Meteorological prediction method, device, system and storage medium

Abstract

The application discloses a meteorological prediction method, device and system and a storage medium. The method comprises the following steps: acquiring a first meteorological parameterization scheme set, and generating a first-layer proxy auxiliary model according to the first meteorological parameterization scheme set; optimizing the first-layer proxy auxiliary model to determine a second meteorological parameterization scheme in the first meteorological parameterization scheme set; generating a second-layer proxy auxiliary model according to the second meteorological parameterization scheme; optimizing the second-layer proxy auxiliary model, and performing meteorological prediction based on the optimized second-layer proxy auxiliary model. By constructing a double-layer proxy auxiliary model, the double-layer proxy auxiliary model is optimized respectively, the meteorological parameterization scheme of the locally optimized configuration of the mesoscale numerical weather prediction model is determined, and the accuracy of meteorological prediction is improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to weather forecasting methods, devices, systems and storage media. Background Technology

[0002] Mesoscale numerical weather prediction models, as an effective means of obtaining high spatiotemporal resolution meteorological forecasts, play a vital role in the safe and efficient development of new energy sources, primarily photovoltaics and wind power. The configuration of physical process parameterization schemes within numerical models is one of the key factors affecting the accuracy of meteorological forecasts; different regions generally require different physical process parameterization schemes.

[0003] The traditional approach is to determine the optimal parameterization scheme from relevant references based on human experience, and then make weather forecasts for a specific study area based on the optimal parameterization scheme. However, the optimal parameterization scheme determined by the traditional approach has a large degree of uncertainty, which leads to a large error in the final weather forecast results.

[0004] Therefore, determining the parameterization scheme for weather forecasting to improve its accuracy is an urgent problem to be solved. Summary of the Invention

[0005] The main objective of this invention is to propose a weather forecasting method, device, system, and storage medium, aiming to solve the problem of how to determine the parameterization scheme for weather forecasting in order to improve the accuracy of weather forecasting.

[0006] To achieve the above objectives, the present invention provides a weather forecasting method, which includes the following steps:

[0007] Obtain a first set of meteorological parameterization schemes, and generate a first-layer proxy auxiliary model based on the first set of meteorological parameterization schemes;

[0008] The first-layer proxy auxiliary model is optimized to determine the second meteorological parameterization scheme from the first meteorological parameterization scheme set;

[0009] A second-layer proxy auxiliary model is generated based on the second meteorological parameterization scheme;

[0010] The second-layer proxy auxiliary model is optimized, and weather forecasts are made based on the optimized second-layer proxy auxiliary model.

[0011] Optionally, the steps for obtaining the first set of meteorological parameterization schemes include:

[0012] Obtain the set of preset meteorological parameterization schemes and the meteorological types to be predicted;

[0013] The set of preset meteorological parameterization schemes is filtered according to the meteorological type to be predicted to obtain the first meteorological parameterization scheme.

[0014] Optionally, the step of generating the first-layer surrogate auxiliary model based on the first meteorological parameterization scheme set includes:

[0015] Determine the types of schemes included in the first meteorological parameterization scheme set, and determine the number of schemes corresponding to each scheme type;

[0016] Based on the scheme type and the number of schemes, an orthogonal experiment is conducted on the first meteorological parameterization scheme set to obtain the target meteorological parameterization scheme set;

[0017] Determine the first meteorological forecast result corresponding to each target meteorological parameterization scheme in the target meteorological parameterization scheme set, and generate a first-layer surrogate auxiliary model based on the target meteorological parameterization scheme set and the first meteorological forecast result.

[0018] Optionally, the step of optimizing the first-layer proxy auxiliary model to determine the second meteorological parameterization scheme from the first set of meteorological parameterization schemes includes:

[0019] The first set of meteorological parameterization schemes is input into the first layer of proxy auxiliary model to obtain the second meteorological forecast result;

[0020] Acquire actual meteorological data, and determine a first prediction error based on the actual meteorological data and the second meteorological forecast result;

[0021] The first-layer proxy auxiliary model is optimized based on the first prediction error to obtain the first-layer target proxy auxiliary model, and the second meteorological parameterization scheme is determined in the first meteorological parameterization scheme set based on the first-layer target proxy auxiliary model.

[0022] Optionally, the step of optimizing the first-layer agent-assisted model based on the first prediction error to obtain the first-layer target agent-assisted model includes:

[0023] Based on the first prediction error, determine the fitness of the first-layer proxy auxiliary model, and update the first meteorological parameterization scheme set based on the fitness.

[0024] The updated set of first meteorological parameterization schemes is input into the first layer surrogate auxiliary model to update the fitness of the first layer surrogate auxiliary model;

[0025] The process continues until the preset conditions are met, at which point the first-layer target agent auxiliary model is obtained.

[0026] Optionally, the step of generating the second-layer surrogate auxiliary model according to the second meteorological parameterization scheme includes:

[0027] The parameters in the second meteorological parameterization scheme are screened to determine the target parameter set;

[0028] Determine the set of parameter values ​​corresponding to each target parameter in the target parameter set;

[0029] The third meteorological forecast result corresponding to the second meteorological parameterization scheme is determined based on the set of parameter values, and a second-layer proxy auxiliary model is generated based on the third meteorological forecast result and the second meteorological parameterization scheme.

[0030] Optionally, the step of determining the set of parameter values ​​corresponding to each target parameter in the target parameter set includes:

[0031] Determine the value range corresponding to each target parameter in the target parameter set, and sample the value range to determine the parameter value set corresponding to each target parameter.

[0032] Optionally, the steps for optimizing the second-layer agent-assisted model include:

[0033] The second meteorological parameterization scheme set is input into the second-layer proxy auxiliary model to obtain the fourth meteorological forecast result;

[0034] Acquire actual meteorological data, and determine the second prediction error based on the actual meteorological data and the fourth meteorological forecast result;

[0035] The second-layer proxy auxiliary model is optimized based on the second prediction error to determine the target parameter value corresponding to each target parameter in the set of parameter values ​​corresponding to each target parameter.

[0036] Optionally, before the step of performing weather forecasting based on the optimized second-layer surrogate-assisted model, the following steps are included:

[0037] Input the target parameter value corresponding to each target parameter into the numerical pattern to verify the optimized second-layer agent auxiliary model and obtain the verification results;

[0038] If the verification result meets the preset conditions, then the following step is executed: perform weather forecasting based on the optimized second-layer agent-assisted model.

[0039] Furthermore, to achieve the above objectives, the present invention also provides a weather forecasting device, the weather forecasting device comprising:

[0040] The acquisition module is used to acquire a first set of meteorological parameterization schemes and generate a first-layer proxy auxiliary model based on the first set of meteorological parameterization schemes.

[0041] The determination module is used to optimize the first-layer proxy auxiliary model in order to determine the second meteorological parameterization scheme from the first meteorological parameterization scheme set;

[0042] The generation module is used to generate a second-layer proxy auxiliary model based on the second meteorological parameterization scheme;

[0043] The prediction module is used to optimize the second-layer proxy auxiliary model and perform weather forecasting based on the optimized second-layer proxy auxiliary model.

[0044] Furthermore, the acquisition module is also used for:

[0045] Obtain the set of preset meteorological parameterization schemes and the meteorological types to be predicted;

[0046] The set of preset meteorological parameterization schemes is filtered according to the meteorological type to be predicted to obtain the first meteorological parameterization scheme.

[0047] Furthermore, the acquisition module is also used for:

[0048] Determine the types of schemes included in the first meteorological parameterization scheme set, and determine the number of schemes corresponding to each scheme type;

[0049] Based on the scheme type and the number of schemes, an orthogonal experiment is conducted on the first meteorological parameterization scheme set to obtain the target meteorological parameterization scheme set;

[0050] Determine the first meteorological forecast result corresponding to each target meteorological parameterization scheme in the target meteorological parameterization scheme set, and generate a first-layer surrogate auxiliary model based on the target meteorological parameterization scheme set and the first meteorological forecast result.

[0051] Furthermore, the determining module is also used for:

[0052] The first set of meteorological parameterization schemes is input into the first layer of proxy auxiliary model to obtain the second meteorological forecast result;

[0053] Acquire actual meteorological data, and determine a first prediction error based on the actual meteorological data and the first meteorological forecast result;

[0054] The first-layer proxy auxiliary model is optimized based on the first prediction error to obtain the first-layer target proxy auxiliary model, and the second meteorological parameterization scheme is determined in the first meteorological parameterization scheme set based on the first-layer target proxy auxiliary model.

[0055] Furthermore, the determining module also includes an optimization module, which is used to:

[0056] Based on the first prediction error, determine the fitness of the first-layer proxy auxiliary model, and update the first meteorological parameterization scheme set based on the fitness.

[0057] The updated set of first meteorological parameterization schemes is input into the first layer surrogate auxiliary model to update the fitness of the first layer surrogate auxiliary model;

[0058] The process continues until the preset conditions are met, at which point the first-layer target agent auxiliary model is obtained.

[0059] Furthermore, the generation module is also used for:

[0060] The parameters in the second meteorological parameterization scheme are screened to determine the target parameter set;

[0061] Determine the set of parameter values ​​corresponding to each target parameter in the target parameter set;

[0062] The third meteorological forecast result corresponding to the second meteorological parameterization scheme is determined based on the set of parameter values, and a second-layer proxy auxiliary model is generated based on the third meteorological forecast result and the second meteorological parameterization scheme.

[0063] Furthermore, the generation module is also used for:

[0064] Determine the value range corresponding to each target parameter in the target parameter set, and sample the value range to determine the parameter value set corresponding to each target parameter.

[0065] Furthermore, the prediction module also includes an optimization module, which is used to:

[0066] The second meteorological parameterization scheme set is input into the second-layer proxy auxiliary model to obtain the fourth meteorological forecast result;

[0067] Acquire actual meteorological data, and determine the second prediction error based on the actual meteorological data and the fourth meteorological forecast result;

[0068] The second-layer proxy auxiliary model is optimized based on the second prediction error to determine the target parameter value corresponding to each target parameter in the set of parameter values ​​corresponding to each target parameter.

[0069] Furthermore, the prediction module also includes a verification module, which is used for:

[0070] Input the target parameter value corresponding to each target parameter into the numerical pattern to verify the optimized second-layer agent auxiliary model and obtain the verification results;

[0071] If the verification result meets the preset conditions, then the following step is executed: perform weather forecasting based on the optimized second-layer agent-assisted model.

[0072] In addition, to achieve the above objectives, the present invention also provides a weather forecasting system, which includes: a memory, a processor, and a weather forecasting program stored in the memory and executable on the processor. When the weather forecasting program is executed by the processor, it implements the steps of the weather forecasting method as described above.

[0073] In addition, to achieve the above objectives, the present invention also provides a storage medium storing a weather forecasting program, which, when executed by a processor, implements the steps of the weather forecasting method described above.

[0074] The meteorological forecasting method proposed in this invention involves obtaining a first set of meteorological parameterization schemes and generating a first-layer surrogate auxiliary model based on this set. The first-layer surrogate auxiliary model is then optimized to determine a second meteorological parameterization scheme from the first set of schemes. A second-layer surrogate auxiliary model is generated based on the second scheme. This second-layer surrogate auxiliary model is further optimized, and meteorological forecasting is performed based on the optimized model. This invention improves the accuracy of meteorological forecasting by constructing a two-layer surrogate auxiliary model and optimizing both layers to determine the optimal meteorological parameterization scheme for a localized mesoscale numerical weather prediction model. Attached Figure Description

[0075] Figure 1 This is a flowchart illustrating the first embodiment of the weather forecasting method of the present invention;

[0076] Figure 2 This is a flowchart illustrating the second embodiment of the weather forecasting method of the present invention;

[0077] Figure 3 This is a schematic diagram of the weather forecasting device of the present invention.

[0078] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0079] Reference Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the meteorological forecasting method of the present invention. The meteorological forecasting method in this embodiment is applied to a meteorological forecasting system, which can be applied to intelligent devices such as terminal devices and PC terminals, for meteorological forecasting of new energy power generation areas, primarily photovoltaic and wind power. For ease of description, a meteorological forecasting system is used as an example. The method includes:

[0080] Step S10: Obtain the first meteorological parameterization scheme set, and generate the first layer proxy auxiliary model based on the first meteorological parameterization scheme set;

[0081] In this embodiment, the meteorological forecasting system first determines the target area, which is a new energy power generation area mainly composed of photovoltaic and wind power, requiring meteorological forecasting. The system then acquires a set of preset meteorological parameterization schemes corresponding to the target area, selects a first set of meteorological parameterization schemes from this set, and generates a first-layer surrogate auxiliary model based on this first set. It should be noted that the meteorological parameterization scheme is a physical process parameterization scheme in a mesoscale numerical weather prediction model. A meteorological parameterization scheme is a data set describing the physical processes of multiple meteorological phenomena or a single meteorological phenomenon, such as a microphysical scheme, a cumulus scheme, or a planetary boundary layer scheme. The meteorological parameterization scheme is one of the important factors affecting the accuracy of meteorological forecasts. A surrogate model is an approximate mathematical model that can replace complex and time-consuming numerical analysis in optimization design. It can also be called a response surface model or an approximate model. For example, the optimization design of aircraft is typically complex and time-consuming. In addition, when doing optimization design, we inevitably encounter some objective functions that are difficult to express with intuitive functional expressions. In this case, a surrogate model can be used to replace the objective function. Using a surrogate model can greatly improve the efficiency of optimization design and reduce the difficulty of optimization.

[0082] Specifically, the steps for obtaining the first set of meteorological parameterization schemes include:

[0083] Step S101: Obtain the set of preset meteorological parameterization schemes and the meteorological type to be predicted;

[0084] Step S102: Based on the weather type to be predicted, the preset meteorological parameterization scheme set is filtered to obtain the first meteorological parameterization scheme.

[0085] In steps S101 and S102, after the meteorological forecasting system determines the target area, it acquires the corresponding set of preset meteorological parameterization schemes and the type of weather to be predicted within the target area. Then, based on the type of weather to be predicted, it filters the preset meteorological parameterization schemes from the set, selecting the one corresponding to the type of weather to be predicted as the first meteorological parameterization scheme. For example, if the type of weather to be predicted is irradiance and wind speed, the preset meteorological parameterization schemes corresponding to irradiance and wind speed are selected from the set of preset meteorological parameterization schemes. From these preset meteorological parameterization schemes corresponding to irradiance and wind speed, based on expert experience, the preset meteorological parameterization scheme that has a greater impact on the prediction results of irradiance and wind speed is selected as the first meteorological parameterization scheme. It should be noted that "greater impact on the prediction results of irradiance and wind speed" means that the meteorological forecast results differ significantly from the actual weather conditions.

[0086] Specifically, the steps for generating the first-layer surrogate auxiliary model based on the first meteorological parameterization scheme set include:

[0087] Step S103: Determine the types of schemes included in the first meteorological parameterization scheme set, and determine the number of schemes corresponding to the scheme types;

[0088] In this step, after determining the first set of meteorological parameterization schemes, the meteorological forecasting system determines the types of schemes included in the first set of meteorological parameterization schemes and the number of schemes corresponding to each type of scheme.

[0089] Step S104: Based on the scheme type and the number of schemes, perform an orthogonal experiment on the first meteorological parameterization scheme set to obtain the target meteorological parameterization scheme set;

[0090] In this step, after the meteorological forecasting system determines the types of schemes included in the first set of meteorological parameterization schemes and the number of schemes corresponding to each type, it conducts orthogonal experiments on the first set of meteorological parameterization schemes based on the scheme types and the number of schemes to obtain the target set of meteorological parameterization schemes. It should be noted that orthogonal experiments are a method for studying multiple factors and multiple levels. It selects some representative points from the full experiment based on orthogonality and conducts experiments on these representative points. These representative points have the characteristics of "uniform dispersion and comparable uniformity". Orthogonal experiments are the main method of fractional factorial design. For example, there are 28 microphysical schemes, 13 cumulus schemes, and 13 planetary boundary layer schemes. Therefore, the first set of meteorological parameterization schemes includes 28 * 13 * 13 = 4732 schemes. Based on expert experience, the meteorological forecasting system selects a predetermined number of representative microphysical schemes from the 28 microphysical schemes, a predetermined number of representative cumulus schemes from the 13 cumulus schemes, and a predetermined number of representative planetary boundary layer schemes from the 13 planetary boundary layer schemes. Orthogonal experiments are then conducted on the selected representative microphysical schemes, cumulus schemes, and planetary boundary layer schemes. Optionally, the orthogonal experiment can involve selecting one microphysical scheme, one cumulus scheme, and one planetary boundary layer scheme to create different configurations. Alternatively, the orthogonal experiment can also involve selecting only one or two of the microphysical, cumulus, and planetary boundary layer schemes to create different configurations. Through orthogonal experiments, the target set of meteorological parameterization schemes is obtained.

[0091] Step S105: Determine the first meteorological forecast result corresponding to each target meteorological parameterization scheme in the target meteorological parameterization scheme set, and generate a first-layer proxy auxiliary model based on the target meteorological parameterization scheme set and the first meteorological forecast result.

[0092] In this step, after determining the set of target meteorological parameterization schemes, the meteorological forecasting system runs the numerical model WRF (The Weather Research and Forecasting Model) on each target meteorological parameterization scheme in the set to obtain the first meteorological forecast result corresponding to each target meteorological parameterization scheme. Then, based on the set of target meteorological parameterization schemes and the first meteorological forecast result, the system uses machine learning methods or deep learning methods, such as generalized linear models and radial basis networks, to generate a first-layer surrogate auxiliary model.

[0093] Specifically, the objective function of the first-layer agent-assisted model is as follows:

[0094]

[0095] Where Y represents the model output, i.e., the first meteorological forecast result corresponding to different target meteorological parameterization schemes, and X... j X k X i This indicates different parameter settings, that is, the types of schemes included in the target meteorological parameterization scheme, a j a j，k a j，k，i The first, second, and third order relationships between parameters are represented by δ, a0 represents the bias (the bias will vary depending on the target meteorological parameterization scheme), δ represents the error (the error between the meteorological prediction result of the first-layer surrogate auxiliary model for the first meteorological parameterization scheme and the first prediction result of the numerical model for the first meteorological parameterization scheme), and n represents the number of parameters.

[0096] Step S20: Optimize the first layer proxy auxiliary model to determine the second meteorological parameterization scheme in the first meteorological parameterization scheme set;

[0097] In this embodiment, after generating the first-layer proxy auxiliary model, the meteorological forecasting system optimizes the first-layer proxy auxiliary model based on whether the target meteorological parameterization scheme corresponding to the first-layer proxy auxiliary model meets the preset conditions, so as to determine the second meteorological parameterization scheme from the first meteorological parameterization scheme set.

[0098] Specifically, step S20 includes:

[0099] Step S201: Input the first meteorological parameterization scheme set into the first layer proxy auxiliary model to obtain the second meteorological forecast result;

[0100] In this step, the meteorological forecasting system inputs each of the first meteorological parameters in the first set of meteorological parameterization schemes into the first-layer proxy auxiliary model, and obtains the second meteorological forecast result corresponding to each of the first meteorological parameters through the first-layer proxy auxiliary model.

[0101] Step S202: Obtain actual meteorological data, and determine the first prediction error based on the actual meteorological data and the second meteorological forecast result;

[0102] In this step, the meteorological forecasting system acquires the actual meteorological data corresponding to the target area, and determines the first forecast error corresponding to each first meteorological parameterization based on the actual meteorological data and each second meteorological forecast result.

[0103] Step S203: Optimize the first-layer proxy auxiliary model based on the first prediction error to obtain the first-layer target proxy auxiliary model, and determine the second meteorological parameterization scheme in the target meteorological parameterization scheme set according to the first-layer target proxy auxiliary model.

[0104] In this step, the meteorological forecasting system determines the fitness of the first-layer surrogate auxiliary model based on the first prediction error corresponding to each first meteorological parameterization, and iteratively optimizes the first-layer surrogate auxiliary model according to the fitness until the first-layer target surrogate auxiliary model is obtained. Then, the system determines the second meteorological parameterization scheme from the target meteorological parameterization scheme set according to the first-layer target surrogate auxiliary model. There may be one or more first-layer target surrogate auxiliary models and one or more second meteorological parameterization schemes.

[0105] Further, the step of optimizing the first-layer agent-assisted model based on the first prediction error to obtain the first-layer target agent-assisted model includes:

[0106] Step S2031: Determine the fitness of the first layer proxy auxiliary model based on the first prediction error, and update the first meteorological parameterization scheme set based on the fitness.

[0107] In this step, the meteorological forecasting system determines the fitness of the first-layer surrogate auxiliary model for each first meteorological parameterization scheme based on the first forecast error corresponding to each first meteorological parameterization and the preset transformation rule, and updates the set of first meteorological parameterization schemes according to the fitness. Specifically, after obtaining the fitness of the first-layer surrogate auxiliary model for each first meteorological parameterization, the meteorological forecasting system compares the fitness of each first meteorological parameterization scheme. If there are different fitness values, it further compares the current cumulative iteration count with the preset iteration count. If the current cumulative iteration count is less than the preset iteration count, the set of first meteorological parameterization schemes is updated. The specific process of updating the first meteorological parameterization scheme set is as follows: n first meteorological parameterization schemes can be selected by roulette wheel selection, or by random selection, or by selecting the top n first meteorological parameterization schemes with higher fitness based on fitness ranking. After obtaining n first meteorological parameterization schemes, a new set of first meteorological parameterization schemes can be generated based on the obtained n first meteorological parameterization schemes by crossover (e.g., swapping the a1 parameter of scheme 1 with the a1 parameter of scheme 2), mutation (e.g., changing the a1 parameter of scheme 1 to the a11 parameter), or some combination of these schemes.

[0108] Step S2032: Input the updated first meteorological parameterization scheme set into the first layer proxy auxiliary model to update the fitness of the first layer proxy auxiliary model;

[0109] In this step, the weather forecasting system inputs each of the first meteorological parameterization schemes in the updated first meteorological parameterization scheme set into the first-layer surrogate auxiliary model to obtain the weather forecast result corresponding to each first meteorological parameterization scheme. Then, based on the weather forecast result corresponding to each first meteorological parameterization scheme and the actual meteorological data, the first forecast error corresponding to each first meteorological parameterization scheme is determined. The fitness of the first-layer surrogate auxiliary model for each first meteorological parameterization scheme is updated based on the first forecast error. The fitness of each updated first meteorological parameterization scheme is compared. If there are different fitness values, the current cumulative iteration count and the preset iteration count are further compared. If the current cumulative iteration count is less than the preset iteration count, the first meteorological parameterization scheme set is updated. The specific process of updating the first meteorological parameterization scheme set is the same as the above steps and will not be repeated here.

[0110] Step S2033, until the preset conditions are met, to obtain the first layer target agent auxiliary model.

[0111] In this step, the meteorological forecasting system iterates through steps S2031 to S2032 until it is determined that the fitness of each updated first meteorological parameterization scheme is the same, or until the current cumulative number of iterations is greater than the preset number of iterations. Then, it exits the iterative cycle and uses the first-layer surrogate auxiliary model obtained in the last iteration as the first-layer target surrogate auxiliary model.

[0112] Step S30: Generate a second-layer proxy auxiliary model based on the second meteorological parameterization scheme;

[0113] In this embodiment, after determining the second meteorological parameterization scheme, the meteorological forecasting system runs the numerical model WRF on the second meteorological parameterization scheme to obtain the meteorological forecasting results corresponding to the second meteorological parameterization scheme, and generates a second-layer surrogate auxiliary model based on the second meteorological parameterization scheme and the corresponding meteorological forecasting results.

[0114] Specifically, step S30 includes:

[0115] Step S301: Filter the parameters in the second meteorological parameterization scheme to determine the target parameter set;

[0116] In this step, the meteorological forecasting system filters the parameters in the second meteorological parameterization scheme. Based on expert experience, it selects the parameters that have a greater impact on the meteorological forecasting results from the parameters in the second meteorological parameterization scheme and determines them as the target parameter set.

[0117] Step S302: Determine the set of parameter values ​​corresponding to each target parameter in the target parameter set;

[0118] In this step, after the meteorological forecasting system determines the target parameter set corresponding to the second meteorological parameterization scheme, it obtains the value range corresponding to each target parameter in the target parameter set, and determines the parameter value set corresponding to each target parameter in the target parameter set based on the value range of each target parameter.

[0119] Further, step S302 includes:

[0120] Step S3021: Determine the value range corresponding to each target parameter in the target parameter set, and sample the value range to determine the parameter value set corresponding to each target parameter.

[0121] In this step, the meteorological forecasting system determines the value range of each target parameter in the target parameter set corresponding to the second meteorological parameterization scheme, and samples according to the value range to determine the parameter value set corresponding to each target parameter; optionally, Gaussian sampling is performed on the value range of each target parameter to determine the parameter value set corresponding to each target parameter; preferably, for the value range of each target parameter, a parameter value is sampled at preset intervals within the value range to determine the parameter value set corresponding to each target parameter.

[0122] Step S303: Determine the third meteorological forecast result corresponding to the second meteorological parameterization scheme based on the parameter value set, and generate a second-layer proxy auxiliary model based on the third meteorological forecast result and the second meteorological parameterization scheme.

[0123] In this step, the weather forecasting system runs a WRF numerical model on the parameter value set corresponding to the second weather parameterization scheme to obtain the third weather forecast result corresponding to the second weather parameterization scheme. Based on the third weather forecast result and the second weather parameterization scheme, a second-layer surrogate auxiliary model is generated. Specifically, the objective function of the second-layer surrogate auxiliary model is as follows:

[0124]

[0125] Where Y represents the model output, i.e., the meteorological forecast result corresponding to the second meteorological parameterization scheme, and X... j X k X i This represents different parameter settings, specifically the parameters included in the second meteorological parameterization scheme, a j a j，k a j，k，i The first, second, and third order relationships between parameters are represented by δ, a0 represents the bias (different second meteorological parameterization schemes will have different biases), δ represents the error (the error between the meteorological prediction results of the second-layer surrogate auxiliary model for the second meteorological parameterization scheme and the third prediction results of the numerical model for the second meteorological parameterization scheme), and n represents the number of parameters.

[0126] Step S40: Optimize the second-layer proxy auxiliary model and perform weather forecasting based on the optimized second-layer proxy auxiliary model.

[0127] In this step, the weather forecasting system optimizes the second-layer surrogate auxiliary model to obtain the second-layer target surrogate auxiliary model, and performs weather forecasting based on the second-layer target surrogate auxiliary model.

[0128] Specifically, the steps for optimizing the second-layer agent-assisted model include:

[0129] Step S401: Input the second meteorological parameterization scheme set into the second-layer proxy auxiliary model to obtain the fourth meteorological forecast result;

[0130] Step S402: Obtain actual meteorological data, and determine the second prediction error based on the actual meteorological data and the fourth meteorological forecast result;

[0131] Step S403: Optimize the second-layer proxy auxiliary model based on the second prediction error to determine the target parameter value corresponding to each target parameter in the parameter value set corresponding to each target parameter.

[0132] In steps S401 to S403, the meteorological forecasting system inputs each of the second meteorological parameterization schemes in the second meteorological parameterization scheme set into the second-layer surrogate auxiliary model to obtain the fourth meteorological forecast result corresponding to each second meteorological parameterization scheme. The meteorological forecasting system acquires the actual meteorological data of the target area, determines the second prediction error corresponding to each fourth meteorological forecast result based on the actual meteorological data, and optimizes the second-layer surrogate auxiliary model with the objective of minimizing the second prediction error using a multi-objective evolutionary optimization algorithm, such as SMS-EMOA. This outputs the most suitable parameter settings for the target area in the second meteorological parameterization scheme, and then determines the target parameter value corresponding to each target parameter from the set of parameter values ​​corresponding to each target parameter in the second meteorological parameterization scheme. Further, the target parameter value corresponding to each target parameter is added to the second-layer surrogate auxiliary model to obtain the second-layer target surrogate auxiliary model, and meteorological forecasting is performed based on the second-layer target surrogate auxiliary model. It should be noted that the optimization process of the second-layer surrogate auxiliary model is similar to the optimization process of the second-layer surrogate auxiliary model, and will not be repeated here.

[0133] The meteorological forecasting system in this embodiment acquires a first set of meteorological parameterization schemes and generates a first-layer surrogate auxiliary model based on the first set of meteorological parameterization schemes. The first-layer surrogate auxiliary model is optimized to determine a second meteorological parameterization scheme from the first set of meteorological parameterization schemes. A second-layer surrogate auxiliary model is generated based on the second meteorological parameterization scheme. The second-layer surrogate auxiliary model is optimized, and meteorological forecasting is performed based on the optimized second-layer surrogate auxiliary model. This invention constructs a first-layer surrogate auxiliary model and a second-layer surrogate auxiliary model based on the meteorological parameterization scheme and the corresponding meteorological forecasting results. The first-layer surrogate auxiliary model and the second-layer surrogate auxiliary model are optimized respectively to determine the meteorological parameterization scheme of the locally optimal mesoscale numerical weather prediction model, thereby improving the accuracy of meteorological forecasting.

[0134] Further, refer to Figure 2The second embodiment of the present invention is proposed. The difference between the second embodiment and the first embodiment is that, before the step of performing weather forecasting based on the optimized second-layer proxy auxiliary model, the following steps are included:

[0135] Step a: Input the target parameter value corresponding to each target parameter into the numerical mode to verify the optimized second-layer agent-assisted model and obtain the verification result;

[0136] Step b: If the verification result meets the preset conditions, then perform the following step: perform weather forecasting based on the optimized second-layer agent-assisted model.

[0137] In this embodiment, after determining the target parameter value corresponding to each target parameter in the second meteorological parameterization scheme, the meteorological forecasting system inputs the target parameter value corresponding to each target parameter into the numerical model to obtain the meteorological forecasting result of the second meteorological parameterization scheme for the target area over a historical period. The system then combines the meteorological forecasting result with the actual meteorological data of the target area over a historical period to verify the optimized second-layer proxy auxiliary model and obtain the verification result. If the verification result is determined to meet the preset conditions, that is, the prediction error between the meteorological forecasting result and the actual meteorological data is less than the preset threshold, then the optimized second-layer proxy auxiliary model is determined to be usable for predicting the meteorology of the target area, and the step of performing meteorological forecasting based on the optimized second-layer proxy auxiliary model is executed.

[0138] The weather forecasting system in this embodiment inputs the target parameter values ​​corresponding to each target parameter into a numerical model to verify the optimized second-layer surrogate auxiliary model and obtain the verification results. If the verification results meet the preset conditions, the system executes the step of performing weather forecasting based on the optimized second-layer surrogate auxiliary model. By inputting the target parameter values ​​corresponding to each target parameter into a numerical model to verify the optimized second-layer surrogate auxiliary model, the reliability of the optimized second-layer surrogate auxiliary model is improved, thereby helping to improve the accuracy of weather forecasting.

[0139] like Figure 3 As shown, the present invention also provides a weather forecasting device. The weather forecasting device of the present invention includes:

[0140] The acquisition module 101 is used to acquire a first set of meteorological parameterization schemes and generate a first-layer proxy auxiliary model based on the first set of meteorological parameterization schemes.

[0141] The determination module 102 is used to optimize the first-layer proxy auxiliary model in order to determine the second meteorological parameterization scheme in the first meteorological parameterization scheme set;

[0142] The generation module 103 is used to generate a second-layer proxy auxiliary model based on the second meteorological parameterization scheme;

[0143] The prediction module 104 is used to optimize the second-layer proxy auxiliary model and make weather predictions based on the optimized second-layer proxy auxiliary model.

[0144] Furthermore, the acquisition module is also used for:

[0145] Obtain the set of preset meteorological parameterization schemes and the meteorological types to be predicted;

[0146] The set of preset meteorological parameterization schemes is filtered according to the meteorological type to be predicted to obtain the first meteorological parameterization scheme.

[0147] Furthermore, the acquisition module is also used for:

[0148] Determine the types of schemes included in the first meteorological parameterization scheme set, and determine the number of schemes corresponding to each scheme type;

[0149] Based on the scheme type and the number of schemes, an orthogonal experiment is conducted on the first meteorological parameterization scheme set to obtain the target meteorological parameterization scheme set;

[0150] Determine the first meteorological forecast result corresponding to each target meteorological parameterization scheme in the target meteorological parameterization scheme set, and generate a first-layer surrogate auxiliary model based on the target meteorological parameterization scheme set and the first meteorological forecast result.

[0151] Furthermore, the determining module is also used for:

[0152] The first set of meteorological parameterization schemes is input into the first layer of proxy auxiliary model to obtain the second meteorological forecast result;

[0153] Acquire actual meteorological data, and determine a first prediction error based on the actual meteorological data and the first meteorological forecast result;

[0154] The first-layer proxy auxiliary model is optimized based on the first prediction error to obtain the first-layer target proxy auxiliary model, and the second meteorological parameterization scheme is determined in the first meteorological parameterization scheme set based on the first-layer target proxy auxiliary model.

[0155] Furthermore, the determining module also includes an optimization module, which is used to:

[0156] Based on the first prediction error, determine the fitness of the first-layer proxy auxiliary model, and update the first meteorological parameterization scheme set based on the fitness.

[0157] The updated set of first meteorological parameterization schemes is input into the first layer surrogate auxiliary model to update the fitness of the first layer surrogate auxiliary model;

[0158] The process continues until the preset conditions are met, at which point the first-layer target agent auxiliary model is obtained.

[0159] Furthermore, the generation module is also used for:

[0160] The parameters in the second meteorological parameterization scheme are screened to determine the target parameter set;

[0161] Determine the set of parameter values ​​corresponding to each target parameter in the target parameter set;

[0162] The third meteorological forecast result corresponding to the second meteorological parameterization scheme is determined based on the set of parameter values, and a second-layer proxy auxiliary model is generated based on the third meteorological forecast result and the second meteorological parameterization scheme.

[0163] Furthermore, the generation module is also used for:

[0164] Determine the value range corresponding to each target parameter in the target parameter set, and sample the value range to determine the parameter value set corresponding to each target parameter.

[0165] Furthermore, the prediction module also includes an optimization module, which is used to:

[0166] The second meteorological parameterization scheme set is input into the second-layer proxy auxiliary model to obtain the fourth meteorological forecast result;

[0167] Acquire actual meteorological data, and determine the second prediction error based on the actual meteorological data and the fourth meteorological forecast result;

[0168] The second-layer proxy auxiliary model is optimized based on the second prediction error to determine the target parameter value corresponding to each target parameter in the set of parameter values ​​corresponding to each target parameter.

[0169] Furthermore, the prediction module also includes a verification module, which is used for:

[0170] Input the target parameter value corresponding to each target parameter into the numerical pattern to verify the optimized second-layer agent auxiliary model and obtain the verification results;

[0171] If the verification result meets the preset conditions, then the following step is executed: perform weather forecasting based on the optimized second-layer agent-assisted model.

[0172] The present invention also provides a weather forecasting system.

[0173] The weather forecasting system includes: a memory, a processor, and a weather forecasting program stored in the memory and executable on the processor. When the weather forecasting program is executed by the processor, it implements the steps of the weather forecasting method as described above.

[0174] The method implemented when the weather forecasting program running on the processor is executed can be referred to in various embodiments of the weather forecasting method of the present invention, and will not be repeated here.

[0175] The present invention also provides a storage medium.

[0176] The storage medium stores a weather forecasting program, which, when executed by a processor, implements the steps of the weather forecasting method described above.

[0177] The method implemented when the weather forecasting program running on the processor is executed can be referred to in various embodiments of the weather forecasting method of the present invention, and will not be repeated here.

[0178] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or system. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or system that includes that element.

[0179] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0180] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0181] The above are merely preferred embodiments of the present invention and do not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A weather forecasting method, characterized in that, The meteorological forecasting method includes the following steps: Obtain a first set of meteorological parameterization schemes, and generate a first-layer proxy auxiliary model based on the first set of meteorological parameterization schemes; The first-layer proxy auxiliary model is optimized to determine the second meteorological parameterization scheme from the first meteorological parameterization scheme set; A second-layer proxy auxiliary model is generated based on the second meteorological parameterization scheme; The second-layer proxy auxiliary model is optimized, and weather forecasting is performed based on the optimized second-layer proxy auxiliary model; The step of generating the first-layer proxy auxiliary model based on the first meteorological parameterization scheme set includes: Determine the types of schemes included in the first meteorological parameterization scheme set, and determine the number of schemes corresponding to each scheme type; Based on the scheme type and the number of schemes, an orthogonal experiment is conducted on the first meteorological parameterization scheme set to obtain the target meteorological parameterization scheme set; Determine the first meteorological forecast result corresponding to each target meteorological parameterization scheme in the target meteorological parameterization scheme set, and generate a first-layer surrogate auxiliary model based on the target meteorological parameterization scheme set and the first meteorological forecast result; The step of generating the second-layer proxy auxiliary model according to the second meteorological parameterization scheme includes: The parameters in the second meteorological parameterization scheme are screened to determine the target parameter set; Determine the set of parameter values ​​corresponding to each target parameter in the target parameter set; The third meteorological forecast result corresponding to the second meteorological parameterization scheme is determined based on the set of parameter values, and a second-layer proxy auxiliary model is generated based on the third meteorological forecast result and the second meteorological parameterization scheme.

2. The weather forecasting method as described in claim 1, characterized in that, The steps for obtaining the first set of meteorological parameterization schemes include: Obtain the set of preset meteorological parameterization schemes and the meteorological types to be predicted; The set of preset meteorological parameterization schemes is filtered according to the meteorological type to be predicted to obtain the first meteorological parameterization scheme.

3. The weather forecasting method as described in claim 1, characterized in that, The step of optimizing the first-layer proxy auxiliary model to determine the second meteorological parameterization scheme from the first meteorological parameterization scheme set includes: The first set of meteorological parameterization schemes is input into the first layer of proxy auxiliary model to obtain the second meteorological forecast result; Acquire actual meteorological data, and determine a first prediction error based on the actual meteorological data and the second meteorological forecast result; The first-layer proxy auxiliary model is optimized based on the first prediction error to obtain the first-layer target proxy auxiliary model, and the second meteorological parameterization scheme is determined in the first meteorological parameterization scheme set based on the first-layer target proxy auxiliary model.

4. The weather forecasting method as described in claim 3, characterized in that, The step of optimizing the first-layer agent-assisted model based on the first prediction error to obtain the first-layer target agent-assisted model includes: Based on the first prediction error, determine the fitness of the first-layer proxy auxiliary model, and update the first meteorological parameterization scheme set based on the fitness. The updated set of first meteorological parameterization schemes is input into the first layer surrogate auxiliary model to update the fitness of the first layer surrogate auxiliary model; The process continues until the preset conditions are met, at which point the first-layer target agent auxiliary model is obtained.

5. The weather forecasting method as described in claim 1, characterized in that, The step of determining the set of parameter values ​​corresponding to each target parameter in the target parameter set includes: Determine the value range corresponding to each target parameter in the target parameter set, and sample the value range to determine the parameter value set corresponding to each target parameter.

6. The weather forecasting method as described in claim 1, characterized in that, The steps for optimizing the second-layer agent-assisted model include: The second meteorological parameterization scheme is input into the second-layer proxy auxiliary model to obtain the fourth meteorological forecast result; Acquire actual meteorological data, and determine the second prediction error based on the actual meteorological data and the fourth meteorological forecast result; The second-layer proxy auxiliary model is optimized based on the second prediction error to determine the target parameter value corresponding to each target parameter in the set of parameter values ​​corresponding to each target parameter.

7. The weather forecasting method as described in claim 6, characterized in that, Before the step of performing weather forecasting based on the optimized second-layer proxy-assisted model, the following steps are included: Input the target parameter value corresponding to each target parameter into the numerical pattern to verify the optimized second-layer agent auxiliary model and obtain the verification results; If the verification result meets the preset conditions, then the following step is executed: perform weather forecasting based on the optimized second-layer agent-assisted model.

8. A weather forecasting device, characterized in that, The weather forecasting device includes: The acquisition module is used to acquire a first set of meteorological parameterization schemes and generate a first-layer proxy auxiliary model based on the first set of meteorological parameterization schemes. The determination module is used to optimize the first-layer proxy auxiliary model in order to determine the second meteorological parameterization scheme from the first meteorological parameterization scheme set; The generation module is used to generate a second-layer proxy auxiliary model based on the second meteorological parameterization scheme; The prediction module is used to optimize the second-layer proxy auxiliary model and make weather forecasts based on the optimized second-layer proxy auxiliary model. The acquisition module is also used for: Determine the types of schemes included in the first meteorological parameterization scheme set, and determine the number of schemes corresponding to each scheme type; Based on the scheme type and the number of schemes, an orthogonal experiment is conducted on the first meteorological parameterization scheme set to obtain the target meteorological parameterization scheme set; Determine the first meteorological forecast result corresponding to each target meteorological parameterization scheme in the target meteorological parameterization scheme set, and generate a first-layer surrogate auxiliary model based on the target meteorological parameterization scheme set and the first meteorological forecast result; The generation module is also used for: The parameters in the second meteorological parameterization scheme are screened to determine the target parameter set; Determine the set of parameter values ​​corresponding to each target parameter in the target parameter set; The third meteorological forecast result corresponding to the second meteorological parameterization scheme is determined based on the set of parameter values, and a second-layer proxy auxiliary model is generated based on the third meteorological forecast result and the second meteorological parameterization scheme.

9. A weather forecasting system, characterized in that, The weather forecasting system includes: a memory, a processor, and a weather forecasting program stored in the memory and executable on the processor. When the weather forecasting program is executed by the processor, it implements the steps of the weather forecasting method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a weather forecasting program, which, when executed by a processor, implements the steps of the weather forecasting method as described in any one of claims 1 to 7.

Images