CN113341481A - Method and device for determining weather forecast result - Google Patents
Method and device for determining weather forecast result Download PDFInfo
- Publication number
- CN113341481A CN113341481A CN202110751609.2A CN202110751609A CN113341481A CN 113341481 A CN113341481 A CN 113341481A CN 202110751609 A CN202110751609 A CN 202110751609A CN 113341481 A CN113341481 A CN 113341481A
- Authority
- CN
- China
- Prior art keywords
- forecast
- meteorological
- sub
- training
- weather
- 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
- 238000000034 method Methods 0.000 title claims description 44
- 238000012549 training Methods 0.000 claims description 168
- 238000001556 precipitation Methods 0.000 claims description 49
- 238000012216 screening Methods 0.000 claims description 29
- 238000003062 neural network model Methods 0.000 claims description 15
- 238000009826 distribution Methods 0.000 claims description 7
- 230000004044 response Effects 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 4
- 239000002609 medium Substances 0.000 description 25
- 230000006870 function Effects 0.000 description 9
- 238000004422 calculation algorithm Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 8
- 230000004913 activation Effects 0.000 description 7
- 230000003287 optical effect Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 238000003064 k means clustering Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000013136 deep learning model Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 239000013307 optical fiber Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 230000005923 long-lasting effect Effects 0.000 description 1
- 239000006837 my medium Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000000704 physical effect Effects 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 238000010561 standard procedure Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Chemical compound O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01W—METEOROLOGY
- G01W1/00—Meteorology
- G01W1/10—Devices for predicting weather conditions
Landscapes
- Environmental & Geological Engineering (AREA)
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Atmospheric Sciences (AREA)
- Biodiversity & Conservation Biology (AREA)
- Ecology (AREA)
- Environmental Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
After the weather forecast sets are received, the ensemble forecasts in the weather forecast sets can be classified based on forecast characteristics indicated by the ensemble forecasts in the weather forecast sets, at least one sub-weather forecast set is obtained, each sub-weather forecast set comprises a plurality of ensemble forecasts with the same or similar indication forecast characteristics, a target forecast member is screened out from each sub-weather forecast set, and then the weather forecast result is determined according to the determined target forecast member. That is, after the weather forecast sets are received, the weather forecast sets are classified, forecast characteristics indicated by the set forecasts in each class are close, then a target forecast member which is most likely to be close to an actual situation is screened out from each class, and then a weather forecast result is obtained according to the screened target forecast member, so that the obtained weather forecast result is more accurate.
Description
Technical Field
The disclosed example relates to the technical field of weather forecasting, in particular to a method and a device for determining weather forecasting results.
Background
The urban heat island effect is caused by the accelerated urbanization process, the physical property of the underlying surface is changed, the surface temperature difference and the vertical speed close to the ground are increased, and the method is favorable for water vapor rising, so that the frequent occurrence of extreme rainfall weather and the enhancement of rainfall are caused. Compared with long-lasting strong precipitation with longer duration and weaker strength, the short-lasting strong precipitation event has short duration and concentrated precipitation, is more likely to cause disasters such as waterlogging and ponding, and the short-imminent precipitation forecast is the key for making advance defense decision when industrial users face sudden strong precipitation meteorological disasters.
With the development of scientific technology, objective description of spatial matching characteristics such as scale, structure, position and the like of live and forecasted strong precipitation events can be given by applying an object recognition spatial inspection technology, so that short-term precipitation forecast can be improved, and the number of meteorological stations is large. Thus, one can calculate the weather data observed by each weather station to obtain ensemble forecasts, each of which may include multiple forecast members.
Disclosure of Invention
This disclosure is provided to introduce concepts in a simplified form that are further described below in the detailed description. This disclosure is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
The embodiment of the disclosure provides a method for determining a weather forecast result, which can screen a large amount of forecast data to obtain target forecast data, and can obtain a weather forecast result according to the target forecast data, so that weather forecast is more accurate.
In a first aspect, an embodiment of the present disclosure provides a method for determining a weather forecast result, including: in response to receiving a weather forecast set, classifying the weather forecast set based on forecast characteristics indicated by each ensemble forecast in the weather forecast set to obtain at least one set of sub-weather forecast sets, wherein the forecast characteristics include: precipitation or wind speed, each sub-meteorological prediction ensemble comprising at least one ensemble prediction; and screening a target forecast member from each sub-meteorological forecast set, and determining a meteorological forecast result according to the screened target forecast member.
With reference to the embodiments of the first aspect, in some embodiments, the screening out a target forecast member from each sub-weather forecast set, and determining a weather forecast result according to the screened out target forecast member includes: and screening out a target forecast member from each sub-meteorological forecast set by utilizing the forecast model group, and determining a meteorological forecast result based on the screened target forecast member by utilizing the forecast model group.
With reference to the embodiments of the first aspect, in some embodiments, the screening out a target forecast member from each of the sub-meteorological forecast sets includes: and screening out a target forecast member from each sub-forecast set according to the forecast characteristic distribution of the forecast members in each sub-meteorological forecast set.
In combination with an embodiment of the first aspect, in some embodiments, the set of prediction models is obtained by: obtaining training samples, wherein the training samples comprise: forecast meteorological data for training and live meteorological data for training; classifying the obtained training samples to obtain N training sample sets, wherein N is a positive integer; and (4) obtaining N forecasting model groups by utilizing each training sample set to preset the initial neural network model.
In combination with an embodiment of the first aspect, in some embodiments, the training of the meteorological features of the sample comprises: a feature for indicating precipitation; and classifying the obtained training samples to obtain N training sample sets, including: and classifying the obtained training samples according to the precipitation indicated by each training sample to obtain N training sample sets.
In combination with an embodiment of the first aspect, in some embodiments, the training of the meteorological features of the sample comprises: the method for classifying the acquired training samples to obtain N training sample sets includes: and classifying the obtained training samples according to the wind speed indicated by each training sample to obtain N training sample sets.
With reference to the embodiments of the first aspect, in some embodiments, obtaining the N prediction model sets by using each training sample set pair preset initial neural network model includes: screening training samples in a preset proportion in each training sample set; and training a preset initial neural network model according to the training samples screened from the training sample sets to obtain the N forecasting model groups.
With reference to embodiments of the first aspect, in some embodiments, the method further includes: and verifying each forecasting model group by using the training samples which are not screened out from each training sample set.
In a second aspect, an embodiment of the present disclosure provides an apparatus for determining a weather forecast result, including: the classification unit is used for responding to the received weather forecast sets, classifying the weather forecast sets based on forecast characteristics indicated by all the forecast in the weather forecast sets, and obtaining at least one group of sub-weather forecast sets, wherein the forecast characteristics comprise: precipitation or wind speed, each sub-meteorological prediction ensemble comprising at least one ensemble prediction; and the screening unit is used for screening out a target forecast member from each sub-meteorological forecast set and determining a meteorological forecast result according to the screened target forecast member.
With reference to the embodiments of the second aspect, in some embodiments, the screening unit is further specifically configured to: and screening out a target forecast member from each sub-meteorological forecast set by utilizing the forecast model group, and determining a meteorological forecast result based on the screened target forecast member by utilizing the forecast model group.
With reference to the embodiments of the second aspect, in some embodiments, the screening unit is further specifically configured to: and screening out a target forecast member from each sub-meteorological forecast set according to the forecast characteristic distribution of the forecast members in each sub-meteorological forecast set.
With reference to the embodiments of the second aspect, in some embodiments, the apparatus further includes a determining unit, configured to obtain the set of prediction models by:
obtaining training samples, wherein the training samples comprise: forecast meteorological data for training and live meteorological data for training; classifying the obtained training samples to obtain N training sample sets, wherein N is a positive integer; and obtaining the N forecasting model groups by utilizing each training sample set to preset an initial neural network model.
With reference to the embodiment of the second aspect, in some embodiments, the determining unit is further specifically configured to: and classifying the obtained training models by using a clustering algorithm to obtain N training sample sets.
In combination with an embodiment of the second aspect, in some embodiments, the training of the meteorological features of the sample comprises: a feature for indicating precipitation, and the determining unit is further specifically configured to: and classifying the obtained training samples according to the precipitation indicated by each training sample to obtain N training sample sets.
In combination with an embodiment of the second aspect, in some embodiments, the training of the meteorological features of the sample comprises: a characteristic for indicating a wind speed, and the determining unit is further specifically configured to: and classifying the obtained training samples according to the wind speed indicated by each training sample to obtain N training sample sets.
With reference to the embodiment of the second aspect, in some embodiments, the determining unit is further specifically configured to: screening training samples in a preset proportion in each training sample set; and training a preset initial neural network model according to the training samples screened from the training sample sets to obtain the N forecasting model groups.
With reference to the embodiment of the second aspect, in some embodiments, the determining unit is further specifically configured to: and verifying each forecasting model group by using the training samples which are not screened out from each training sample set.
In a third aspect, an embodiment of the present disclosure provides an electronic device, including: one or more processors; a storage device for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement the method for determining weather forecast results as described above in the first aspect.
In a fourth aspect, the disclosed embodiments provide a computer readable medium, on which a computer program is stored, which when executed by a processor, implements the steps of the method for determining weather forecast results as described above in the first aspect.
According to the method and the device for determining the weather forecast result, after the weather forecast sets are received, the ensemble forecasts in the weather forecast sets can be classified based on forecast characteristics indicated by the ensemble forecasts in the weather forecast sets, at least one sub-weather forecast set is obtained, each sub-weather forecast set comprises a plurality of ensemble forecasts with the same or similar indication forecast characteristics, a target forecast member is screened out from each sub-weather forecast set, and then the weather forecast result is determined according to the determined target forecast member. That is, after the weather forecast sets are received, the weather forecast sets are classified, so that the forecast characteristics indicated by the set forecasts in each category are close to each other, a target forecast member which is most likely to be close to the actual situation is screened out from each category, and then the weather forecast results are obtained according to the screened target forecast member, so that the obtained weather forecast results are more accurate.
Drawings
The above and other features, advantages and aspects of various embodiments of the present disclosure will become more apparent by referring to the following detailed description when taken in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numbers refer to the same or similar elements. It should be understood that the drawings are schematic and that elements and features are not necessarily drawn to scale.
FIG. 1 is a flow diagram of one embodiment of a method of determining weather forecast results according to the present disclosure;
FIG. 2 is a schematic flow chart diagram illustrating another embodiment of a method for determining weather forecast results according to the present disclosure;
FIG. 3 is a schematic diagram of a model connection structure of an embodiment of the method for determining weather forecast results of the present disclosure;
fig. 4 is a schematic connection diagram of the weather forecast result determination device of the present disclosure.
Detailed Description
Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the disclosure are for illustration purposes only and are not intended to limit the scope of the disclosure.
It should be understood that the various steps recited in the method embodiments of the present disclosure may be performed in a different order, and/or performed in parallel. Moreover, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the present disclosure is not limited in this respect.
The term "include" and variations thereof as used herein are open-ended, i.e., "including but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Relevant definitions for other terms will be given in the following description.
It should be noted that the terms "first", "second", and the like in the present disclosure are only used for distinguishing different devices, modules or units, and are not used for limiting the order or interdependence relationship of the functions performed by the devices, modules or units.
It is noted that references to "a" and "an" in this disclosure are intended to be illustrative rather than limiting, and that those skilled in the art will recognize that reference to "one or more" unless explicitly stated otherwise by context.
The names of messages or information exchanged between devices in the embodiments of the present disclosure are for illustrative purposes only, and are not intended to limit the scope of the messages or information.
Referring to fig. 1, a flow of an embodiment of a method of determining weather forecast results according to the present disclosure is shown. The subject of execution of the weather forecast result determination may be a terminal device and/or a server, as shown in fig. 1, and the weather forecast result determination includes the following steps:
Here, forecast characteristics indicated by ensemble forecasts include: precipitation or wind speed.
Here, each set of sub-meteorological forecasts comprises at least one ensemble forecast.
In some embodiments, after receiving a set of weather forecasts, multiple forecasts may be included in the set of weather forecasts, with some forecasts indicating similar forecast characteristics and some forecasts indicating more distant forecast characteristics. For example, forecast members in one ensemble forecast A may each indicate precipitation at a time in the future with similar indicated precipitation, while forecast members in another ensemble forecast B also indicate precipitation with similar precipitation as indicated by ensemble forecast A, at which time ensemble forecast A and ensemble forecast B may be divided into the same subset of weather forecast ensembles. In other words, ensemble forecasts, which may be understood to have the same or similar forecast characteristics, may be divided within the same sub-ensemble of meteorological forecasts. And the meteorological features indicated by the forecast members in each ensemble forecast are similar.
That is, the weather forecast ensemble includes many forecasts, however, the amount of precipitation indicated may vary from one forecast to another. For example, some forecasts may indicate a greater amount of precipitation, while some forecasts may indicate a lesser amount of precipitation. Accordingly, the wind speed indicated by different ensemble forecasts may also be different.
In some embodiments, the classification criteria for precipitation and the standard method for wind speed may be different, and therefore, after receiving the set of weather forecasts, the set of weather forecasts may be classified based on forecast characteristics indicated by each of the set forecasts in the set of weather forecasts, such that at least one sub-set of weather forecasts may also be obtained.
In some embodiments, if the forecast characteristics are precipitation, multiple precipitation ranges may be partitioned, and predictions for which the indicated precipitation is within the same range may be partitioned into the same sub-meteorological forecast set. Accordingly, when the forecast characteristic is wind speed, the wind speed may be divided into a plurality of ranges, and the ensemble forecast with the indicated wind speed in the same range may be divided into the same sub-meteorological forecast ensemble.
And 102, screening a target forecast member from each sub-meteorological forecast set, and determining a meteorological forecast result according to the screened target forecast member.
In some embodiments, there may be one target forecast member in each sub-meteorological forecast set, and the forecast characteristics indicated by the target forecast member may represent forecast characteristics indicated by a majority of forecast members in the sub-meteorological forecast set (where forecast members of a plurality of ensemble forecasts may be included). That is, the forecast characteristics indicated by most forecast members in the sub-forecast collection may be similar to the forecast characteristics indicated by the target forecast member. Thus, the forecast characteristics indicated by the forecast member are most likely the truest forecast characteristics.
In some embodiments, since the forecast characteristics indicated by each target forecast member may be the most true forecast characteristics, the determination of the weather forecast result according to the screened target forecast members is more accurate.
It can be seen that after a weather forecast set is received, ensemble forecasts in the weather forecast set can be classified based on forecast characteristics indicated by the ensemble forecasts in the weather forecast set to obtain at least one sub-weather forecast set, each sub-weather forecast set comprises a plurality of ensemble forecasts with the same or similar indication forecast characteristics, a target forecast member is screened out from each sub-weather forecast set, and then a weather forecast result is determined according to the determined target forecast member. That is, after the weather forecast sets are received, the weather forecast sets are classified, so that the forecast characteristics indicated by the set forecasts in each category are close to each other, a target forecast member which is most likely to be close to the actual situation is screened out from each category, and then the weather forecast results are obtained according to the screened target forecast member, so that the obtained weather forecast results are more accurate.
In some embodiments, the forecast period for each forecast member in the ensemble forecast may be the same.
In some embodiments, when classifying the ensemble forecasts in the weather forecast ensemble, a clustering method may be directly utilized, that is, after the clustering method processes the weather forecast ensemble, at least one cluster may be obtained, and each cluster may be understood as a sub-weather forecast ensemble.
As an example, the clustering method used may be the K-means clustering method.
IN some embodiments, a spatial forecasting station may obtain weather forecast data using RMAPS-IN (which may be based on real-time weather observations, numerical forecast fusion integration techniques, etc.), and after obtaining forecast data, may obtain multiple forecast members using different parameterization schemes (herein, it may be understood that different perturbation schemes are used), and multiple forecast members.
In some embodiments, the perturbation schemes used in generating the forecast members are different, so that the forecast characteristics indicated by the forecast members are different, and accordingly, the forecast characteristics indicated by some forecast members may be closer to the actual situation, while the forecast characteristics indicated by some forecast members may be far from the actual situation, so that the purpose of determining the target forecast member from the sub-weather forecast set is to screen out the forecast member whose indicated forecast characteristics are most likely to be close to the actual situation.
In some embodiments, the step 102 (selecting a target forecast member from each of the sub-weather forecast sets, and determining a weather forecast result according to the selected target forecast member) may specifically include: and screening a target forecast member from each sub-meteorological forecast set by using the forecast model group, and determining a meteorological forecast result based on the screened target forecast member by using the forecast model group.
In some embodiments, the set of prediction models may be understood as a set of models that has been trained in advance, and the set of prediction models may be used to find a target prediction member in the sub-meteorological prediction set that is most likely to reflect the sub-meteorological prediction set, for example, a prediction member with the highest similarity to each prediction member may be found by using the idea of linear regression, and the prediction member may be determined as the target prediction member. As an example, the target forecast member most likely reflects the true forecast situation.
That is, the prediction model set can be selected according to the actual situation, in the practical application, the neural network model is utilized to select the most representative member from the plurality of members, and for the simplicity of the description, the specific determination mode of the estimated accuracy is not limited, and only the reasonable setting is needed according to the actual situation. For example, the set of prediction models may be a set of linear regression models that identify a member of the set that indicates a prediction characteristic that is similar to the prediction characteristic indicated by a majority of the members of the set.
In some embodiments, the step 102 (selecting a target forecast member from each of the sub-weather forecast sets, and determining a weather forecast result according to the selected target forecast member) may specifically include: and screening out a target forecast member from each sub forecast member according to the forecast characteristic distribution of the forecast members in each sub meteorological forecast set.
In some embodiments, according to the distribution of the forecast characteristics of each forecast member, it is convenient to know that the forecast characteristics of those forecast members are located near the distribution center, so that the forecast member whose forecast characteristics are indicated in the distribution center can be determined as the target forecast member.
It should be noted that although each sub-forecast aggregate includes a plurality of forecasts, the aggregate forecast includes a plurality of forecast members, so that a target forecast member can be screened out from each sub-meteorological forecast member directly according to the forecast characteristic distribution of each forecast member.
There are many ways to screen out target forecast members from multiple forecast members, and for the sake of brevity of the description, detailed description is omitted here. For example, in some implementations, a neural network model may also be utilized to screen out one target forecast member from each of the subset forecast members.
In some embodiments, there may be multiple sub-models within each set of prediction models, and each sub-prediction model may correspond to a sub-meteorological prediction set, i.e., the sub-meteorological prediction sets to which each sub-prediction model is adapted may be different. For example, a certain sub-meteorological forecast model can more efficiently and accurately screen out target forecast members from a sub-meteorological forecast set aiming at heavy rain, and a certain forecast model group can more accurately screen out target forecast members from a sub-meteorological forecast set aiming at heavy wind. That is, each sub-forecasting model may have better judgment ability for a specific meteorological feature, so that the target forecasting member can be better screened from the sub-meteorological forecasting set indicating the specific meteorological feature.
In some embodiments, the classification criteria for classifying an ensemble forecast of a weather forecast ensemble may be based on characteristics of the sub-forecast models that have been obtained. For example, three sub-forecasting models are obtained, a sub-forecasting model a for heavy rain weather, a sub-forecasting model B for light rain weather, and a sub-forecasting model C for medium rain weather. At this time, the received weather forecast aggregate can be divided into 3 sub weather forecast aggregates, the forecast characteristics indicated by each of the forecast aggregates in one of the sub weather forecast aggregates are heavy rain, the forecast characteristics indicated by each of the forecast aggregates in one of the sub weather forecast aggregates are light rain, and the forecast characteristics indicated by each of the forecast aggregates in one of the sub weather forecast aggregates are my medium rain. Of course, the number of forecasts in the 3 sub-sets of weather forecasts may vary, and it is also possible that a particular set of sub-weather forecasts is an empty set, e.g., none of the forecast characteristics indicated by the forecasts may be a drizzle.
In some embodiments, the ensemble predictions in each sub-meteorological prediction ensemble may be input into the corresponding sub-meteorological prediction model, so that the target prediction member output by each sub-meteorological prediction model may be more accurate, and the meteorological prediction result determined by using the prediction model set may be more accurate.
In some embodiments, 10 sub-prediction models may be established for precipitation and 9 sub-prediction models may be established for wind speed. That is, the precipitation amount may be classified into ten grades according to the characteristics of the precipitation amount, and the characteristics of the precipitation amount may include at least: the precipitation quantity is large, and the precipitation is consumed. Of course, precipitation rain intensity, etc. may also be considered in some implementations. That is, the precipitation amount may be divided into 10 levels according to the characteristics of the precipitation amount, each level corresponds to one sub-prediction model, the wind speed may be divided into 9 levels according to the characteristics of the wind speed (the characteristics of the wind speed at least include the magnitude and the generation and the elimination of the wind speed), and correspondingly, each level may also correspond to one sub-prediction model. Of course, how many sub-prediction models need to be established can also be set according to actual conditions.
In some implementations, with continuing reference to fig. 2, fig. 2 shows a flowchart of obtaining a set of forecasting models, which may be obtained in the manner described in steps 201-203, as shown in fig. 2:
Here, the training samples may include: forecast weather data for training and live weather data for training.
In some embodiments, the forecast data for training may be understood as an ensemble forecast for training, which may include different types of ensemble forecasts as well as different kinds of ensemble forecasts. For example, ensemble prediction and live data received over a historical period of time may be selected as training samples. For example, an ensemble forecast for the previous month weather is acquired, and live data of the previous month weather is acquired as training samples. Of course, the data of the past time intervals are specifically selected as training samples, which are not limited here, and only need to be set reasonably according to the situation.
In some embodiments, when the forecasting model group is mainly used for selecting target forecasting members for ensemble forecasting of heavy rainfall in summer, ensemble forecasting data in summer in the past year can be used as training samples, so that the judgment capability of the finally obtained forecasting model can be more accurate.
Here, N may be a positive integer.
In some embodiments, the meteorological features indicated by each training sample are different, i.e., the characteristics of each training sample are different. Therefore, the obtained training samples can be classified according to the meteorological features indicated by the training samples to obtain N training sample sets, so that the meteorological features of the training samples in each training sample set in the N training sample sets are relatively close to each other, and the accuracy of the trained forecasting model can be improved.
In some embodiments, the weather station to which the weather data corresponds may also be considered in classifying the training forecasted weather data. As an example, the observation data is station data, and the weather stations may be classified for more intuitive corresponding results and for determining whether the trained forecasting model meets the requirements.
In some embodiments, in the process of classifying the training samples, the classification in step 101 may also be adopted, so that the obtained model group may better identify the target forecast members in each sub-meteorological forecast set.
In some embodiments, the training samples in each training sample set have close meteorological features, so the trained prediction model set can have strong recognition accuracy.
In some embodiments, the obtained training samples may be classified by using a clustering algorithm, so as to obtain N training sample sets.
In some embodiments, the obtained training models may be classified by an iteratively solved cluster analysis algorithm, for example, a K-means clustering algorithm (K-means clustering algorithm) to classify the obtained training models. K cluster centroids (cluster centroids) can be selected as { μ 1, μ 2, μ 3.. μ k-1, μ k }. epsilon.Rn, and the following process is iterated until convergence { for each sample i, the class c to which it should belong is calculated(i)：＝argminj||x(i)-μjFor each class j, re-computing the centroid of that class
Until the centroid is unchanged or varies little.
Here, the k value may be obtained by calculating an interval statistic (GAP STATISITC).
In some embodiments, the calculation of the optimal number of classifications of k-means may continue by calculating an interval statistic (GAP STATISTIC). For example, gap (k) ═ E (log (Dk)) -log (Dk); here, E · (log (dk))) is a expectation of logDk, and can be generally generated using monte carlo simulations. The basic process of the algorithm may be that firstly, random samples as many as the number of original samples are randomly generated in the area where the samples are located according to uniform distribution, and a Dk is obtained for the K-means. Then, E (logDk) can be calculated approximately by repeating the above steps. However, in practice gap (k) can be seen as the difference between the loss of the random sample and the loss of the actual sample. Assuming that the optimal cluster number corresponding to the actual sample is kb, the loss of the actual sample should be relatively small, and the difference between the random sample loss and the actual sample loss is also the maximum response, i.e. the k value corresponding to the maximum value obtained by gap (k) is the optimal cluster number.
In some embodiments, the value of k may be equal to N.
In some embodiments, the initial neural network model may be a reverse transmission deep learning model, which may be a model transmission structure of a single model of the reverse transmission deep learning model as shown in fig. 3 for ease of understanding; the model may be composed of four fully connected layers, an activation function of each layer is equal to relu, 20% of Dropout is added between the first three layers (which may prevent an overfitting phenomenon during model training), the last layer is linked by fully connected layers, and Dense (219, activation ═ relu ') → Dropout (0.2) → Dense (128, activation ═ relu') → Dropout (0.2) → Dense (64, activation ═ relu ') -Dropout (0.2) → Dense (32, activation ═ relu') → Dropout (0.2) → Dense (16, activation ═ relu ') -Dense (1, activation ═ reimbus'), a loss function is equal to 'relu'), an optimization function may be equal to the number of samples, and the number of iterations may be selected according to a number of learning times of 1000000.
In some embodiments, training samples can be classified according to different meteorological features indicated by the training samples, so that certain meteorological features in the training sample sets can be closer, and the forecasting model trained by using the training sample sets can have higher accuracy.
In some embodiments, the meteorological features of the training samples may include: a feature for indicating precipitation; at this time, the obtained training samples may be classified according to the precipitation amount indicated by each training sample, so as to obtain N training sample sets.
Here, the training samples may be classified based on the feature of the precipitation amount and a k-means clustering algorithm to obtain N training sample sets.
In some embodiments, by obtaining a sample set corresponding to each precipitation amount, a plurality of prediction model sets may be trained by using training samples, that is, each obtained prediction model set may correspond to a certain precipitation interval. For example, the precipitation interval corresponding to the forecast model group a is 1-3mm per hour, and if the precipitation indicated by the received multiple forecast reports is all within the interval, the multiple forecast reports can be all input into the forecast model group a, and the forecast model group a can output a target forecast member. In this way, each forecasting model group only processes specific forecasting data, so that the accuracy of the forecasting model group can be improved.
In some embodiments, the meteorological features of the training samples may include: and the characteristics are used for indicating the wind speed, and at this time, the obtained training samples can be classified according to the wind speed indicated by each training sample, so that N training sample sets are obtained.
In some embodiments, the method for classifying the training samples indicating the wind speed and the method for classifying the training samples indicating the precipitation amount may be the same classification method or different classification methods, and only needs to be selected reasonably according to actual conditions.
In some embodiments, the training samples may further include other types of feature data, and accordingly, the training samples may also be classified according to other types of data features to obtain N training sample sets.
In some embodiments, by obtaining N training sample sets, N prediction model sets can be obtained, which not only makes the recognition capability of obtaining the prediction model sets more accurate, but also improves the processing efficiency of processing the prediction data by using the N prediction model sets.
In some embodiments, training samples in a preset proportion in each training sample set may be screened; the preset initial neural network model can be trained according to the training samples screened from each training sample set to obtain N prediction model groups.
In some embodiments, the training samples in the preset proportion in each training sample set are screened out, so that the number of training samples used for training the forecasting model group can be reduced, and the training efficiency of the forecasting model group can be improved.
In some embodiments, the set of prediction models may be verified using unscreened training samples in the set of training samples.
In some embodiments, the prediction capability of each prediction model group can be determined by verifying each prediction model group by using the training samples not screened out from each training sample set, so that the judgment capability obtained by using the prediction model group can be more accurate.
In some embodiments, since the prediction model groups are verified by using the training samples that are not screened out from the training sample sets, that is, the training samples are divided into the training samples and the verification samples, and in this way, the prediction model groups can be more accurately verified.
In some embodiments, the set of forecast models trained through steps 201-203 can be used to filter the ensemble forecasts to determine target forecast members in the ensemble forecasts. In this way, the set forecast is better utilized to determine the weather forecast result.
With further reference to fig. 4, as an implementation of the methods shown in the above figures, the present disclosure provides a weather forecast result determination apparatus, which corresponds to the method embodiment shown in fig. 1, and which can be applied to various electronic devices.
As shown in fig. 4, the weather forecast result determination device of the present embodiment includes: a classifying unit 401, configured to, in response to receiving a weather forecast set, classify the weather forecast set based on forecast characteristics indicated by each ensemble forecast in the weather forecast set, so as to obtain at least one set of sub-weather forecast sets, where the forecast characteristics include: precipitation or wind speed, each sub-meteorological prediction ensemble comprising at least one ensemble prediction; a screening unit 402, configured to screen out a target forecast member from each sub-weather forecast set, and determine a weather forecast result according to the screened target forecast member.
In some optional embodiments, the screening unit 402 is further specifically configured to: and screening out a target forecast member from each sub-meteorological forecast set by utilizing the forecast model group, and determining a meteorological forecast result based on the screened target forecast member by utilizing the forecast model group.
In some optional embodiments, the screening unit 402 is further specifically configured to: and screening a target forecast member from each sub-meteorological forecast set according to the forecast characteristic distribution of each forecast member in each sub-meteorological forecast set.
In some optional embodiments, the apparatus further comprises a determining unit 403, configured to obtain the set of prediction models by:
obtaining training samples, wherein the training samples comprise: forecast meteorological data for training and live meteorological data for training;
classifying the obtained training samples to obtain N training sample sets, wherein N is a positive integer;
and obtaining the N forecasting model groups by utilizing each training sample set to preset an initial neural network model.
In some optional embodiments, the determining unit 403 is further specifically configured to: and classifying the obtained training models by using a clustering algorithm to obtain N training sample sets.
In some alternative embodiments, the meteorological features of the training samples include: a feature for indicating precipitation, and the determining unit 403 is further specifically configured to: and classifying the obtained training samples according to the precipitation indicated by each training sample to obtain N training sample sets.
In some alternative embodiments, the meteorological features of the training samples include: a feature for indicating a wind speed, and the determining unit 403 is further configured to: and classifying the obtained training samples according to the wind speed indicated by each training sample to obtain N training sample sets.
In some optional embodiments, the determining unit 403 is further specifically configured to: screening training samples in a preset proportion in each training sample set; and training a preset initial neural network model according to the training samples screened from the training sample sets to obtain the N forecasting model groups.
In some optional embodiments, the determining unit 403 is further specifically configured to: and verifying each forecasting model group by using the training samples which are not screened out from each training sample set.
It should be noted that the computer readable medium of the present disclosure can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In contrast, in the present disclosure, a computer readable signal medium may comprise a propagated data signal with computer readable program code embodied therein, either in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, optical cables, RF (radio frequency), etc., or any suitable combination of the foregoing.
In some embodiments, the clients, servers may communicate using any currently known or future developed network Protocol, such as HTTP (HyperText Transfer Protocol), and may interconnect with any form or medium of digital data communication (e.g., a communications network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the Internet (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed network.
The computer readable medium may be embodied in the electronic device; or may exist separately without being assembled into the electronic device.
The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: in response to the received ensemble forecast, determining target forecast members in the ensemble forecast by using a pre-built forecast model group, wherein the forecast model group is used for calculating the forecast accuracy of each forecast member; and determining a weather forecast result according to the determined target forecast member.
Computer program code for carrying out operations for the present disclosure may be written in any combination of one or more programming languages, including but not limited to an object oriented programming language such as Java, Smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units described in the embodiments of the present disclosure may be implemented by software or hardware. Where the name of a unit does not in some cases constitute a limitation on the unit itself, for example, classification unit 401 may also be described as a "unit that classifies forecast members of an ensemble forecast".
The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), systems on a chip (SOCs), Complex Programmable Logic Devices (CPLDs), and the like.
In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing description is only exemplary of the preferred embodiments of the disclosure and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the disclosure herein is not limited to the particular combination of features described above, but also encompasses other embodiments in which any combination of the features described above or their equivalents does not depart from the spirit of the disclosure. For example, the above features and (but not limited to) the features disclosed in this disclosure having similar functions are replaced with each other to form the technical solution.
Further, while operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are included in the above discussion, these should not be construed as limitations on the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination.
Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.
Claims (10)
1. A method for determining a weather forecast result, comprising:
in response to receiving a set of weather forecasts, classifying the set of weather forecasts based on forecast characteristics indicated by each of the set forecasts in the set of weather forecasts to obtain at least one set of sub-sets of weather forecasts, wherein the forecast characteristics include: precipitation or wind speed, each sub-meteorological prediction ensemble comprising at least one ensemble prediction;
and screening a target forecast member from each sub-meteorological forecast set, and determining a meteorological forecast result according to the screened target forecast member.
2. The method of claim 1, wherein the selecting a target forecast member from each of the sub-weather forecast collections, and determining weather forecast results based on the selected target forecast member, comprises:
and screening a target forecast member from each sub-meteorological forecast set by using a forecast model group, and determining a meteorological forecast result based on the screened target forecast member by using the forecast model group.
3. The method of claim 1, wherein the screening out a target forecast member from each of the sub-meteorological forecast sets comprises:
and screening out a target forecast member from each sub-meteorological forecast set according to the forecast characteristic distribution of the forecast members in each sub-meteorological forecast set.
4. The method of claim 2, wherein the set of prediction models is obtained by:
obtaining training samples, wherein the training samples comprise: forecast meteorological data for training and live meteorological data for training;
classifying the obtained training samples to obtain N training sample sets, wherein N is a positive integer;
and (4) obtaining N forecasting model groups by utilizing each training sample set to preset the initial neural network model.
5. The method of claim 4, wherein training the meteorological features of the sample comprises: a feature for indicating precipitation; and classifying the obtained training samples to obtain N training sample sets, including:
and classifying the obtained training samples according to the precipitation indicated by each training sample to obtain the N training sample sets.
6. The method of claim 4, wherein training the meteorological features of the sample comprises: features for indicating wind speed, and classifying the acquired training samples to obtain N training sample sets, including:
and classifying the obtained training samples according to the wind speed indicated by each training sample to obtain the N training sample sets.
7. The method of claim 4, wherein obtaining the N sets of prediction models by using the preset initial neural network model for each training sample set comprises:
screening training samples in a preset proportion in each training sample set;
and training a preset initial neural network model according to the training samples screened from the training sample sets to obtain the N forecasting model groups.
8. An apparatus for determining a weather forecast result, comprising:
the classification unit is used for responding to the received weather forecast sets, classifying the weather forecast sets based on forecast characteristics indicated by all the forecast in the weather forecast sets, and obtaining at least one group of sub-weather forecast sets, wherein the forecast characteristics comprise: precipitation or wind speed, each sub-meteorological prediction ensemble comprising at least one ensemble prediction;
and the screening unit is used for screening out a target forecast member from each sub-meteorological forecast set and determining a meteorological forecast result according to the screened target forecast member.
9. An electronic device comprising a processor and a memory, said memory storing computer readable instructions which, when executed by said processor, perform the steps of the method of any of claims 1-7.
10. A computer-readable medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110751609.2A CN113341481A (en) | 2021-07-02 | 2021-07-02 | Method and device for determining weather forecast result |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110751609.2A CN113341481A (en) | 2021-07-02 | 2021-07-02 | Method and device for determining weather forecast result |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113341481A true CN113341481A (en) | 2021-09-03 |
Family
ID=77482323
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110751609.2A Pending CN113341481A (en) | 2021-07-02 | 2021-07-02 | Method and device for determining weather forecast result |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113341481A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170261646A1 (en) * | 2016-03-08 | 2017-09-14 | China Institute Of Water Resources And Hydropower Research | Self-correcting multi-model numerical rainfall ensemble forecasting method |
| CN107392349A (en) * | 2017-06-01 | 2017-11-24 | 中国水利水电科学研究院 | A kind of DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM method and device |
| CN110019124A (en) * | 2017-11-21 | 2019-07-16 | 中国电力科学研究院有限公司 | A kind of set weather forecast revision method and system |
| CN111461444A (en) * | 2020-04-07 | 2020-07-28 | 上海电气风电集团股份有限公司 | Prediction method, system, medium and electronic device for unit power of wind power plant |
| CN111723929A (en) * | 2020-05-08 | 2020-09-29 | 广州华工中云信息技术有限公司 | Numerical prediction product correction method, device and system based on neural network |
| CN111929748A (en) * | 2020-09-17 | 2020-11-13 | 南京浦蓝大气环境研究院有限公司 | Meteorological element forecasting method and system |
| CN112232554A (en) * | 2020-09-28 | 2021-01-15 | 中国科学院空天信息创新研究院 | Construction method of local short-term rainfall forecast model based on BP neural network |
-
2021
- 2021-07-02 CN CN202110751609.2A patent/CN113341481A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170261646A1 (en) * | 2016-03-08 | 2017-09-14 | China Institute Of Water Resources And Hydropower Research | Self-correcting multi-model numerical rainfall ensemble forecasting method |
| CN107392349A (en) * | 2017-06-01 | 2017-11-24 | 中国水利水电科学研究院 | A kind of DATA PROCESSING IN ENSEMBLE PREDICTION SYSTEM method and device |
| CN110019124A (en) * | 2017-11-21 | 2019-07-16 | 中国电力科学研究院有限公司 | A kind of set weather forecast revision method and system |
| CN111461444A (en) * | 2020-04-07 | 2020-07-28 | 上海电气风电集团股份有限公司 | Prediction method, system, medium and electronic device for unit power of wind power plant |
| CN111723929A (en) * | 2020-05-08 | 2020-09-29 | 广州华工中云信息技术有限公司 | Numerical prediction product correction method, device and system based on neural network |
| CN111929748A (en) * | 2020-09-17 | 2020-11-13 | 南京浦蓝大气环境研究院有限公司 | Meteorological element forecasting method and system |
| CN112232554A (en) * | 2020-09-28 | 2021-01-15 | 中国科学院空天信息创新研究院 | Construction method of local short-term rainfall forecast model based on BP neural network |
Non-Patent Citations (1)
| Title |
|---|
| 杜钧等: "集合预报方法在暴雨研究和预报中的应用", 《气象科技进展》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112118551B (en) | Equipment risk identification method and related equipment | |
| CN111461410B (en) | Air quality prediction method and device based on transfer learning | |
| Jalalkamali | Using of hybrid fuzzy models to predict spatiotemporal groundwater quality parameters | |
| CN112285807B (en) | Meteorological information prediction method and device | |
| CN104281882A (en) | Method and system for predicting social network information popularity on basis of user characteristics | |
| CN109799550B (en) | Method and device for predicting rainfall intensity | |
| CN107886160B (en) | BP neural network interval water demand prediction method | |
| CN111435463A (en) | Data processing method and related equipment and system | |
| CN111784044A (en) | Landslide prediction method, device, equipment and storage medium | |
| CN112910690A (en) | Network traffic prediction method, device and equipment based on neural network model | |
| Shao et al. | The Traffic Flow Prediction Method Using the Incremental Learning‐Based CNN‐LTSM Model: The Solution of Mobile Application | |
| CN108805174A (en) | clustering method and device | |
| CN112468324B (en) | Graph convolution neural network-based encrypted traffic classification method and device | |
| CN116596095A (en) | Training method and device of carbon emission prediction model based on machine learning | |
| Noor et al. | A non-local model output statistics approach for the downscaling of CMIP5 GCMs for the projection of rainfall in Peninsular Malaysia | |
| Shahdad et al. | Drought forecasting using new advanced ensemble-based models of reduced error pruning tree | |
| CN112990311A (en) | Method and device for identifying admitted client | |
| CN117117833A (en) | Photovoltaic output power prediction method and device, electronic equipment and storage medium | |
| CN118037000A (en) | Urban resource dynamic scheduling method and system based on digital economy | |
| CN113127472B (en) | Method and system for real-time deduplication statistics of number of drivers with large reporting amount | |
| Shinkuma et al. | Data assessment and prioritization in mobile networks for real-time prediction of spatial information using machine learning | |
| CN113341481A (en) | Method and device for determining weather forecast result | |
| CN112231299A (en) | Method and device for dynamically adjusting feature library | |
| Dhillon et al. | Seasonal precipitation forecasting for water management in the Kosi Basin, India using large-scale climate predictors | |
| CN113723712B (en) | Wind power prediction method, system, equipment and medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210903 |