CN117194957A - Ultra-short-term prediction method based on satellite inversion radiation data technology - Google Patents
Ultra-short-term prediction method based on satellite inversion radiation data technology Download PDFInfo
- Publication number
- CN117194957A CN117194957A CN202311291774.XA CN202311291774A CN117194957A CN 117194957 A CN117194957 A CN 117194957A CN 202311291774 A CN202311291774 A CN 202311291774A CN 117194957 A CN117194957 A CN 117194957A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- forecast
- satellite inversion
- satellite
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000005855 radiation Effects 0.000 title claims abstract description 76
- 238000000034 method Methods 0.000 title claims abstract description 73
- 238000005516 engineering process Methods 0.000 title claims abstract description 29
- 238000010248 power generation Methods 0.000 claims abstract description 28
- 230000009897 systematic effect Effects 0.000 claims description 26
- 238000012549 training Methods 0.000 claims description 23
- 238000010586 diagram Methods 0.000 claims description 22
- 238000012937 correction Methods 0.000 claims description 20
- 238000010801 machine learning Methods 0.000 claims description 19
- 238000004422 calculation algorithm Methods 0.000 claims description 18
- 238000009826 distribution Methods 0.000 claims description 17
- 238000011156 evaluation Methods 0.000 claims description 15
- 238000013136 deep learning model Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 10
- 238000004458 analytical method Methods 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 9
- 238000005457 optimization Methods 0.000 claims description 8
- 238000013473 artificial intelligence Methods 0.000 claims description 6
- 238000000611 regression analysis Methods 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 6
- 230000002159 abnormal effect Effects 0.000 claims description 4
- 238000013527 convolutional neural network Methods 0.000 claims description 4
- 238000005286 illumination Methods 0.000 claims description 4
- 238000000528 statistical test Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 3
- 125000004122 cyclic group Chemical group 0.000 claims description 3
- 230000000694 effects Effects 0.000 claims description 3
- 238000012854 evaluation process Methods 0.000 claims description 3
- 238000012417 linear regression Methods 0.000 claims description 3
- 238000007637 random forest analysis Methods 0.000 claims description 3
- 230000001932 seasonal effect Effects 0.000 claims description 3
- 238000007619 statistical method Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000007787 long-term memory Effects 0.000 claims description 2
- 238000012821 model calculation Methods 0.000 claims description 2
- 230000006403 short-term memory Effects 0.000 claims description 2
- 238000011160 research Methods 0.000 abstract description 3
- 206010063385 Intellectualisation Diseases 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010224 classification analysis Methods 0.000 description 2
- 238000007621 cluster analysis Methods 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 239000000443 aerosol Substances 0.000 description 1
- 239000008264 cloud Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000015654 memory Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000010998 test method Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Abstract
The invention relates to the technical field of natural science research prediction, and discloses an ultra-short-term prediction method based on satellite inversion radiation data technology. The ultra-short-term prediction method based on the satellite inversion radiation data technology acquires the incident radiation data of the earth surface sun in real time, breaks through the limitation of the traditional radiation observation method, and provides accurate data support for photovoltaic power generation operation.
Description
Technical Field
The invention relates to the technical field of natural science research prediction, in particular to an ultra-short-term prediction method based on a satellite inversion radiation data technology.
Background
The national meteorological satellite center stably provides ground incident solar radiation inversion products from 3 months and 12 days in 2018, is one of L2-level quantitative inversion products of wind clouds 4A, takes influence parameters such as clouds, aerosols, water vapor content, earth surface albedo, earth surface altitude and the like into consideration, has a spatial resolution of 4km, has a highest time resolution of up to 15min (40 observation times in the whole day, and is observed once every 15min before and after every 3h of whole point except for each whole point time observation), and the earth surface solar radiation inversion products of the wind clouds 4A comprise 3 elements such as total radiation, horizontal plane direct radiation, scattered radiation and the like. The method can effectively solve the problems of sparse ground radiation observation sites and long observation time interval, can be used as a new means for carrying out high space-time resolution ground solar incident radiation observation in the northern river basin, and provides possibility for relevant research and service development.
The incident radiation of the surface solar is one of FY-4A L2 quantitative inversion products, and a new radiation observation source is provided for developing photovoltaic power modeling.
The traditional radiation observation method is difficult to accurately measure and predict the solar incident radiation on the earth surface due to various limitations, such as sparsity of observation equipment, uncertainty of environmental factors and the like, and therefore, a corresponding technical scheme needs to be designed for solving the problems.
Disclosure of Invention
(one) solving the technical problems
Aiming at the defects of the prior art, the invention provides an ultra-short-term prediction method based on a satellite inversion radiation data technology, which solves the technical problems that the traditional radiation observation method is difficult to accurately measure and predict the incident radiation of the earth surface sun due to various limitations of sparsity of observation equipment, uncertainty of environmental factors and the like.
(II) technical scheme
In order to achieve the above purpose, the invention is realized by the following technical scheme: a method for ultra-short-term prediction based on satellite inversion radiation data technology comprises the following steps:
s1, performing applicability analysis on earth surface solar incident radiation inverted by a satellite by utilizing an artificial intelligence technology, and evaluating the accuracy and stability of satellite inversion data by utilizing historical data through a machine learning algorithm to determine the capability of the satellite inversion data for ultra-short-term prediction;
s2, correcting satellite inversion data by adopting a PDF method, optimizing a photovoltaic power model, and correcting systematic errors by comparing probability density distribution of the satellite inversion data and real radiation observation data;
s3, checking and correcting meteorological elements of the multi-mode forecasting product, evaluating the accuracy and stability of the multi-mode forecasting product by using a checking method of statistical checking and graph comparison, and correcting systematic errors of forecasting data by using a PDF method to reduce the systematic errors of mode forecasting;
s4, applying the corrected satellite inversion data and the corrected multimode forecast data to a photovoltaic power generation weather service business, and constructing a photovoltaic power forecast model by combining the actual condition of photovoltaic power generation, wherein the model is used for providing a high-level weather forecast service for photovoltaic power generation.
Preferably, in step S1, applicability analysis is performed on satellite inverted surface solar incident radiation by using artificial intelligence technology, and the specific method comprises the following steps:
constructing a deep learning model aiming at satellite inversion data characteristics, and automatically learning and identifying satellite inversion data characteristics by the model;
in the evaluation process, the deep learning model is used for receiving satellite inversion data as input, outputting an evaluation result of the accuracy and the stability of the satellite inversion data, automatically classifying and clustering the satellite inversion data according to the evaluation result, classifying the data into different categories or clusters, and independently evaluating and analyzing each category or cluster;
for different data types or clusters, the model automatically learns and recognizes different characteristics and modes, automatically evaluates the applicability of the model, and is suitable for data in different geographic areas or in different seasons;
the deep learning model outputs applicability analysis results of satellite inversion data, including accuracy, stability and reliability of the data, and applicability evaluation results for different data categories or clusters.
Preferably, the deep learning model adopts a convolutional neural network, a cyclic neural network or a long-term and short-term memory network structure.
Preferably, in step S1, the machine learning algorithm evaluates satellite inversion data using the historical data, including the following method steps:
selecting a machine learning algorithm suitable for processing satellite inversion data, wherein the machine learning algorithm comprises a support vector machine, a random forest and a gradient lifting tree, performing feature engineering on historical data, extracting suitable features and using the features for training and evaluating a model;
selecting representative data from the historical data as a training set for training a machine learning model;
training a machine learning model by utilizing a training set, and evaluating and optimizing the model by adjusting model parameters and adopting a cross verification technology;
evaluating the trained model by using an independent test set, and comparing the accuracy and stability of satellite inversion data predicted by the model and real radiation observation data;
adjusting and optimizing the model according to the evaluation result;
the training set and the test set are updated continually, and the model is trained and evaluated continually using the new data.
Preferably, the evaluation index includes, but is not limited to, mean square error, root mean square error, and mean absolute error; the model adjustments and optimizations include, but are not limited to, adding features, changing model structures, and adjusting hyper-parameters.
Preferably, in step S2, the method step of comparing the probability density distribution of the satellite inversion data and the real radiation observation data includes:
carrying out data cleaning, missing value filling and outlier processing pretreatment on satellite inversion data and real radiation observation data;
respectively constructing probability density functions by utilizing the preprocessed satellite inversion data and the real radiation observation data;
comparing the constructed probability density functions, analyzing the similarity and the difference of the probability density functions, and calculating the distance, intersection and union between two PDFs, wherein the distance, intersection and union are used for evaluating the errors and the differences of satellite inversion data and real radiation observation data;
determining systematic errors between satellite inversion data and real radiation observation data according to the comparison result;
performing systematic error correction on satellite inversion data according to the magnitude and the direction of the systematic error;
and after systematic error correction is carried out, comparing probability density distribution of satellite inversion data and real radiation observation data, and evaluating correction effect.
Preferably, the method of systematic error correction includes, but is not limited to, linear regression correction of data and correction using interpolation methods.
Preferably, in step S3, the specific method for statistical test includes:
average comparison: comparing the average value of the forecast value and the real observed value of each forecast mode, and calculating absolute error and relative error indexes to measure the accuracy of forecast;
variance comparison: comparing the variance of the forecast values of each forecast mode with the variance of the real observed values, wherein a smaller variance indicates that the forecast values are relatively stable, and a larger variance indicates that the forecast values fluctuate more;
correlation coefficient comparison: calculating a correlation coefficient between the forecast value and the real observed value, wherein the closer the correlation coefficient is to 1, the more relevant the forecast mode and the real observed value are;
regression analysis: establishing a regression equation between the prediction mode and the real observed value through regression analysis;
the specific method for graph comparison comprises the following steps:
timing diagram: the forecast values and the real observed values of the various forecast modes are drawn into a time chart, and the forecast quality is evaluated by observing the trend and fluctuation conditions of the time chart.
Scatter plot: and drawing the forecast values and the real observed values of the forecast modes into a scatter diagram, and evaluating the accuracy and the stability of the forecast modes by observing the distribution condition of the scatter diagram.
Box line diagram: and drawing the forecasting values of the forecasting modes into a box diagram, and evaluating the forecasting stability by observing the box body and abnormal value conditions of the box diagram.
Preferably, in step S4, the specific method steps for constructing the photovoltaic power forecast model in combination with photovoltaic power generation include the following steps:
incorporating more influencing factors into the model including geographic location, climate characteristics, seasonal variations, and weather systems;
dynamically adjusting model parameters according to real-time satellite inversion data and multi-mode forecast data by using a dynamic optimization algorithm;
introducing uncertainty factors of satellite inversion data errors, multimode forecast data errors, geographic positions and climate characteristic changes, and processing and analyzing the uncertainty factors by a probability statistical method;
and the model calculation and storage are distributed to a plurality of calculation nodes by adopting a distributed architecture, so that parallel calculation and rapid training of the model are realized.
Preferably, the more influencing factors further comprise wind speed, wind direction, air pressure, temperature, humidity, solar radiation data, type of equipment, performance parameters of equipment, running state of equipment, illumination time, solar altitude angle, source of data, data accuracy and data reliability.
(III) beneficial effects
Compared with the prior art, the invention has the beneficial effects that: the satellite inversion data technology is adopted, solar incident radiation data of the earth surface are obtained in real time, the limitation of the traditional radiation observation method is broken through, and more accurate data support is provided for the operation of photovoltaic power generation; the PDF method is adopted to correct satellite inversion data, so that the accuracy of a photovoltaic power model is improved, and powerful support is provided for the optimal operation of photovoltaic power generation; the weather elements of the multi-mode forecast product are inspected and corrected, the accuracy of forecast is further improved, and a more reliable data basis is provided for weather forecast service of photovoltaic power generation; the satellite inversion data and the multi-mode forecast data are applied to the photovoltaic power generation weather service business, and a high-level weather forecast service is provided for photovoltaic power generation.
Drawings
FIG. 1 is a schematic diagram of the method steps of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Referring to fig. 1, the embodiment of the invention provides a technical scheme: a method for ultra-short-term prediction based on satellite inversion radiation data technology comprises the following steps:
s1, performing applicability analysis on earth surface solar incident radiation inverted by a satellite by utilizing an artificial intelligence technology, and evaluating the accuracy and stability of satellite inversion data by utilizing historical data through a machine learning algorithm to determine the capability of the satellite inversion data for ultra-short-term prediction;
s2, correcting satellite inversion data by adopting a PDF (probability density matching) method, optimizing a photovoltaic power model, and correcting systematic errors by comparing probability density distribution of the satellite inversion data and real radiation observation data so as to improve accuracy and reliability of the satellite inversion data, thereby improving accuracy of the photovoltaic power model;
s3, checking and correcting meteorological elements of the multi-mode forecasting product, evaluating the accuracy and stability of the multi-mode forecasting product by using a checking method of statistical checking and graph comparison, correcting systematic errors of forecasting data by using a PDF method, reducing the systematic errors of mode forecasting, and improving forecasting accuracy;
s4, applying the corrected satellite inversion data and the corrected multimode forecast data to a photovoltaic power generation weather service business, and constructing an efficient and accurate photovoltaic power forecast model by combining the actual condition of photovoltaic power generation, wherein the model is used for providing a high-level weather forecast service for photovoltaic power generation.
Further improved, in step S1, applicability analysis is performed on incident radiation of earth' S surface sun inverted by satellite by using artificial intelligence technology, and the specific method comprises the following steps:
constructing a deep learning model aiming at satellite inversion data characteristics, and automatically learning and identifying satellite inversion data characteristics by the model;
in the evaluation process, the deep learning model is used for receiving satellite inversion data as input, outputting an evaluation result of the accuracy and the stability of the satellite inversion data, automatically classifying and clustering the satellite inversion data according to the evaluation result, classifying the data into different categories or clusters, and independently evaluating and analyzing each category or cluster;
for different data types or clusters, the model automatically learns and recognizes different characteristics and modes, automatically evaluates the applicability of the model, and is suitable for data in different geographic areas or in different seasons;
the deep learning model outputs applicability analysis results of satellite inversion data, including accuracy, stability, reliability aspects of the data, and applicability evaluation results for different data categories or clusters.
The accuracy and the stability of the data can be improved, automation and intellectualization can be realized, and the efficiency and the accuracy of prediction are greatly improved.
Further improved, the deep learning model adopts a Convolutional Neural Network (CNN), a cyclic neural network (RNN) or a long-short-term memory network (LSTM) structure, and the accuracy and the stability of the deep learning model are improved through training historical data.
Further improved, in step S1, the machine learning algorithm evaluates satellite inversion data using historical data comprising the method steps of:
selecting a machine learning algorithm suitable for processing satellite inversion data, wherein the machine learning algorithm comprises a support vector machine, a random forest and a gradient lifting tree, and performing feature engineering on historical data according to the characteristics of the algorithm to extract suitable features for training and evaluating a model;
selecting representative data from the historical data as a training set for training a machine learning model, wherein the training set comprises satellite inversion data and real radiation observation data so as to facilitate model learning and comparison;
training a machine learning model by utilizing a training set, enabling the model to better fit historical data by adjusting model parameters, and evaluating and optimizing the model by adopting a cross verification technology;
evaluating the trained model by using an independent test set, and comparing the accuracy and stability of satellite inversion data predicted by the model and real radiation observation data;
adjusting and optimizing the model according to the evaluation result, and improving the accuracy and stability of the model;
the training set and the testing set are continuously updated, and the model is continuously trained and evaluated by using new data so as to improve the applicability and accuracy of the model.
The method for evaluating satellite inversion data by utilizing the historical data through the machine learning algorithm can not only improve the accuracy and stability of the data, but also realize automation and intellectualization, and greatly improve the prediction efficiency and accuracy. Meanwhile, the method can also carry out classification and cluster analysis according to factors such as different geographic areas, different seasons and the like, and provides finer and accurate reference basis for subsequent ultra-short-term prediction.
Further refinements, the evaluation metrics include, but are not limited to, mean Square Error (MSE), root Mean Square Error (RMSE), and Mean Absolute Error (MAE); the model adjustments and optimizations include, but are not limited to, adding features, changing model structures, and adjusting hyper-parameters.
Further improved, in step S2, the method step of comparing the probability density distribution of the satellite inversion data and the real radiation observation data includes:
carrying out data cleaning, missing value filling and outlier processing pretreatment on satellite inversion data and real radiation observation data so as to ensure the quality and reliability of the data;
respectively constructing Probability Density Functions (PDFs) of the preprocessed satellite inversion data and the real radiation observation data by utilizing the preprocessed satellite inversion data and the real radiation observation data, wherein the probability density functions can describe probability distribution conditions of the data so as to reflect characteristics and rules of the data;
comparing the constructed probability density functions, analyzing the similarity and the difference of the probability density functions, and calculating the distance, intersection and union between two PDFs, wherein the distance, intersection and union are used for evaluating the errors and the differences of satellite inversion data and real radiation observation data;
determining systematic errors between satellite inversion data and real radiation observation data according to comparison results, wherein the systematic errors generally refer to obvious errors such as overall offset, trend mismatch and the like of the data;
according to the magnitude and direction of the systematic error, performing systematic error correction on satellite inversion data to reduce the difference between the satellite inversion data and real radiation observation data;
and after systematic error correction is carried out, the probability density distribution of the satellite inversion data and the actual radiation observation data is compared again, and the correction effect is estimated.
By comparing the probability density distribution of the satellite inversion data and the actual radiation observation data, the accuracy and stability of the satellite inversion data are more accurately estimated, and a more reliable reference basis is provided for subsequent ultra-short-term prediction. Meanwhile, the method can also carry out classification and cluster analysis according to factors such as different geographic areas, different seasons and the like, further refine the comparison of probability density distribution and the correction of systematic errors, and provide finer and more accurate service for ultra-short-term prediction.
Further refinements, the systematic error correction method includes, but is not limited to, linear regression correction of the data and correction using interpolation methods.
Further improved, in step S3, the specific method for statistical test includes:
average comparison: comparing the average value of the forecast value and the real observed value of each forecast mode to evaluate the accuracy of forecast, and calculating absolute error and relative error indexes to measure the accuracy of forecast;
variance comparison: comparing the variance of the forecast values of each forecast mode with the variance of the real observed values to evaluate the stability of the forecast, wherein a smaller variance indicates that the forecast values are relatively stable, and a larger variance indicates that the forecast values fluctuate more;
correlation coefficient comparison: calculating a correlation coefficient between the forecast value and the real observed value to evaluate the forecast capability of the forecast mode, wherein the closer the correlation coefficient is to 1, the more relevant the forecast mode and the real observed value are;
regression analysis: establishing a regression equation between the prediction mode and the real observed value through regression analysis to evaluate the prediction capability of the prediction mode, wherein the accuracy and stability of the prediction mode can be reflected by the coefficient and intercept of the regression equation;
the specific method for graph comparison comprises the following steps:
timing diagram: and drawing the forecast values and the real observed values of the forecast modes into a time sequence diagram to intuitively evaluate the accuracy and the stability of forecast, and evaluating the forecast quality by observing the trend and the fluctuation condition of the time sequence diagram.
Scatter plot: the forecast values and the real observed values of the forecast modes are drawn into scatter diagrams to evaluate the difference and the correlation among the forecast modes, and the accuracy and the stability of the forecast modes are evaluated by observing the distribution condition of the scatter diagrams.
Box line diagram: the forecasting values of each forecasting mode are drawn into a box diagram to intuitively evaluate the forecasting stability, the box of the box diagram represents the quartile range of the forecasting values, the abnormal values are represented out of line, and the forecasting stability is evaluated by observing the box of the box diagram and the abnormal value conditions.
By the statistical test and graph comparison test method, the accuracy and stability of the multi-mode forecast product can be comprehensively evaluated, so that a reliable reference basis is provided for subsequent correction and forecast, the size of the forecast value is considered, the fluctuation condition and the distribution condition of the forecast value are also considered, and the performance of the forecast product can be evaluated more accurately.
Further improved, in step S4, the specific method steps for constructing the photovoltaic power forecast model in combination with photovoltaic power generation include the following steps:
incorporating more influencing factors into the model, including geographic location, climate characteristics, seasonal variations, and weather systems, to increase predictive power and accuracy of the model;
dynamically adjusting model parameters according to real-time satellite inversion data and multi-mode forecast data by utilizing a dynamic optimization algorithm so as to realize self-adaptive optimization of the model, and improving the prediction accuracy and efficiency of the model;
introducing uncertainty factors of satellite inversion data errors, multimode forecast data errors, geographic positions and climate characteristic changes, and processing and analyzing the uncertainty factors by a probability statistical method so as to improve the prediction accuracy of the model;
and the distributed architecture is adopted to distribute the calculation and storage of the model to a plurality of calculation nodes so as to improve the calculation efficiency and response speed of the model, realize the parallel calculation and rapid training of the model and improve the prediction capability of the model.
Through implementation of the technical scheme, other influencing factors such as geographic positions and the like can be considered, a dynamic optimization algorithm is adopted, uncertainty factors are considered, a distributed architecture is established to improve the calculation efficiency of the model, and therefore an efficient and accurate photovoltaic power forecasting model is constructed.
Specifically, more influencing factors include wind speed, wind direction, air pressure, temperature, humidity, solar radiation data, type of equipment, equipment performance parameters, equipment running state, illumination time, solar altitude angle, data sources, data accuracy and data reliability.
The solar radiation receiving and photovoltaic power generation efficiency can be influenced, and the data such as the intensity, time and spectral distribution of the solar radiation are incorporated into the model so as to more accurately predict the power output of the photovoltaic power generation; the characteristics of the photovoltaic power generation device can also affect its power output; the geographic position and the terrain have important influence on the photovoltaic power generation efficiency, and factors such as climate, illumination time, solar altitude angle and the like under different geographic positions and terrain conditions are included in a photovoltaic power forecast model; the data quality has important influence on the accuracy and reliability of the model, and the data sources, accuracy, reliability and other factors are included in the photovoltaic power forecasting model.
The ultra-short-term prediction method is a prediction method for predicting the future within 4 hours, and is mainly used for establishing a prediction model by analyzing historical meteorological data, then predicting the future meteorological data by using the model, and is widely applied to the fields of energy, traffic, agriculture and the like, thereby providing important help for the production and life of people.
Overview: the satellite inversion data technology is adopted, solar incident radiation data of the earth surface are obtained in real time, the limitation of the traditional radiation observation method is broken through, and more accurate data support is provided for the operation of photovoltaic power generation; the PDF method is adopted to correct satellite inversion data, so that the accuracy of a photovoltaic power model is improved, and powerful support is provided for the optimal operation of photovoltaic power generation; the weather elements of the multi-mode forecast product are inspected and corrected, the accuracy of forecast is further improved, and a more reliable data basis is provided for weather forecast service of photovoltaic power generation; the satellite inversion data and the multi-mode forecast data are applied to the photovoltaic power generation weather service business, so that a high-level weather forecast service is provided for photovoltaic power generation, and more accurate data support is provided for the operation of photovoltaic power generation.
While the fundamental and principal features of the invention and advantages of the invention have been shown and described, it will be apparent to those skilled in the art that the invention is not limited to the details of the foregoing exemplary embodiments, but may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.
Furthermore, it should be understood that although the present disclosure describes embodiments, not every embodiment is provided with a separate embodiment, and that this description is provided for clarity only, and that the disclosure is not limited to the embodiments described in detail below, and that the embodiments described in the examples may be combined as appropriate to form other embodiments that will be apparent to those skilled in the art.
Claims (10)
1. The ultra-short-term prediction method based on the satellite inversion radiation data technology is characterized by comprising the following steps of:
s1, performing applicability analysis on earth surface solar incident radiation inverted by a satellite by utilizing an artificial intelligence technology, and evaluating the accuracy and stability of satellite inversion data by utilizing historical data through a machine learning algorithm to determine the capability of the satellite inversion data for ultra-short-term prediction;
s2, correcting satellite inversion data by adopting a PDF method, optimizing a photovoltaic power model, and correcting systematic errors by comparing probability density distribution of the satellite inversion data and real radiation observation data;
s3, checking and correcting meteorological elements of the multi-mode forecasting product, evaluating the accuracy and stability of the multi-mode forecasting product by using a checking method of statistical checking and graph comparison, and correcting systematic errors of forecasting data by using a PDF method to reduce the systematic errors of mode forecasting;
s4, applying the corrected satellite inversion data and the corrected multimode forecast data to a photovoltaic power generation weather service business, and constructing a photovoltaic power forecast model by combining the actual condition of photovoltaic power generation, wherein the model is used for providing a high-level weather forecast service for photovoltaic power generation.
2. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 1, wherein the method comprises the following steps: in step S1, applicability analysis is carried out on earth surface solar incident radiation inverted by satellites by utilizing an artificial intelligence technology, and the specific method comprises the following steps:
constructing a deep learning model aiming at satellite inversion data characteristics, and automatically learning and identifying satellite inversion data characteristics by the model;
in the evaluation process, the deep learning model is used for receiving satellite inversion data as input, outputting an evaluation result of the accuracy and the stability of the satellite inversion data, automatically classifying and clustering the satellite inversion data according to the evaluation result, classifying the data into different categories or clusters, and independently evaluating and analyzing each category or cluster;
for different data types or clusters, the model automatically learns and recognizes different characteristics and modes, automatically evaluates the applicability of the model, and is suitable for data in different geographic areas or in different seasons;
the deep learning model outputs applicability analysis results of satellite inversion data, including accuracy, stability and reliability of the data, and applicability evaluation results for different data categories or clusters.
3. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 2, wherein the method comprises the following steps: the deep learning model adopts a convolutional neural network, a cyclic neural network or a long-term and short-term memory network structure.
4. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 1, wherein the method comprises the following steps: in step S1, the machine learning algorithm evaluates satellite inversion data using the historical data, including the following method steps:
selecting a machine learning algorithm suitable for processing satellite inversion data, wherein the machine learning algorithm comprises a support vector machine, a random forest and a gradient lifting tree, performing feature engineering on historical data, extracting suitable features and using the features for training and evaluating a model;
selecting representative data from the historical data as a training set for training a machine learning model;
training a machine learning model by utilizing a training set, and evaluating and optimizing the model by adjusting model parameters and adopting a cross verification technology;
evaluating the trained model by using an independent test set, and comparing the accuracy and stability of satellite inversion data predicted by the model and real radiation observation data;
adjusting and optimizing the model according to the evaluation result;
the training set and the test set are updated continually, and the model is trained and evaluated continually using the new data.
5. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 4, wherein the method comprises the following steps: the evaluation index includes, but is not limited to, mean square error, root mean square error, and mean absolute error; the model adjustments and optimizations include, but are not limited to, adding features, changing model structures, and adjusting hyper-parameters.
6. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 1, wherein the method comprises the following steps: in step S2, the method steps for comparing the probability density distribution of the satellite inversion data and the real radiation observation data include:
carrying out data cleaning, missing value filling and outlier processing pretreatment on satellite inversion data and real radiation observation data;
respectively constructing probability density functions by utilizing the preprocessed satellite inversion data and the real radiation observation data;
comparing the constructed probability density functions, analyzing the similarity and the difference of the probability density functions, and calculating the distance, intersection and union between two PDFs, wherein the distance, intersection and union are used for evaluating the errors and the differences of satellite inversion data and real radiation observation data;
determining systematic errors between satellite inversion data and real radiation observation data according to the comparison result;
performing systematic error correction on satellite inversion data according to the magnitude and the direction of the systematic error;
and after systematic error correction is carried out, comparing probability density distribution of satellite inversion data and real radiation observation data, and evaluating correction effect.
7. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 6, wherein the method comprises the following steps: the systematic error correction method includes, but is not limited to, linear regression correction of data and correction using interpolation methods.
8. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 1, wherein the method comprises the following steps: in step S3, the specific method for statistical test includes:
average comparison: comparing the average value of the forecast value and the real observed value of each forecast mode, and calculating absolute error and relative error indexes to measure the accuracy of forecast;
variance comparison: comparing the variance of the forecast values of each forecast mode with the variance of the real observed values, wherein a smaller variance indicates that the forecast values are relatively stable, and a larger variance indicates that the forecast values fluctuate more;
correlation coefficient comparison: calculating a correlation coefficient between the forecast value and the real observed value, wherein the closer the correlation coefficient is to 1, the more relevant the forecast mode and the real observed value are;
regression analysis: establishing a regression equation between the prediction mode and the real observed value through regression analysis;
the specific method for graph comparison comprises the following steps:
timing diagram: the forecast values and the real observed values of the various forecast modes are drawn into a time chart, and the forecast quality is evaluated by observing the trend and fluctuation conditions of the time chart.
Scatter plot: and drawing the forecast values and the real observed values of the forecast modes into a scatter diagram, and evaluating the accuracy and the stability of the forecast modes by observing the distribution condition of the scatter diagram.
Box line diagram: and drawing the forecasting values of the forecasting modes into a box diagram, and evaluating the forecasting stability by observing the box body and abnormal value conditions of the box diagram.
9. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 1, wherein the method comprises the following steps: in step S4, the specific method steps for constructing the photovoltaic power forecast model by combining photovoltaic power generation include the following steps:
incorporating more influencing factors into the model including geographic location, climate characteristics, seasonal variations, and weather systems;
dynamically adjusting model parameters according to real-time satellite inversion data and multi-mode forecast data by using a dynamic optimization algorithm;
introducing uncertainty factors of satellite inversion data errors, multimode forecast data errors, geographic positions and climate characteristic changes, and processing and analyzing the uncertainty factors by a probability statistical method;
and the model calculation and storage are distributed to a plurality of calculation nodes by adopting a distributed architecture, so that parallel calculation and rapid training of the model are realized.
10. The ultra-short term prediction method based on satellite inversion radiation data technology according to claim 9, wherein: the more influencing factors also comprise wind speed, wind direction, air pressure, temperature, humidity, solar radiation data, equipment type, equipment performance parameters, equipment running state, illumination time, solar altitude angle, data source, data accuracy and data reliability.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291774.XA CN117194957A (en) | 2023-10-08 | 2023-10-08 | Ultra-short-term prediction method based on satellite inversion radiation data technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311291774.XA CN117194957A (en) | 2023-10-08 | 2023-10-08 | Ultra-short-term prediction method based on satellite inversion radiation data technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117194957A true CN117194957A (en) | 2023-12-08 |
Family
ID=88988757
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311291774.XA Pending CN117194957A (en) | 2023-10-08 | 2023-10-08 | Ultra-short-term prediction method based on satellite inversion radiation data technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117194957A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117473877A (en) * | 2023-12-27 | 2024-01-30 | 青岛市生态与农业气象中心(青岛市气候变化中心) | Lightning three-dimensional radiation source position inversion method based on stationary satellite data |
-
2023
- 2023-10-08 CN CN202311291774.XA patent/CN117194957A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117473877A (en) * | 2023-12-27 | 2024-01-30 | 青岛市生态与农业气象中心(青岛市气候变化中心) | Lightning three-dimensional radiation source position inversion method based on stationary satellite data |
| CN117473877B (en) * | 2023-12-27 | 2024-03-22 | 青岛市生态与农业气象中心(青岛市气候变化中心) | Lightning three-dimensional radiation source position inversion method based on stationary satellite data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Photovoltaic power forecasting with a hybrid deep learning approach | |
| Al-Dahidi et al. | Assessment of artificial neural networks learning algorithms and training datasets for solar photovoltaic power production prediction | |
| CN108921339B (en) | Quantile regression-based photovoltaic power interval prediction method for genetic support vector machine | |
| CN110929953A (en) | Photovoltaic power station ultra-short term output prediction method based on cluster analysis | |
| CN114707688A (en) | Photovoltaic power ultra-short-term prediction method based on satellite cloud chart and space-time neural network | |
| Zhang et al. | Optimal graph structure based short-term solar PV power forecasting method considering surrounding spatio-temporal correlations | |
| CN117194957A (en) | Ultra-short-term prediction method based on satellite inversion radiation data technology | |
| CN114510513A (en) | Short-term meteorological forecast data processing method for ultra-short-term photovoltaic power prediction | |
| Ogbulezie et al. | A review of regression models employed for predicting diffuse solar radiation in North-Western Africa | |
| CN115081557A (en) | Night aerosol optical thickness estimation method and system based on ground monitoring data | |
| CN114330934A (en) | Model parameter self-adaptive GRU new energy short-term power generation power prediction method | |
| He et al. | A power forecasting approach for PV plant based on irradiance index and LSTM | |
| CN115759389A (en) | Day-ahead photovoltaic power prediction method based on weather type similar day combination strategy | |
| Bai et al. | Evolutionary algorithm-based error parameterization methods for data assimilation | |
| CN116451821A (en) | Deep learning-based large-area new energy generation power prediction method and system | |
| Omar et al. | Seasonal clustering forecasting technique for intelligent hourly solar irradiance systems | |
| Yang et al. | Kalman filter photovoltaic power prediction model based on forecasting experience | |
| CN114626640A (en) | Natural gas load prediction method and system based on characteristic engineering and LSTM neural network | |
| Liu et al. | Research of photovoltaic power forecasting based on big data and mRMR feature reduction | |
| CN116565840A (en) | High-precision wind speed soft measurement method for wind power prediction of wind power plant | |
| CN116663393A (en) | Random forest-based power distribution network continuous high-temperature fault risk level prediction method | |
| CN116029419A (en) | Deep learning-based long-term new energy daily average generation power prediction method and system | |
| Yan et al. | Two-phase neural network model for pollution concentrations forecasting | |
| Hu et al. | Grain yield predict based on GRA-AdaBoost-SVR model | |
| Mazrooei et al. | Utilizing probabilistic downscaling methods to develop streamflow forecasts from climate forecasts |
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 |