A multi-region-based numerical weather warning method and system

Abstract

The application provides a multi-region-based numerical weather warning method and system, and relates to the technical field of data processing.The method comprises the following steps: collecting the topographic conditions of a target region by means of unmanned aerial vehicle cruising, determining N regional topographic information for feature extraction, obtaining M regional features, collecting real-time data by means of a meteorological detector, determining N regional initial values, assigning numerical values according to the M regional features to obtain a feature numerical value set, traversing historical meteorological data, matching the meteorological data traversal result with the feature numerical value set, obtaining a meteorological feature numerical value set, generating a multi-level warning instruction, and performing multi-level weather warning according to the multi-level warning instruction.The application solves the technical problem that the weather features of multiple regions cannot be controlled in real time in the prior art, resulting in low weather warning efficiency and accuracy, and achieves rational and accurate control based on numerical weather, thereby improving the weather warning efficiency and accuracy.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method and system for numerical weather warning based on multiple regions. Background Technology

[0002] With the development of weather warning technology, weather warnings are used to provide advance warnings of weather disasters. Weather disasters refer to those caused by meteorological factors that severely damage human life, property, and the natural environment. Common weather disasters include typhoons, rainstorms, floods, flash floods, mudslides, droughts, sandstorms, lightning, hail, and smog. While the causes and formation mechanisms of these disasters vary, they are all closely related to meteorological factors. However, current technologies suffer from the inability to monitor weather characteristics in multiple regions in real time, resulting in low efficiency and accuracy of weather warnings. Summary of the Invention

[0003] This application provides a multi-regional numerical weather warning method and system to address the technical problem in the prior art that the inability to control weather characteristics in multiple regions in real time leads to low efficiency and accuracy of weather warnings.

[0004] In view of the above problems, this application provides a numerical weather warning method and system based on multiple regions.

[0005] Firstly, this application provides a multi-regional numerical weather warning method, the method comprising: collecting terrain information of a target area through a drone patrol to determine N regional terrain information; traversing the N regional terrain information to extract features and obtain M regional features; determining N regional initial values ​​based on real-time data of the N regional terrain information collected by a meteorological detector; obtaining a feature value set, the feature value set being obtained by assigning values ​​to the N regional initial values ​​based on the M regional features; traversing historical meteorological data and matching the meteorological data traversal results with the feature value set to obtain a meteorological feature value set; generating multi-level alarm instructions based on the meteorological feature value set; and issuing multi-level weather warnings based on the multi-level alarm instructions.

[0006] Secondly, this application provides a multi-regional numerical weather warning system, the system comprising: an information determination module, which is used to collect the terrain information of the target area through the drone's patrol and determine the terrain information of N regions; a feature extraction module, which is used to traverse the terrain information of the N regions to extract features and obtain M regional features; a data acquisition module, which is used to determine the initial values ​​of the N regions based on the real-time data of the terrain information of the N regions collected by the meteorological detector; a numerical assignment module, which is used to obtain a feature value set, which is obtained by assigning values ​​to the initial values ​​of the N regions based on the M regional features; a matching module, which is used to traverse historical meteorological data and match the meteorological data traversal results with the feature value set to obtain a meteorological feature value set; and a weather warning module, which is used to generate multi-level alarm instructions based on the meteorological feature value set and to issue multi-level weather warnings based on the multi-level alarm instructions.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] This application provides a multi-regional numerical weather warning method and system, which relates to the field of data processing technology. It solves the technical problem that the existing technology cannot control the weather characteristics of multiple regions in real time, resulting in low efficiency and accuracy of weather warnings. It realizes rational and precise control based on numerical weather, and improves the efficiency and accuracy of weather warnings. Attached Figure Description

[0009] Figure 1 This application provides a schematic flowchart of a multi-regional numerical weather warning method;

[0010] Figure 2 This application provides a schematic diagram of a multi-regional numerical weather warning system.

[0011] Explanation of reference numerals in the attached diagram: Information determination module 1, feature extraction module 2, data acquisition module 3, numerical assignment module 4, matching module 5, weather warning module 6. Detailed Implementation

[0012] This application provides a numerical weather warning method and system based on multiple regions to solve the technical problem that the existing technology cannot control the weather characteristics of multiple regions in real time, resulting in low efficiency and accuracy of weather warnings.

[0013] Example 1

[0014] like Figure 1As shown in the figure, this application provides a multi-regional numerical weather warning method. This method is applied to a multi-regional numerical weather warning system, which is communicatively connected to a drone and a meteorological detector. The method includes:

[0015] Step A100: Collect terrain information of the target area through the drone patrol, and determine the terrain information of N areas;

[0016] Furthermore, step A300 of this application also includes:

[0017] Step A110: Determine the drone's cruise route based on the target area map;

[0018] Step A120: Acquire images in the target area according to the cruise route to obtain a set of area images;

[0019] Step A130: Analyze the terrain of the target area in the image set of the region and determine the terrain information of the N regions.

[0020] In this application, the numerical weather warning method based on multiple regions provided in the embodiments of this application is applied to a numerical weather warning system based on multiple regions. The numerical weather warning system based on multiple regions is communicatively connected to a drone and a meteorological detector, which are used to collect meteorological environmental parameters.

[0021] Furthermore, the drone patrol to collect terrain information of the target area refers to first determining the drone's patrol route based on a map of the target area. This involves creating a map of the target area, identifying target locations with various terrain features, and then determining the drone's patrol route based on the actual terrain. During map creation, local weather conditions, such as wind speed, precipitation, cloud cover, and humidity, need to be considered to adjust the drone's patrol route accordingly. Simultaneously, images are collected from multiple points within the target area along the patrol route. Terrain analysis is then performed on the image set transmitted back by the drone along the patrol route. This involves analyzing and determining the surface terrain features and elevation variations within the collected image set. Based on this analysis, N regional terrain information points are obtained within the target area, where N is a positive integer greater than 1. This information serves as an important reference for later implementation of multi-regional numerical weather prediction for weather warnings.

[0022] Step A200: Traverse the geomorphological information of the N regions to extract features and obtain M regional features;

[0023] In this application, in order to provide more accurate weather warnings for N regions, it is necessary to traverse and access each regional geomorphological data node contained in the geomorphological information of the N regions. Geomorphological features are extracted based on the geomorphological information contained in the N regions. These geomorphological features can be mountainous, plateau, basin, plain, hilly, etc. Regional features corresponding to the N regions are extracted from the geomorphological conditions of different regions and summarized as M regional features, where M is a positive integer less than or equal to N. This provides a guarantee for realizing weather warnings based on numerical weather prediction for multiple regions.

[0024] Step A300: Determine the initial values ​​for the N regions based on the real-time data of the topographic information collected by the meteorological detector.

[0025] Furthermore, step A300 of this application also includes:

[0026] Step A310: Collect real-time atmospheric data in the first area using the meteorological detector, wherein the real-time atmospheric data includes precipitation data, wind direction data, cloud cover data, and temperature data;

[0027] Step A320: Perform weight training on the real-time atmospheric data to obtain weight training data;

[0028] Step A330: Use the weighted training data as the initial atmospheric value in the first region, and iterate thereafter to determine the initial values ​​for the N regions.

[0029] In this application, to more accurately determine weather values, real-time atmospheric data is first collected in a first region using a meteorological detector. The first region refers to a randomly selected area from N regions, and the real-time atmospheric data includes precipitation, wind direction, cloud cover, and temperature data. Before issuing a weather warning, weight training is required based on different meteorological data collected by the meteorological detector. This weight training requires a large amount of data aggregation and precise weight determination before targeted calculation. For example, the weight ratios of precipitation, wind direction, cloud cover, and temperature data can be: First influence coefficient: Second influence coefficient: Third influence coefficient: Fourth influence coefficient is 3: With a weight ratio of 4:2:5, the impact parameters after the weight training process are respectively: first impact parameter * 0.3, second impact parameter * 0.4, third impact parameter * 0.2, and fourth impact parameter * 0.5. Based on this weight training data, the weather conditions requiring warning are determined. Furthermore, the weight training results are used as the initial atmospheric values ​​in the first region. On this basis, a second region is extracted from N regions. The second region is a region randomly extracted from the N regions, and the second region is different from the first region. Similarly, the initial atmospheric values ​​in the second region are determined. Thus, the N regions are traversed and calculated to determine the initial values ​​of the N regions, laying a solid foundation for subsequent implementation of weather warnings based on numerical weather data from multiple regions.

[0030] Step A400: Obtain the feature value set, which is obtained by assigning values ​​to the N initial values ​​of the regions based on the features of the M regions;

[0031] Furthermore, step A400 of this application also includes:

[0032] Step A410: Construct a feature numerical subnetwork, synchronize the M regional features to the numerical allocation layer within the feature numerical subnetwork, allocate the numerical values ​​using the N initial values ​​of the regions, and obtain the feature numerical set. The feature numerical subnetwork includes a data input layer, a numerical allocation layer, and a data output layer.

[0033] Step A420: Label the features of the M regions to obtain a constructed dataset, wherein the constructed dataset includes a training set and a validation set;

[0034] Step A430: Supervised training and validation of the feature numerical subnetwork are performed using the training set and validation set until the feature numerical subnetwork meets the convergence algorithm, and then the feature numerical subnetwork is obtained.

[0035] In this application, to ensure the accuracy of regional numerical allocation based on regional characteristics across N regions within the target area, a feature numerical subnetwork is first constructed. This subnetwork includes a data input layer, a numerical allocation layer, and a data output layer. Furthermore, different regional parameters are assigned to the M regional features, i.e., data annotation, to obtain the constructed dataset. The feature numerical subnetwork is a neural subnetwork in machine learning that can continuously iterate and optimize itself. It is trained using a training set and a validation dataset. Each training dataset in the constructed dataset includes a training set and a validation set corresponding to the M regional features, with the validation set being the validation data that corresponds one-to-one with the training set.

[0036] Furthermore, the feature numerical subnetwork construction process is as follows: each set of training data in the training set is input into the feature numerical subnetwork, and the output of the feature numerical subnetwork is adjusted by the validation data corresponding to this set of training data. When the output result of the feature numerical subnetwork is consistent with the validation, the training of the current set ends. The training of the feature numerical subnetwork is completed when all the training data in the training set is completed.

[0037] To ensure the convergence and accuracy of the feature numerical subnetwork, the convergence process can be defined as the output data in the feature numerical subnetwork converging at a point through a convergence algorithm. Convergence is considered achieved when the data approaches a certain value. The convergence algorithm is as follows:

[0038]

[0039] Among them, S j Let i be the convergence value of the feature-valued subnetwork when it conforms to the convergence algorithm, j be the first feature value, and w be the second feature value. ij Let b be the weight between i and j. j The threshold value for j.

[0040] The first feature value is the numerical value assigned to any one of the M regional features. The second feature value is the numerical value assigned to any one of the M regional features that is different from the first feature. The convergence value when the feature numerical subnetwork meets the convergence algorithm is determined by the above convergence algorithm. At the same time, the accuracy of the feature numerical subnetwork can be tested by testing the feature numerical subnetwork with a test dataset. For example, the test accuracy can be set to 80%. When the test accuracy of the test dataset meets 80%, the feature numerical subnetwork is completed.

[0041] Finally, the features of M regions are synchronized to the numerical allocation layer in the constructed feature numerical sub-network. The initial values ​​of the N regions are used to allocate values. The different values ​​allocated by the numerical allocation layer indicate the regional numerical situation under different features, thereby obtaining the feature numerical set. This plays a role in promoting weather early warning based on numerical weather of multiple regions.

[0042] Step A500: Traverse the historical meteorological data, and match the meteorological data traversal results with the feature value set to obtain the meteorological feature value set;

[0043] In this application, the system retrieves meteorological data of the target area within a historical time period and uses the time sequence of the recorded meteorological data as the traversal route to sequentially access each meteorological data node contained in the historical meteorological data. The traversal access result is recorded as the meteorological data traversal result. Furthermore, the meteorological data traversal result is matched with the feature dataset. This means that the similarity between the meteorological features contained in the meteorological data traversal result and the features contained in the feature value set is verified. If the similarity between the meteorological features contained in the meteorological data traversal result and the features contained in the feature value set is greater than 80%, then the value corresponding to the matched feature in the feature value set is assigned to the meteorological features in the meteorological data traversal result and output. This process is repeated iteratively, and the features in the meteorological data traversal result are sequentially assigned values ​​and recorded as the meteorological feature value set for output. This serves as reference data for subsequent weather warnings based on numerical weather data from multiple regions.

[0044] Step A600: Generate multi-level alarm instructions based on the meteorological feature data set, and issue multi-level weather warnings based on the multi-level alarm instructions.

[0045] Furthermore, step A600 of this application also includes:

[0046] Step A610: Serialize the meteorological feature data set according to the numerical value to obtain a meteorological feature data set sequence;

[0047] Step A620: Match the alarm intensity based on the sequence of the meteorological feature data set, and generate a first alarm command according to the alarm intensity;

[0048] Step A630: Control the system by using the first alarm command to search for matching weather data and issue a first weather warning based on the matching weather data.

[0049] In this application, to improve the accuracy of weather warnings, the meteorological feature data set is first serialized according to its numerical value. This means that the meteorological feature values ​​within the set are sorted in descending order according to their numerical values, and the meteorological feature data set with the descending order sequence is denoted as the meteorological feature data set sequence. Further, the warning intensity is matched based on the meteorological feature data set sequence. This means that the warning intensity to be matched is first divided into four levels: Level 1, Level 2, Level 3, and Level 4. Level 1 is the highest warning level, indicating that the weather will be extremely severe; Level 2 is the next lowest warning level, indicating that the weather will be severe; Level 3 is the next lowest warning level, indicating that the weather will be relatively severe; and Level 4 is the next lowest warning level, indicating that the weather will be moderate. Example The meteorological feature in the first order of the meteorological feature numerical set sequence is the strongest and has the highest value. Therefore, the alarm intensity to be matched is level one. Further, a first alarm command is generated based on the current alarm intensity, and there is a one-to-one correspondence between the first alarm command and the alarm intensity. This determines the multi-level alarm command. At the same time, the system uses the first alarm command matched with the current alarm intensity from the multi-level alarm command to search for weather data that matches the current alarm intensity in the big data, extracts it, and records it as the matching weather data that will be issued as an early warning. Finally, the first weather warning is issued to the corresponding area in N regions based on the matching weather data. On this basis, the same iteration is performed, and multiple levels of weather warnings are issued to N regions according to the multi-level alarm command, which improves the accuracy of weather warning based on numerical weather data of multiple regions in the later stage.

[0050] In summary, the numerical weather warning method based on multiple regions provided in this application has at least the following technical effects: it realizes rational and precise control based on numerical weather, and improves the efficiency and accuracy of weather warnings.

[0051] Example 2

[0052] Based on the same inventive concept as the multi-regional numerical weather warning method in the foregoing embodiments, such as Figure 2 As shown, this application provides a multi-regional numerical weather warning system, the system comprising:

[0053] Information determination module 1 is used to collect the terrain information of the target area through the drone's patrol and determine the terrain information of N areas.

[0054] Feature extraction module 2 is used to traverse the N regional geomorphological information to extract features and obtain M regional features;

[0055] Data acquisition module 3 is used to determine the initial values ​​of N regions based on the real-time data of the topographic information of the N regions collected by the meteorological detector;

[0056] Numerical assignment module 4 is used to obtain a feature numerical set, which is obtained by assigning numerical values ​​to the N initial values ​​of the regions based on the characteristics of the M regions.

[0057] Matching module 5 is used to traverse historical meteorological data and match the meteorological data traversal results with the feature value set to obtain the meteorological feature value set.

[0058] Weather warning module 6 is used to generate multi-level alarm instructions based on the meteorological feature data set, and to issue multi-level weather warnings based on the multi-level alarm instructions.

[0059] Furthermore, the system also includes:

[0060] A route planning module, which is used to plan the cruise route of the UAV based on a target area map;

[0061] Image acquisition module, the image acquisition module is used to acquire images in the target area according to the cruise route, and obtain a set of regional images;

[0062] The terrain information module is used to analyze the terrain of the target area in the image set of the region and determine the terrain information of the N regions.

[0063] Furthermore, the system also includes:

[0064] The real-time data module is used to collect real-time atmospheric data in the first area through the meteorological detector, wherein the real-time atmospheric data includes precipitation data, wind direction data, cloud cover data, and temperature data.

[0065] A weight training module is used to perform weight training on the real-time atmospheric data to obtain weight training data.

[0066] An iterative module is used to take the weighted training data as the initial atmospheric value in the first region, and then iterate to determine the initial values ​​of the N regions.

[0067] Furthermore, the system also includes:

[0068] The sub-network module is used to construct a feature numerical sub-network, synchronize the features of the M regions to the numerical allocation layer within the feature numerical sub-network, and perform numerical allocation through the initial values ​​of the N regions to obtain the feature numerical set. The feature numerical sub-network includes a data input layer, a numerical allocation layer, and a data output layer.

[0069] A data annotation module is used to annotate the features of the M regions to obtain a constructed dataset, wherein the constructed dataset includes a training set and a validation set;

[0070] A convergence module is used to supervise the training and validation of the feature numerical subnetwork using the training set and validation set until the feature numerical subnetwork meets the convergence algorithm, and then obtain the feature numerical subnetwork.

[0071] Furthermore, the system also includes:

[0072] A serialization processing module is used to serialize the meteorological feature value set according to the value size to obtain a meteorological feature value set sequence.

[0073] An alarm intensity matching module is used to match alarm intensity based on the sequence of meteorological feature values ​​and generate a first alarm command based on the alarm intensity.

[0074] The early warning module is used to control the system to search for matching weather data through the first alarm command, and to issue a first weather warning based on the matching weather data.

[0075] Through the foregoing detailed description of a multi-regional numerical weather warning method, those skilled in the art can clearly understand that this embodiment provides a multi-regional numerical weather warning system. As the device disclosed in the embodiment corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant details can be found in the method section.

[0076] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A numerical weather warning method based on multiple regions, characterized in that, The method is applied to a multi-regional numerical weather warning system, which is communicatively connected to a drone and a meteorological instrument. The method includes: The drone patrol collects terrain information of the target area to determine the terrain information of N areas. By traversing the geomorphological information of the N regions, feature extraction is performed to obtain M regional features; Based on the real-time data of the geomorphological information of the N regions collected by the meteorological detector, the initial values ​​of the N regions are determined. A feature value set is obtained by assigning numerical values ​​to the N initial values ​​of the regions based on the characteristics of the M regions. Historical meteorological data is traversed, and the meteorological feature value set is matched with the meteorological data traversal result to obtain the meteorological feature value set. The meteorological feature value set is obtained by assigning values ​​to the features in the meteorological data traversal result in sequence and then outputting the meteorological feature value set. Multi-level alarm commands are generated based on the meteorological feature data set, and multi-level weather warnings are issued based on the multi-level alarm commands. Based on real-time data of geomorphological information collected by the meteorological detector in the N regions, initial values ​​for the N regions are determined. The method further includes: The meteorological detector collects real-time atmospheric data in the first area, including precipitation data, wind direction data, cloud cover data, and temperature data. The real-time atmospheric data is used for weight training to obtain weight training data; The weighted training data is used as the initial atmospheric value in the first region, and the initial values ​​of the N regions are determined iteratively. The method also includes: A feature numerical subnetwork is constructed, and the features of the M regions are synchronized to the numerical allocation layer within the feature numerical subnetwork. The numerical allocation is performed using the initial values ​​of the N regions to obtain the feature numerical set. The feature numerical subnetwork includes a data input layer, a numerical allocation layer, and a data output layer. Data annotation is performed on the features of the M regions to obtain a constructed dataset, wherein the constructed dataset includes a training set and a validation set; The feature numerical subnetwork is trained and validated using the training set and validation set until it meets the convergence algorithm, at which point the feature numerical subnetwork is obtained.

2. The method as described in claim 1, characterized in that, The method involves using drones to patrol and collect terrain information of a target area, determining terrain information for N areas. Other methods include: The drone's cruise route is determined based on a map of the target area; The cruise route is used to acquire images in the target area, resulting in a set of regional images. Analyze the terrain of the target area in the image set of the region to determine the terrain information of the N regions.

3. The method as described in claim 1, characterized in that, The convergence algorithm is as follows: in, Let be the convergence value of the feature-valued subnetwork when it conforms to the convergence algorithm, where i is the first feature value and j is the second feature value. The weights between i and j The threshold value for j.

4. The method as described in claim 1, characterized in that, The method further includes generating multi-level alarm intensity commands based on the meteorological feature data set: The meteorological feature data set is serialized according to its numerical value to obtain a meteorological feature data set sequence. Based on the sequence matching of the meteorological feature data set, an alarm intensity is generated, and a first alarm command is generated according to the alarm intensity. The system searches for matching weather data using the first alarm command and issues a first weather warning based on the matching weather data.

5. A numerical weather warning system based on multiple regions, characterized in that, The multi-regional numerical weather warning system is communicatively connected to drones and meteorological detectors. The system includes: An information determination module is used to collect terrain information of N areas by the drone's patrol and collect terrain information of the target area. The feature extraction module is used to traverse the N regional geomorphological information to extract features and obtain M regional features; The data acquisition module is used to determine the initial values ​​of the N regions based on the real-time data of the topographic information of the N regions collected by the meteorological detector. A numerical assignment module is used to obtain a feature numerical set, which is obtained by assigning numerical values ​​to the N initial values ​​of the regions based on the features of the M regions. The matching module is used to traverse historical meteorological data, match the meteorological data traversal results with the feature value set, and obtain the meteorological feature value set. The meteorological feature value set is obtained by assigning values ​​to the features in the meteorological data traversal results in sequence and then outputting the meteorological feature value set. A weather warning module is used to generate multi-level alarm instructions based on the meteorological feature data set, and to issue multi-level weather warnings based on the multi-level alarm instructions. The system also includes: The real-time data module is used to collect real-time atmospheric data in the first area through the meteorological detector, wherein the real-time atmospheric data includes precipitation data, wind direction data, cloud cover data, and temperature data. A weight training module is used to perform weight training on the real-time atmospheric data to obtain weight training data. An iterative module is used to take the weighted training data as the initial atmospheric value in the first region, and iterate thereto determine the initial values ​​of the N regions. The sub-network module is used to construct a feature numerical sub-network, synchronize the features of the M regions to the numerical allocation layer within the feature numerical sub-network, and perform numerical allocation through the initial values ​​of the N regions to obtain the feature numerical set. The feature numerical sub-network includes a data input layer, a numerical allocation layer, and a data output layer. A data annotation module is used to annotate the features of the M regions to obtain a constructed dataset, wherein the constructed dataset includes a training set and a validation set; A convergence module is used to supervise the training and validation of the feature numerical subnetwork using the training set and validation set until the feature numerical subnetwork meets the convergence algorithm, and then obtain the feature numerical subnetwork.

Images