CN112684519A - Weather forecasting method and device, computer equipment and storage medium - Google Patents
Weather forecasting method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN112684519A CN112684519A CN202011615078.6A CN202011615078A CN112684519A CN 112684519 A CN112684519 A CN 112684519A CN 202011615078 A CN202011615078 A CN 202011615078A CN 112684519 A CN112684519 A CN 112684519A
- Authority
- CN
- China
- Prior art keywords
- latitude
- longitude
- predicted
- data
- model
- 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.)
- Granted
Links
- 238000013277 forecasting method Methods 0.000 title claims abstract description 14
- 238000000034 method Methods 0.000 claims abstract description 79
- 238000012549 training Methods 0.000 claims abstract description 57
- 238000005070 sampling Methods 0.000 claims abstract description 21
- 238000012545 processing Methods 0.000 claims description 13
- 230000002093 peripheral effect Effects 0.000 claims description 11
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical compound OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 description 9
- 230000008569 process Effects 0.000 description 9
- 238000010801 machine learning Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 230000005855 radiation Effects 0.000 description 3
- 101100135611 Arabidopsis thaliana PAP12 gene Proteins 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000763 evoking effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000013441 quality evaluation Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a weather forecasting method, a weather forecasting device, computer equipment and a storage medium, belonging to the technical field of weather forecasting, wherein the method comprises the steps of receiving the forecasting time and the forecasting longitude and latitude input by a user, and calculating earth-day relationship information and earth-month related information according to the forecasting time and the forecasting longitude and latitude; searching corresponding topographic features according to the predicted longitude and latitude; sampling data according to the data distribution to be used as a training data set of the model; introducing the land-day relationship information, the land-month related information and the landform and landform characteristics into a weather forecast model as characteristic variables; performing model hyper-parameter search on the characteristic variables by using a parameter search tool to obtain optimal hyper-parameters; performing model training according to the optimal hyper-parameter and the training data set to obtain a prediction model; receiving real-time weather forecast information as prediction data; and using the prediction model and the prediction data to predict weather forecast.
Description
Technical Field
The present invention relates to the field of weather prediction, and in particular, to a method and an apparatus for weather prediction, a computer device, and a storage medium.
Background
The meteorological model correction refers to a process of correcting the existing medium-long term weather forecast data by mining the modes and rules in the historical observation data.
The current meteorological model correction method mainly comprises two methods:
1) correction using ANO method
ANO (historical data based model leveling integral correction) corrects meteorological data by taking into account the characteristics of atmospheric physical motion and using numerical methods to approximate solutions. The method comprehensively considers the internal power and process constraint of atmospheric motion, the forcing action of external sources such as surface (sea surface) power, thermal process, solar radiation and the like, and partially eliminates the mode system error by utilizing historical observation data and the idea of weather disturbance distance integral, so that the forecast field is closer to reality.
2) Correction using MOS method
Mos (model Output statistics) is widely used as an offline statistical correction method (Glahn and Lowry,1972), and the principle is to directly use the quality evaluation of a historical numerical forecast product to establish a statistical forecast relationship with the local weather element actual measurement values at the same time, and further use a real-time numerical forecast product and the statistical relationship to make local weather forecast. The error correction statistical model is a mode output statistical method, which is called MOS method for short. In the MOS method, the correctness of data needs to be confirmed, the forecasting factors and the number need to be adjusted, and the statistical forecasting equation is continuously checked and improved to improve the forecasting quality.
The first mode correction method is low in accuracy rate depending on massive data analysis and manual experience, and a formula generated by a mode needs to be corrected regularly;
the second machine learning method needs a large amount of historical data support, so that the demand of the model on computing resources is huge;
due to the constraint conditions, the correction of the high-resolution meteorological data at the grid level cannot be completed quickly and efficiently.
Disclosure of Invention
The embodiment of the invention provides a weather forecasting method, a weather forecasting device, computer equipment and a storage medium. The following presents a simplified summary in order to provide a basic understanding of some aspects of the disclosed embodiments. This summary is not an extensive overview and is intended to neither identify key/critical elements nor delineate the scope of such embodiments. Its sole purpose is to present some concepts in a simplified form as a prelude to the more detailed description that is presented later.
According to an embodiment of the present invention, there is provided a weather forecasting method including:
receiving the predicted time input by a user and the corresponding predicted longitude and latitude;
respectively calculating earth-day relationship information and earth-month related information according to the predicted time and the predicted longitude and latitude, and searching corresponding topographic features according to the predicted longitude and latitude;
performing data sampling on historical contemporaneous data according to data distribution to serve as a training data set;
taking the land-day relationship information, the land-month related information and the landform and landform characteristics as characteristic variables, and performing model hyper-parameter search on the characteristic variables by using a parameter search tool to obtain optimal hyper-parameters;
performing model training according to the optimal hyper-parameter and the training data set to obtain a prediction model;
receiving real-time weather forecast information as prediction data;
and predicting weather forecast by using the prediction model and the prediction data, and outputting the obtained weather forecast.
Further, the air conditioner is provided with a fan,
the calculating the earth-day relationship information according to the predicted time and the predicted longitude and latitude comprises calculating a solar azimuth angle, an earth-day distance and solar irradiance according to the predicted time and the predicted longitude and latitude respectively.
Further, the air conditioner is provided with a fan,
and the step of calculating the earth-moon related information according to the predicted time and the predicted longitude and latitude comprises calculating a moon azimuth angle and calculating an earth-moon distance.
Further, the air conditioner is provided with a fan,
searching for corresponding topographic features according to the predicted longitude and latitude, wherein the steps of searching for the altitude of a preset grid point and the altitude of the preset grid point according to the current point peripheral distance of the longitude and latitude to obtain the topographic features are included; according to the topographic features, carrying out binarization to generate the negative/positive features of the current point; converting the topographic features into geomorphic features according to the national land classification.
Further, the peripheral distance ranges from 0.1 ° or 0.01 °.
And the step of inquiring the altitude of the preset grid point at the peripheral distance of the current point where the longitude and the latitude are located according to the longitude and the latitude comprises the step of inquiring the altitudes of 4 and 12 grid points at the periphery.
Further, the weather forecasting method is characterized in that,
the generating of the yin/yang characteristics of the current point comprises: judging the characteristics of the male surface and the female surface by inquiring the altitude of 4 grid points around the female surface, Af1<Af4Being the yin side, Af1>Af4Is a sunny side; judging the characteristics of the male surface and the female surface by inquiring the altitude of 12 grid points on the periphery, At3<At10Being the yin side, At3>At10Is a sunny side;
wherein,
Af1: in the method for inquiring the altitudes of 4 surrounding grid points, the altitudes of the point positions right above the current longitude and latitude;
Af4: in the method for inquiring the altitudes of 4 surrounding grid points, the altitudes of the point positions right below the current longitude and latitude;
At3: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of a point position right above the current longitude and latitude is inquired;
At10: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of the point position right below the current longitude and latitude is obtained.
Further, the air conditioner is provided with a fan,
the data sampling of the historical contemporaneous data according to the data distribution comprises the following steps: and (3) proportionally selecting 1-3 months of historical contemporaneous forecast data by using a rejection sampling method for the data which obey normal distribution as a training data set of the model.
Furthermore, a parameter searching tool adopted by the model hyper-parameter searching is a tree-structure Parzen estimation method. Further, the model training method is to train the model by using a limit gradient lifting tree framework.
According to a second aspect of the present invention, there is provided a weather forecasting apparatus comprising,
the first receiving module is used for receiving the predicted time input by the user and the corresponding predicted longitude and latitude;
the historical data processing module is connected with the first receiving module and used for respectively calculating earth-day relationship information and earth-month related information according to the predicted time and the predicted longitude and latitude and searching corresponding topographic features according to the predicted longitude and latitude;
the sampling module is used for sampling data of historical contemporaneous data according to data distribution to serve as a training data set;
the super-parameter searching module is connected with the historical data processing module and is used for performing model super-parameter searching on the characteristic variables by using a parameter searching tool by taking the land-day relationship information, the land-month related information and the topographic features as characteristic variables to obtain optimal super-parameters;
the model training module is respectively connected with the sampling module and the hyper-parameter searching module and is used for carrying out model training according to the optimal hyper-parameter and the training data set to obtain a prediction model;
the second receiving module is used for receiving the real-time weather forecast message as prediction data;
and the prediction module is respectively connected with the model training module and the second receiving module and is used for predicting weather forecast by using the prediction model and the prediction data and outputting the obtained weather forecast.
According to a third aspect of the present invention, there is provided a computer apparatus comprising: a processor and a memory, and a computer program stored in the memory and executable in the processor, the processor executing the program to implement the steps of the method as claimed in any one of the above.
According to a fourth aspect of the present invention, there is provided a computer readable storage medium comprising a computer program stored therein, the program being for implementing a method as claimed in any one of the preceding claims.
The technical scheme provided by the embodiment of the invention has the following beneficial effects:
according to the embodiment, the invention provides a weather forecasting method, which is characterized in that the geographical and daily relationship information and the geographical and monthly related information are calculated according to the predicted time and the predicted longitude and latitude, the topographic and geomorphic information corresponding to the predicted longitude and latitude is searched and input into a machine learning model for training, and the restriction of the model on the requirement of a large amount of historical data is relieved. Even if a small amount of historical data is used for training, the accuracy of the traditional machine learning model can be achieved or even exceeded. The new method combines the space and geophysical information with the machine learning model for the first time, so that the time and the cost of model training are greatly reduced, and a new thought and solution are developed for mode prediction.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and together with the description, serve to explain the principles of the invention.
FIG. 1 is a flow diagram illustrating a method of weather forecasting according to an exemplary embodiment;
FIG. 2 is a diagram illustrating a method of selecting four grid points on the perimeter, according to an exemplary embodiment;
FIG. 3 is a diagram illustrating a method for selecting twelve grid points on the periphery according to an exemplary embodiment
FIG. 4 is a flow diagram illustrating a method of model training in accordance with an exemplary embodiment;
FIG. 5 is a graphical illustration of a comparison of accuracy within 2 degrees of model training in three ways in accordance with an exemplary embodiment;
FIG. 6 is a diagram illustrating the impact of a physical environment on accuracy gain in accordance with an exemplary embodiment;
FIG. 7 is a graph illustrating a year 2020, month 8 temperature accuracy, according to an exemplary embodiment;
FIG. 8 is a graph illustrating a temperature accuracy of 9 months in 2020, according to an exemplary embodiment;
FIG. 9 is a graph illustrating a 10 month 2020 temperature accuracy, according to an exemplary embodiment;
FIG. 10 is a graph illustrating a 2019 month 11 temperature accuracy, according to an exemplary embodiment.
Detailed Description
The following description and the drawings sufficiently illustrate specific embodiments of the invention to enable those skilled in the art to practice them. The examples merely typify possible variations. Individual components and functions are optional unless explicitly required, and the sequence of operations may vary. Portions and features of some embodiments may be included in or substituted for those of others. The scope of embodiments of the invention encompasses the full ambit of the claims, as well as all available equivalents of the claims. Embodiments may be referred to herein, individually or collectively, by the term "invention" merely for convenience and without intending to voluntarily limit the scope of this application to any single invention or inventive concept if more than one is in fact disclosed. Herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed. The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the structures, products and the like disclosed by the embodiments, the description is relatively simple because the structures, the products and the like correspond to the parts disclosed by the embodiments, and the relevant parts can be just described by referring to the method part.
It should be noted that: the relative arrangement of the components and steps, the numerical expressions, and numerical values set forth in these embodiments do not limit the scope of the present disclosure unless specifically stated otherwise.
It should be noted that although the various steps of the methods of the present disclosure are depicted in the drawings in a particular order, this does not require or imply that these steps must be performed in this particular order, or that all of the depicted steps must be performed, to achieve desirable results. Additionally or alternatively, certain steps may be omitted, multiple steps combined into one step execution, and/or one step broken down into multiple step executions, etc.
The invention is further described with reference to the following figures and examples:
in a first aspect of the present invention, a weather forecasting method as shown in fig. 1 includes:
s1: receiving the predicted time input by a user and the corresponding predicted longitude and latitude;
s2: respectively calculating earth-day relationship information and earth-month related information according to the predicted time and the predicted longitude and latitude, and searching corresponding topographic features according to the predicted longitude and latitude;
s3: performing data sampling on historical contemporaneous data according to data distribution to serve as a training data set;
s4: taking the land-day relationship information, the land-month related information and the topographic features as characteristic variables, and performing model hyper-parameter search on the characteristic variables by using a parameter search tool to obtain optimal hyper-parameters;
s5: performing model training according to the optimal hyper-parameter and the training data set to obtain a prediction model;
s6: receiving real-time weather forecast information as prediction data;
s7: and using the prediction model and the prediction data to predict weather forecast and outputting the obtained weather forecast.
According to the scheme, further, the calculating of the earth-day relationship information according to the predicted time and the predicted longitude and latitude comprises the steps of calculating a solar azimuth angle, calculating an earth-day distance and calculating solar irradiance;
the calculated solar azimuth angle is the sun azimuth angle,
wherein,
φs: the solar azimuth angle;
θs: solar altitude;
h: calculating a time angle of time;
δ: declination of the sun;
phi: a geographic latitude;
the calculated distance between the earth and the day is,
D＝(1-0.01672cos(0.9856(day-4)))×AU
wherein,
d: the distance between the ground and the day;
day: the date to be calculated;
AU: 1 astronomical unit equal to the distance between the near-sun points of the earth;
the calculated solar irradiance is such that,
ID＝SR×(R/D)2
wherein,
ID: solar irradiance;
SR: solar surface radiation;
r: the solar radius;
d: the distance between the ground and the day.
According to the scheme, further, the step of calculating the earth-moon related information according to the predicted time and the longitude and latitude comprises the steps of calculating a moon azimuth angle and calculating an earth-moon distance.
According to the scheme, further, the step of acquiring the topographic and geomorphic characteristics according to the predicted longitude and latitude comprises the step of inquiring and calculating the peripheral distance and the self elevation according to the longitude and latitude so as to obtain the topographic characteristics; according to the topographic features, carrying out binarization to obtain the negative/positive features of the current point; as shown in table 1, the topographic features are converted into topographic features according to the national land classification.
TABLE 1 conversion of topographic features into geomorphic features according to national land classifications
According to the above aspect, further, the peripheral distance range may be equal to or greater than 0.01 ° and equal to or less than 0.1 °. In the present invention, the angle may be 0.1 ° or 0.01 °.
According to the above solution, further, as shown in fig. 2 and fig. 3, the calculating the peripheral distance and the altitude thereof according to the latitude and longitude query includes querying the altitudes of 4 and 12 grid points on the periphery.
According to the above scheme, further, the method for obtaining the yin/yang characteristics of the current point comprises the following steps: judging the characteristics of the yin surface/the yang surface by inquiring the altitude of 4 grid points on the periphery, wherein Af1< Af4 is the yin surface, and Af1> Af4 is the yang surface; judging the characteristics of the yin side/the yang side by inquiring the altitude of 12 grid points on the periphery, wherein At3< At10 is the yin side, and At3> At10 is the yang side;
wherein,
af 1: in the method for inquiring the altitudes of 4 surrounding grid points, the altitudes of the point positions right above the current longitude and latitude;
af 4: in the method for inquiring the altitudes of 4 surrounding grid points, the altitudes of the point positions right below the current longitude and latitude;
at 3: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of a point position right above the current longitude and latitude is inquired;
at 10: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of the point position right below the current longitude and latitude is obtained.
According to the above scheme, further, the method for sampling data according to data distribution includes: assuming that the data are proportional to the normal distribution, selecting 1-3 months of contemporaneous historical forecast data by using a rejection sampling method to serve as a training data set of the model.
The principle and process of the rejection sampling method are as follows:
TABLE 2 rejection sampling principle and procedure
According to the scheme, further, a parameter searching tool adopted by the model hyper-parameter searching is a tree-structure Parzen estimation method. The tree structure Parzen estimation method is a method for searching and optimizing hyper-parameters of a machine learning model, and is a sequential model-based optimization (SMBO). The SMBO method sequentially constructs a model from historical index data to estimate the performance of the hyper-parameters, and then selects new hyper-parameters based on this model.
The principle and process of the tree structure Parzen method are as follows:
TABLE 3 principle and procedure of the Parzen method for tree structure
According to the scheme, further, as shown in fig. 4, the model training method is to train the model by using a limit gradient lifting tree framework.
Training models using extreme gradient lifting tree (XGboost) framework
The principle and process of XGboost training are as follows:
XGboost is an abbreviation of Extreme Grander Boosting Tree, and is used for optimizing and improving Boosting Tree
TABLE 4 development Process of extreme gradient Lift Tree (XGboost) framework
According to the invention, the land-day relationship information and the land-month related information are calculated according to the prediction time and the prediction longitude and latitude, the terrain and landform information corresponding to the prediction longitude and latitude is searched and input to the machine learning model for training, the data according to the model training is enriched, the state of the terrain and landform can be depicted, and the relation of weather changing along with the terrain is reflected. The technical scheme can relieve the restriction of the model on the requirement of a large amount of historical data, and can achieve or even exceed the accuracy of the conventional machine learning model under the condition of using a small amount of historical data for training. The method provided by the invention combines the space and geophysical information with the machine learning model, so that the time and cost of model training are greatly reduced, the working efficiency is improved, and the accuracy of weather forecast is ensured.
Example 1
As shown in fig. 5, by correcting the air temperatures of a plurality of observation stations of 2 thousand in China using the EC data of three years of history in 2017, 2018 and 2019, it was found that using different amounts of altitude data around the stations resulted in different accuracy rates.
Selecting the temperature accuracy rate change within 2 degrees in 6 months in 2019;
12p is the training accuracy of the 12-point model around the station;
4p is the training accuracy of the nearest 4-point model around the station;
4pi is the model training accuracy rate after interpolation of the nearest 4 points around the station.
Geographic information should influence local weather phenomena, so that terrain and landform information is added for model training, and the accuracy and stability of result discovery are further improved.
As shown in fig. 6, the change of weather should be the result of the physical environment influence of the earth, and the accuracy of weather forecast is further improved after the information about the relationship between the earth and the day and the information about the earth and the moon are added to the model.
The data of 2020-09-02 to 2020-09-08 for one week is used for verification,
alti is the gain of model training accuracy after interpolation of 4 nearest neighbors around a station compared with the EC original forecast;
sun is the accuracy gain after the sun azimuth angle, the earth-sun distance and the solar irradiance are increased;
the sun + moon is the accuracy gain after the lunar azimuth angle and the lunar distance are increased on the basis of the solar azimuth angle, the terrestrial distance and the solar irradiance.
As shown in fig. 7, 8, 9 and 10, the robustness of the model was further examined by using less data training, and it was found that the model can achieve more stable accuracy output even when only 3 months of historical data is used.
The model is trained by using data of 3 months at 8, 9 and 10 months in 2019, and verified by using data of 8, 9, 10 and 11 months in 2020,
wherein EC is the original accuracy;
model is the Model prediction accuracy.
According to the experimental results, a new model training method is obtained, the AI model is enabled to automatically simulate the meteorological physical process of the earth, the AI model is regarded as a black box simulator, and the weather forecast result with high accuracy can be quickly obtained by adopting end-to-end input and output design.
In a second aspect of the present invention, a weather forecasting apparatus includes,
a first receiving module 201, configured to receive a predicted time input by a user and a corresponding predicted longitude and latitude;
a historical data processing module 202, connected to the first receiving module 201, configured to respectively calculate information about a geographical-day relationship and information about a geographical-month relationship according to the predicted time and the predicted longitude and latitude, and search for corresponding topographic features according to the predicted longitude and latitude;
in the embodiment of the present invention, the historical data processing module 202 includes a ground-day data processing sub-module, a ground-month data processing sub-module, and a geomorphic data processing sub-module. Wherein,
and the earth-day data processing submodule is used for calculating the solar azimuth angle, the earth-day distance and the solar irradiance according to the predicted time and the predicted longitude and latitude.
The method for calculating the solar azimuth angle comprises the following steps of calculating a solar altitude angle and a time angle of the predicted time according to the predicted time and the predicted longitude and latitude, and calculating the solar azimuth angle according to the solar altitude angle, the time angle of the predicted time, the solar declination and the geographical latitude, wherein the method comprises the following steps:
wherein phi iss: the solar azimuth angle; thetas: solar altitude; h: predicting a time angle of time; δ: declination of the sun; phi: a geographic latitude;
the method for calculating the distance between the earth and the day is that,
D＝(1-0.01672cos(0.9856(day-4)))×AU
wherein, D: the distance between the ground and the day; day: a predicted date; AU: 1 astronomical unit equal to the distance between the near-sun points of the earth; the method for calculating the solar irradiance is that,
ID＝SR×(R/D)2
wherein, ID: solar irradiance; SR: solar surface radiation; r: the solar radius; d: the distance between the ground and the day.
And the earth-moon data processing submodule is used for calculating the moon azimuth angle and calculating the earth-moon distance according to the prediction time and the longitude and latitude.
The landform data processing submodule is used for inquiring and predicting the elevation of the preset grid point of the peripheral distance of the longitude and the latitude and the self elevation according to the predicted longitude and latitude to obtain the landform characteristics; according to the topographic features, carrying out binarization to generate the negative/positive features of the current point; wherein the peripheral distance range may be equal to or greater than 0.01 ° and equal to or less than 0.1 °. A method for generating the features of the internal surface and external surface of current point includes such steps as inquiring the altitude of 4 lattice points around the current point to judge the features of internal surface and external surface, Af1<Af4Being the yin side, Af1>Af4Is a sunny side; judging the characteristics of the male surface and the female surface by inquiring the altitude of 12 grid points on the periphery, At3<At10Being the yin side, At3>At10Is a sunny side;
wherein,
Af1: in the altitude method for inquiring 4 surrounding grid points, the points right above the current longitude and latitudeThe altitude of the bit;
Af4: in the method for inquiring the altitudes of 4 surrounding grid points, the altitudes of the point positions right below the current longitude and latitude;
At3: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of a point position right above the current longitude and latitude is inquired;
At10: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of the point position right below the current longitude and latitude is obtained.
The sampling module 203 is used for performing data sampling on historical contemporaneous data according to data distribution to serve as a training data set;
in the invention, the contemporaneous historical forecast data of 1-3 months is selected as a training data set of the model by proportionally using a rejection sampling method for the data which obey normal distribution.
The hyper-parameter searching module 204 is connected with the historical data processing module 202 and is used for performing model hyper-parameter searching on the characteristic variables by using a parameter searching tool by taking the earth-day relationship information, the earth-month related information and the topographic features as characteristic variables to obtain optimal hyper-parameters;
in the present invention, the hyper-parameter search module 204 is configured to perform model hyper-parameter search on the feature variables by using a tree structure Parzen estimation method with the information of the earth-day relationship, the information of the earth-month relationship, and the topographic features as the feature variables, so as to obtain the optimal hyper-parameters.
The model training module 205 is connected to the sampling module 203 and the hyper-parameter searching module 204, and configured to perform model training according to the optimal hyper-parameter and the training data set to obtain a prediction model;
in the present invention, the model training module 205 is configured to perform model training using an extreme gradient lifting tree framework training model according to the optimal hyper-parameter and the training data set, so as to obtain a prediction model.
A second receiving module 206, configured to receive the real-time weather forecast message as prediction data;
and the prediction module 207 is connected to the model training module 205 and the second receiving module 206, and configured to perform weather forecast prediction by using the prediction model and the prediction data, and output the obtained weather forecast.
According to a third aspect of the invention, a computer device comprises: a processor and a memory, and a computer program stored in the memory and executable in the processor, the processor executing the steps of the program to implement a weather forecasting method, wherein the weather forecasting method comprises,
receiving the predicted time input by a user and the corresponding predicted longitude and latitude;
respectively calculating earth-day relationship information and earth-month related information according to the predicted time and the predicted longitude and latitude, and searching corresponding topographic features according to the predicted longitude and latitude;
performing data sampling on historical contemporaneous data according to data distribution to serve as a training data set;
taking the land-day relationship information, the land-month related information and the landform and landform characteristics as characteristic variables, and performing model hyper-parameter search on the characteristic variables by using a parameter search tool to obtain optimal hyper-parameters;
performing model training according to the optimal hyper-parameter and the training data set to obtain a prediction model;
receiving real-time weather forecast information as prediction data;
and predicting weather forecast by using the prediction model and the prediction data, and outputting the obtained weather forecast.
According to a fourth aspect of the present invention, a computer-readable storage medium includes a computer program stored therein for implementing a method of weather forecasting.
It is to be understood that the present invention is not limited to the procedures and structures described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the invention is limited only by the appended claims.
Claims (10)
1. A weather forecasting method, comprising:
receiving the predicted time input by a user and the corresponding predicted longitude and latitude;
respectively calculating earth-day relationship information and earth-month related information according to the predicted time and the predicted longitude and latitude, and searching corresponding topographic features according to the predicted longitude and latitude;
performing data sampling on historical contemporaneous data according to data distribution to serve as a training data set;
taking the land-day relationship information, the land-month related information and the landform and landform characteristics as characteristic variables, and performing model hyper-parameter search on the characteristic variables by using a parameter search tool to obtain optimal hyper-parameters;
performing model training according to the optimal hyper-parameter and the training data set to obtain a prediction model;
receiving real-time weather forecast information as prediction data;
and predicting weather forecast by using the prediction model and the prediction data, and outputting the obtained weather forecast.
2. Weather forecast method according to claim 1,
the calculating the earth-day relationship information according to the predicted time and the predicted longitude and latitude comprises calculating a solar azimuth angle, an earth-day distance and solar irradiance according to the predicted time and the predicted longitude and latitude respectively.
3. Weather forecast method according to claim 1,
and the step of calculating the earth-moon related information according to the predicted time and the predicted longitude and latitude comprises calculating a moon azimuth angle and calculating an earth-moon distance.
4. Weather forecast method according to claim 1,
searching for corresponding topographic features according to the predicted longitude and latitude, wherein the steps of searching for the altitude of a preset grid point and the altitude of the preset grid point according to the peripheral distance of the current point where the current point is located according to the predicted longitude and latitude are carried out to obtain the topographic features; according to the topographic features, carrying out binarization to generate the negative/positive features of the current point; converting the topographic features into geomorphic features according to the national land classification.
5. The weather forecasting method according to claim 4, wherein the peripheral distance range is 0.1 ° or less, and 0.01 ° or more.
6. Weather forecasting method according to claim 4,
and the step of inquiring the altitude of the preset grid point at the peripheral distance of the current point where the longitude and the latitude are located according to the longitude and the latitude comprises the step of inquiring the altitudes of 4 and 12 grid points at the periphery.
7. Weather forecasting method according to claim 6,
the generating of the yin/yang characteristics of the current point comprises: judging the characteristics of the male surface and the female surface by inquiring the altitude of 4 grid points around the female surface, Af1<Af4Being the yin side, Af1>Af4Is a sunny side; judging the characteristics of the male surface and the female surface by inquiring the altitude of 12 grid points on the periphery, At3<At10Being the yin side, At3>At10Is a sunny side;
wherein,
Af1: in the method for inquiring the altitudes of 4 surrounding grid points, the altitudes of the point positions right above the current longitude and latitude;
Af4: in the method for inquiring the altitudes of 4 surrounding grid points, the altitudes of the point positions right below the current longitude and latitude;
At3: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of a point position right above the current longitude and latitude is inquired;
At10: in the method for inquiring the altitude of 12 surrounding grid points, the altitude of the point position right below the current longitude and latitude is obtained.
8. A weather forecast apparatus, comprising,
the first receiving module is used for receiving the predicted time input by the user and the corresponding predicted longitude and latitude;
the historical data processing module is connected with the first receiving module and used for respectively calculating earth-day relationship information and earth-month related information according to the predicted time and the predicted longitude and latitude and searching corresponding topographic features according to the predicted longitude and latitude;
the sampling module is used for sampling data of historical contemporaneous data according to data distribution to serve as a training data set;
the super-parameter searching module is connected with the historical data processing module and is used for performing model super-parameter searching on the characteristic variables by using a parameter searching tool by taking the land-day relationship information, the land-month related information and the topographic features as characteristic variables to obtain optimal super-parameters;
the model training module is respectively connected with the sampling module and the hyper-parameter searching module and is used for carrying out model training according to the optimal hyper-parameter and the training data set to obtain a prediction model;
the second receiving module is used for receiving the real-time weather forecast message as prediction data;
and the prediction module is respectively connected with the model training module and the second receiving module and is used for predicting weather forecast by using the prediction model and the prediction data and outputting the obtained weather forecast.
9. A computer device, comprising: processor and memory, and a computer program stored in the memory and executable in the processor, wherein execution of the program by the processor enables implementation of the steps of the method of any one of claims 1 to 7.
10. A computer-readable storage medium, comprising a computer program stored in the computer-readable storage medium, the program being for implementing the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011615078.6A CN112684519B (en) | 2020-12-30 | 2020-12-30 | Weather forecasting method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011615078.6A CN112684519B (en) | 2020-12-30 | 2020-12-30 | Weather forecasting method and device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112684519A true CN112684519A (en) | 2021-04-20 |
| CN112684519B CN112684519B (en) | 2022-07-05 |
Family
ID=75455407
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011615078.6A Active CN112684519B (en) | 2020-12-30 | 2020-12-30 | Weather forecasting method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112684519B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114545528A (en) * | 2022-03-09 | 2022-05-27 | 北京墨迹风云科技股份有限公司 | Meteorological numerical model element forecasting and post-correcting method and device based on machine learning |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1437732A (en) * | 2000-04-18 | 2003-08-20 | 卡梅尔系统有限责任公司 | Space weather prediction system and method |
| US20170131435A1 (en) * | 2015-11-05 | 2017-05-11 | Heriot-Watt University | Localized weather prediction |
| CN106709587A (en) * | 2015-11-13 | 2017-05-24 | 中国电力科学研究院 | Direct radiation prediction method based on conventional weather forecast |
| CN109239808A (en) * | 2018-08-01 | 2019-01-18 | 平安科技（深圳）有限公司 | Weather forecast method, device, computer equipment and storage medium |
| CN110298498A (en) * | 2019-06-14 | 2019-10-01 | 水利部交通运输部国家能源局南京水利科学研究院 | A kind of Flood of small drainage area forecasting procedure for establishing data-driven with support vector machines |
-
2020
- 2020-12-30 CN CN202011615078.6A patent/CN112684519B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1437732A (en) * | 2000-04-18 | 2003-08-20 | 卡梅尔系统有限责任公司 | Space weather prediction system and method |
| US20170131435A1 (en) * | 2015-11-05 | 2017-05-11 | Heriot-Watt University | Localized weather prediction |
| CN106709587A (en) * | 2015-11-13 | 2017-05-24 | 中国电力科学研究院 | Direct radiation prediction method based on conventional weather forecast |
| CN109239808A (en) * | 2018-08-01 | 2019-01-18 | 平安科技（深圳）有限公司 | Weather forecast method, device, computer equipment and storage medium |
| CN110298498A (en) * | 2019-06-14 | 2019-10-01 | 水利部交通运输部国家能源局南京水利科学研究院 | A kind of Flood of small drainage area forecasting procedure for establishing data-driven with support vector machines |
Non-Patent Citations (1)
| Title |
|---|
| 王水 等: "《空间天气研究的主要科学问题》", 《中国科学技术大学学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114545528A (en) * | 2022-03-09 | 2022-05-27 | 北京墨迹风云科技股份有限公司 | Meteorological numerical model element forecasting and post-correcting method and device based on machine learning |
| CN114545528B (en) * | 2022-03-09 | 2024-02-06 | 北京墨迹风云科技股份有限公司 | Machine learning-based correction method and device after meteorological numerical mode element forecast |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112684519B (en) | 2022-07-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sun et al. | Estimating the spatial distribution of solar photovoltaic power generation potential on different types of rural rooftops using a deep learning network applied to satellite images | |
| Carneiro et al. | Review on photovoltaic power and solar resource forecasting: current status and trends | |
| Zhang et al. | Future projections of offshore wind energy resources in China using CMIP6 simulations and a deep learning-based downscaling method | |
| US10509868B2 (en) | Operating a solar power generating system | |
| CN113222019B (en) | Meteorological forecast data processing method, device and equipment for transmission line tower | |
| CN115049171B (en) | Photovoltaic power prediction method and system based on feature migration | |
| Isaksson et al. | Solar power forecasting with machine learning techniques | |
| CN112580982A (en) | Ecological protection red line implementation assessment based on multi-temporal remote sensing and CASA model | |
| CN115586591B (en) | Automatic forecasting method and device for storm surge | |
| CN115189356B (en) | Wind and light power prediction correction method and device based on atmospheric circulation stable state | |
| CN112684519B (en) | Weather forecasting method and device, computer equipment and storage medium | |
| CN115544889A (en) | Numerical mode precipitation deviation correction method based on deep learning | |
| CN116029419A (en) | Deep learning-based long-term new energy daily average generation power prediction method and system | |
| CN113984198B (en) | Shortwave radiation prediction method and system based on convolutional neural network | |
| Achbab et al. | Developing and applying a GIS-Fuzzy AHP assisted approach to locating a hybrid renewable energy system with high potential: Case of Dakhla region–Morocco | |
| Krantz et al. | Memorandum on evaluating global climate models for studying regional climate change in California | |
| CN114819264A (en) | Photovoltaic power station irradiance ultra-short term prediction method based on space-time dependence and storage medium | |
| Bawazir et al. | Optimum PV distributed generation based on grid and geographical area: A case study of Aden governorate, Yemen | |
| CN117421871A (en) | Offshore wind power potential evaluation method and device and computer equipment | |
| CN114676893B (en) | Photovoltaic power station solar irradiance short-term prediction method based on optimal graph structure and storage medium | |
| Graabak et al. | Developing a wind and solar power data model for Europe with high spatial-temporal resolution | |
| Lee et al. | 24-Hour photovoltaic generation forecasting using combined very-short-term and short-term multivariate time series model | |
| Ganvir et al. | Prediction of Global Horizontal Irradiance based on eXplainable Artificial Intelligence | |
| Hayhoe et al. | STAR-ESDM: A Generalizable Approach to Generating High-Resolution Climate Projections through Signal Decomposition | |
| Yagli et al. | Using combinational methods for forecast improvement in PV power plants |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Shao Kejia Inventor after: Wang Tian Inventor after: Huang Yaohai Inventor before: Shao Kejia Inventor before: Wang Yi Inventor before: Huang Yaohai |