day accuracy of the photovoltaic power generation ultra-short term power prediction result are checked by utilizing a photovoltaic power station ultra-short term power prediction check index accuracy calculation formula, and the photovoltaic power station ultra-short term power prediction check index accuracy calculation formula refers to a fifteenth chapter of a second chapter of a national energy office in China regulatory agency document:
the solar accuracy rate of the photovoltaic power station in the 4 th hour of ultra-short period power prediction is greater than or equal to 90%, and when the solar accuracy rate is less than 90%, the following formula is adopted for checking:
wherein: p (P) Mi For the actual power at time i, P Pi For predicting 4 th hour (i moment) predicted value of ultra-short-term power, cap is available capacity of a photovoltaic power station, n is number of samples in a power generation period, and P N And (5) loading capacity for the photovoltaic power station.
The month accuracy is the arithmetic average of the day accuracy.
As shown in fig. 7, the month accuracy of the ultra-short-term prediction results before and after sunset encoding was compared. It is obvious from the figure that the month accuracy is improved in different degrees in the sunrise and sunset interval after the sunrise and sunset codes are applied. This result fully demonstrates the effectiveness and practicality of the proposed sunrise and sunset encoding algorithm.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention 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 invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. 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.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.

Claims (10)

1. The ultra-short-term power prediction method for photovoltaic power generation considering sunrise and sunset time is characterized by comprising the following steps of:
collecting original data, wherein the original data comprises actual measurement photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
preprocessing the original data to obtain a training set and a testing set;
constructing a photovoltaic power generation ultra-short-term power prediction characteristic by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic power generation ultra-short-term power prediction characteristic comprises a sunrise and sunset characteristic code, and the sunrise and sunset characteristic code judges whether the prediction time is close to sunrise and sunset;
establishing a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts photovoltaic power generation power of four hours in the future at a prediction time point;
training a photovoltaic power generation ultra-short term power prediction model by using a training set to obtain a photovoltaic power generation ultra-short term power prediction result;
and verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
2. The photovoltaic power generation ultra-short term power prediction method considering sunrise and sunset time according to claim 1, wherein the photovoltaic ultra-short term power prediction features further comprise an original feature, a statistical feature and a time code;
the original features are specifically actual measurement photovoltaic power generation power of a predicted day in a sliding time window, the statistical features are specifically features extracted from the original features by using a statistical method, and the statistical method comprises an average value, a median and a standard deviation;
the time code comprises minutes, hours, weeks, months and years corresponding to the predicted time point; the sunrise and sunset characteristic codes use cosine codes or time distances, the cosine codes specifically code each day in one year into a value from 0 to 1, a cosine function is used for obtaining periodic codes, and sunrise and sunset time shows periodic change in one year; the time distance is the time difference between the current time and the sunrise and sunset time.
3. The photovoltaic power generation ultra-short term power prediction method considering sunrise and sunset time according to claim 1, wherein the photovoltaic power generation ultra-short term power prediction model considering sunrise and sunset time established by utilizing photovoltaic power generation ultra-short term power prediction characteristics specifically comprises: firstly, inputting original data in a historical time window into a BiLSTM layer, capturing time sequence information and sequence relation of the original data in the historical time window by the BiLSTM layer, and then focusing Attention on the time sequence information and sequence relation related to prediction by a photovoltaic power generation ultra-short-term power prediction model through an Attention mechanism, and learning fluctuation and trend of photovoltaic power generation in different time periods.
4. The ultra-short term power prediction method for photovoltaic power generation taking into account sunrise and sunset time according to claim 1, wherein the Attention mechanism identifies key features and secondary features from data within a preselected historical time window and extracts key features and assigns weights to the key features and secondary features, the weight assigned by the key features being higher than that of the secondary features.
5. The photovoltaic power generation ultra-short term power prediction method considering sunrise and sunset time according to claim 1, wherein the photovoltaic power generation ultra-short term power prediction result of the photovoltaic power generation ultra-short term power prediction model is evaluated after the photovoltaic power generation ultra-short term power prediction result is verified by using a verification set, the evaluation comprises calculation and analysis of performance indexes, the performance indexes comprise average absolute errors, the accuracy of the prediction result of the photovoltaic power generation ultra-short term power prediction model at the time of 0 is 100%, and the photovoltaic power generation ultra-short term power prediction model is a perfect model.
6. The photovoltaic power generation ultra-short term power prediction method considering sunrise and sunset time according to claim 1, wherein the month accuracy and the day accuracy of the photovoltaic power generation ultra-short term power prediction result are checked by using a photovoltaic power station ultra-short term power prediction check index accuracy calculation formula.
7. The photovoltaic power generation ultra-short term power prediction method considering sunrise and sunset time according to claim 1, wherein the preprocessing is specifically to clean raw data to obtain a data set, and divide the data set into a training set and a test set; the cleaning is specifically to delete or repair any missing, abnormal or inaccurate data points in the original data.
8. The photovoltaic power generation ultra-short-term power prediction system taking sunrise and sunset time into consideration is characterized by comprising a data acquisition module, wherein the data acquisition module is used for acquiring original data, and the original data comprise actual photovoltaic power generation power of a photovoltaic station on a prediction day and sunrise and sunset time of the prediction day;
the preprocessing module is used for preprocessing the original data to obtain a training set and a testing set;
the characteristic construction module is used for constructing photovoltaic power generation ultra-short-term power prediction characteristics by utilizing a sliding time window, wherein the sliding time window specifically uses original data in different historical time windows according to preset selection, the photovoltaic ultra-short-term power prediction characteristics comprise sunrise and sunset characteristic codes, and the sunrise and sunset characteristic codes judge whether the prediction time is close to sunrise and sunset;
the model building module is used for building a photovoltaic power generation ultra-short-term power prediction model based on BiLSTM and attribute mechanisms by utilizing the photovoltaic power generation ultra-short-term power prediction characteristics; the photovoltaic power generation ultra-short term power prediction model predicts photovoltaic power generation power of four hours in the future at a prediction time point;
the model training module is used for training a photovoltaic power generation ultra-short-term power prediction model by using the training set to obtain a photovoltaic power generation ultra-short-term power prediction result;
and the result verification module is used for verifying the photovoltaic power generation ultra-short term power prediction result by using the test set.
9. An electronic device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the photovoltaic power generation ultrashort term power prediction method taking sunrise and sunset time into account of any of claims 1-7 when executing the computer program.
10. A computer readable storage medium storing a computer program which, when executed by a processor, implements the sunrise and sunset time considered photovoltaic power generation ultrashort term power prediction method of any one of claims 1 to 7.
CN202311255468.0A 2023-09-26 2023-09-26 Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time Pending CN117277304A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311255468.0A CN117277304A (en) 2023-09-26 2023-09-26 Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311255468.0A CN117277304A (en) 2023-09-26 2023-09-26 Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time

Publications (1)

Publication Number Publication Date
CN117277304A true CN117277304A (en) 2023-12-22

Family

ID=89210015

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311255468.0A Pending CN117277304A (en) 2023-09-26 2023-09-26 Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time

Country Status (1)

Country Link
CN (1) CN117277304A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117767295A (en) * 2023-12-25 2024-03-26 泰富江苏共享网络科技有限公司 Method for predictive analysis of stable power generation time period of photovoltaic power station

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117767295A (en) * 2023-12-25 2024-03-26 泰富江苏共享网络科技有限公司 Method for predictive analysis of stable power generation time period of photovoltaic power station

Similar Documents

Publication Publication Date Title
CN113962364B (en) Multi-factor power load prediction method based on deep learning
CN110263866B (en) Power consumer load interval prediction method based on deep learning
Azadi et al. Verifying the performance of artificial neural network and multiple linear regression in predicting the mean seasonal municipal solid waste generation rate: A case study of Fars province, Iran
CN110766212B (en) Ultra-short-term photovoltaic power prediction method for historical data missing electric field
CN103117546B (en) A kind of Ultrashort-term slide prediction method for wind power
CN108388962B (en) Wind power prediction system and method
CN108196317B (en) Meteorological prediction method for micro-grid system
CN114510513A (en) Short-term meteorological forecast data processing method for ultra-short-term photovoltaic power prediction
US10770898B2 (en) Methods and systems for energy use normalization and forecasting
CN116151459B (en) Power grid flood prevention risk probability prediction method and system based on improved Transformer
CN117277304A (en) Photovoltaic power generation ultra-short-term power prediction method and system considering sunrise and sunset time
CN109840633A (en) Photovoltaic output power predicting method, system and storage medium
CN113962441A (en) Short-term irradiance prediction method and prediction system based on historical data analysis
CN117526274A (en) New energy power prediction method, electronic equipment and storage medium in extreme climate
CN116029419A (en) Deep learning-based long-term new energy daily average generation power prediction method and system
CN117114438A (en) Building area energy system cold and hot load data driving prediction method with flexibility and interpretability
Alharbi et al. Short-term wind speed and temperature forecasting model based on gated recurrent unit neural networks
CN108960522A (en) A kind of photovoltaic power generation quantity prediction analysis method
Li et al. Deep learning model for short-term photovoltaic power forecasting based on variational mode decomposition and similar day clustering
CN114895380A (en) Solar radiation prediction method, device, equipment and medium
Gao et al. Spatio-temporal interpretable neural network for solar irradiation prediction using transformer
Deng et al. A survey of the researches on grid-connected solar power generation systems and power forecasting methods based on ground-based cloud atlas
Ghassemi et al. Optimal surrogate and neural network modeling for day-ahead forecasting of the hourly energy consumption of university buildings
CN113984198A (en) Short wave radiation prediction method and system based on convolutional neural network
CN105205572A (en) Photovoltaic power generation output forecasting method based on EMD (Empirical Mode Decomposition) and Elman algorithm

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