Typhoon generation prediction method, device, equipment and medium
The method combines spatiotemporal and target detection neural networks to efficiently and accurately predict typhoon generation time and location, overcoming the inefficiencies and inaccuracies of current forecasting techniques.
Patent Information
- Application Number
- CN202411237961.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-05
- Publication Date
- 2024-12-03
AI Technical Summary
Existing typhoon prediction methods struggle to simultaneously achieve efficient and accurate forecasting of typhoon generation time and location, with numerical weather forecasting requiring significant computational resources and historical similarity methods providing low accuracy.
A typhoon generation prediction method utilizing a spatiotemporal prediction neural network for meteorological data and a target detection neural network for typhoon recognition, dividing the problem into meteorological prediction and typhoon identification, and sequentially training these networks to create a multi-source meteorological prediction model and typhoon recognition model.
Enables high-efficiency and high-accuracy prediction of typhoon generation time and location by integrating spatiotemporal and target detection neural networks, effectively addressing the limitations of existing methods.
Smart Images

Figure CN119065033A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of meteorological forecasting, and in particular to a typhoon generation forecasting method, device, equipment and medium. Background Art
[0002] Typhoons are one of the most common and most harmful natural disasters in my country's southeastern coastal areas. The occurrence of typhoon weather will cause significant losses to the social economy and threaten people's lives and property. Therefore, efficient and accurate prediction of the time and location of typhoon formation is of great practical significance for enhancing social disaster prevention and mitigation capabilities, improving public safety, and guiding scientific decision-making and resource management.
[0003] In the relevant technical scheme, there are two methods for predicting typhoon formation: the first is the numerical weather forecast method, and the second is the similarity weather retrieval method. The numerical weather forecast method is the most commonly used method in modern meteorological forecasting. This method derives future atmospheric changes through given initial conditions and boundary values and atmospheric dynamics differential equations. It has a high accuracy rate, but it has a large demand for computing resources and a long calculation time, and requires the support of a large computer platform. The similarity weather retrieval method is to find weather cases similar to the current moment in historical data, and to approximate the current typhoon formation process through the historical weather evolution process. This method is highly dependent on rich historical data, and can only give approximate results, with a low accuracy rate.
[0004] The above two methods are difficult to simultaneously meet the needs of efficiently and accurately predicting the time and location of typhoon generation. Therefore, how to achieve these requirements in the typhoon generation prediction task is a technical problem that people in this field urgently need to solve. Summary of the invention
[0005] The purpose of the present invention is to provide a typhoon generation prediction method, device, equipment and medium, which can simultaneously meet the needs of efficiently and accurately predicting the typhoon generation time and location.
[0006] In order to solve the above technical problems, the present invention provides a typhoon generation prediction method, the method comprising:
[0007] Obtain and process a variety of original meteorological data at different historical moments in the target area to obtain a meteorological sample set;
[0008] Inputting the meteorological data sample set into a spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model; the multi-source meteorological prediction model is the first part of the typhoon generation prediction model;
[0009] Acquire and process historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with typhoon locations;
[0010] The typhoon data training set is input into the target detection neural network for training and testing to obtain a typhoon recognition model; the typhoon recognition model is the second part of the typhoon generation prediction model;
[0011] The meteorological data to be measured in the target area is input into the typhoon generation prediction model, and the time and location of future typhoon generation are output.
[0012] In the first aspect, in the above-mentioned typhoon generation prediction method provided by the present invention, a variety of original meteorological data of the target area at different historical moments are obtained and processed to obtain a meteorological sample set, including:
[0013] Acquire multi-source meteorological data including satellite data, ocean temperature data, and ERA5 multiple meteorological element data at different historical moments of the target area to obtain the original meteorological data;
[0014] Deleting noise data in the original meteorological data and supplementing missing data to obtain preprocessed data;
[0015] Arrange and group the preprocessed data in chronological order, each group having data of a set number of consecutive moments, and obtain a total of multiple groups of sample data; the data of the set number of consecutive moments includes input feature data and predicted label data;
[0016] The multiple groups of sample data are divided into a training set and a test set according to a set ratio to obtain the meteorological sample set.
[0017] On the other hand, in the above-mentioned typhoon generation prediction method provided by the present invention, the meteorological data sample set is input into the spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model, including:
[0018] Inputting the training set in the meteorological data sample set into the spatiotemporal prediction neural network for training to obtain a trained first model;
[0019] The test set in the meteorological data sample set is input into the first model for testing, the test result is obtained and it is judged whether the training of the first model is completed according to the specified threshold parameters; if so, the multi-source meteorological prediction model is obtained; if not, retraining is performed.
[0020] On the other hand, in the above-mentioned typhoon generation prediction method provided by the present invention, the historical typhoon meteorological data of the target area is obtained and processed to obtain a typhoon data sample set marked with the typhoon position, including:
[0021] Acquire satellite image data of the target area during the early stages of typhoon formation to obtain historical typhoon meteorological data;
[0022] The typhoon meteorological data are processed to obtain a typhoon data sample set marked with typhoon positions.
[0023] On the other hand, in the above-mentioned typhoon generation prediction method provided by the present invention, the typhoon meteorological data is processed to obtain a typhoon data sample set marked with a typhoon position, including:
[0024] Performing feature engineering on the typhoon meteorological data, marking the location of the typhoon on the typhoon meteorological data as a data label, and obtaining an initial typhoon sample set;
[0025] The initial typhoon sample set is divided into a training set and a test set according to a specified ratio to obtain the typhoon data sample set.
[0026] On the other hand, in the above-mentioned typhoon generation prediction method provided by the present invention, the typhoon data training set is input into the target detection neural network for training and testing to obtain a typhoon recognition model, including:
[0027] Inputting the training set in the typhoon data sample set into the target detection neural network for training to obtain a trained second model;
[0028] The test set in the typhoon data sample set is input into the second model for testing, the test result is obtained and it is judged whether the training of the second model is completed according to the specified threshold parameters; if so, the typhoon recognition model is obtained; if not, retraining is performed.
[0029] On the other hand, in the above-mentioned typhoon generation prediction method provided by the present invention, the meteorological data to be measured in the target area is input into the typhoon generation prediction model, and the time and location of future typhoon generation are output, including:
[0030] Inputting the meteorological data to be measured in the target area into the multi-source meteorological prediction model of the typhoon generation prediction model to obtain meteorological prediction results at multiple moments in the future;
[0031] The satellite image data in the meteorological forecast results are sequentially input into the typhoon identification model of the typhoon generation prediction model in chronological order to obtain the time and location of future typhoon generation.
[0032] In order to solve the above technical problems, the present invention further provides a typhoon generation prediction device, the device comprising:
[0033] A meteorological sample set acquisition module is used to acquire and process a variety of original meteorological data at different historical moments in the target area to obtain a meteorological sample set;
[0034] A multi-source meteorological prediction model training module is used to input the meteorological data sample set into the spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model; the multi-source meteorological prediction model is the first part of the typhoon generation prediction model;
[0035] A typhoon data sample set acquisition module is used to acquire and process the historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with the typhoon position;
[0036] A typhoon identification model training module is used to input the typhoon data training set into the target detection neural network for training and testing to obtain a typhoon identification model; the typhoon identification model is the second part of the typhoon generation prediction model;
[0037] The typhoon generation prediction model inference module is used to input the meteorological data to be measured in the target area into the typhoon generation prediction model and output the time and location of future typhoon generation.
[0038] In order to solve the above technical problems, the present invention further provides a typhoon generation prediction device, the device comprising:
[0039] Memory for storing computer programs;
[0040] A processor is used to implement the steps of the above-mentioned typhoon generation prediction method when executing the computer program.
[0041] In order to solve the above technical problem, the present invention also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned typhoon generation prediction method are implemented.
[0042] It can be seen from the above technical scheme that the present invention provides a typhoon generation prediction method, which includes: obtaining and processing a variety of original meteorological data at different historical moments in the target area to obtain a meteorological sample set; inputting the meteorological data sample set into a spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model; the multi-source meteorological prediction model is the first part of the typhoon generation prediction model; obtaining and processing the historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with the typhoon position; inputting the typhoon data training set into a target detection neural network for training and testing to obtain a typhoon recognition model; the typhoon recognition model is the second part of the typhoon generation prediction model; inputting the meteorological data to be tested in the target area into the typhoon generation prediction model, and outputting the time and location of future typhoon generation.
[0043] The beneficial effect of the present invention lies in that the above-mentioned typhoon generation prediction method provided by the present invention divides the typhoon generation prediction problem into two problems, namely meteorological prediction and typhoon identification, and solves them in sequence, respectively training and testing the spatiotemporal prediction neural network and the target detection neural network to obtain a typhoon generation prediction model including a multi-source meteorological prediction model and a typhoon identification model, and inputting the meteorological data to be measured in the target area into the typhoon generation prediction model, the time and location of future typhoon generation can be predicted, with high efficiency and high accuracy, thus simultaneously meeting the needs of efficiently and accurately predicting the time and location of typhoon generation.
[0044] In addition, the present invention also provides a corresponding typhoon generation prediction device, a typhoon generation prediction equipment and a computer-readable storage medium for the typhoon generation prediction method, which have the same or corresponding technical features as the above-mentioned typhoon generation prediction method and have the same effects as above. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] In order to more clearly illustrate the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0046] Figure 1 A flow chart of a typhoon generation prediction method provided by an embodiment of the present invention;
[0047] Figure 2 A flowchart for obtaining a multi-source weather forecast model provided by an embodiment of the present invention;
[0048] Figure 3 A flow chart for obtaining a typhoon identification model provided by an embodiment of the present invention;
[0049] Figure 4 An overall framework diagram of a typhoon generation prediction model based on multi-source meteorological data provided by an embodiment of the present invention;
[0050] Figure 5 A schematic diagram of prediction results of a typhoon formation prediction model provided by an embodiment of the present invention;
[0051] Figure 6 A schematic diagram of the structure of a typhoon generation prediction device provided by an embodiment of the present invention;
[0052] Figure 7 A schematic diagram of the structure of a typhoon generation prediction device provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0053] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0054] In order to enable those skilled in the art to better understand the solution of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods. Figure 1 A flow chart of a typhoon generation prediction method provided by an embodiment of the present invention, such as Figure 1 As shown, the method includes:
[0055] S101. Acquire and process a variety of original meteorological data of a target area at different historical moments to obtain a meteorological sample set.
[0056] In the implementation, first obtain a variety of meteorological data at different historical moments in the target area to obtain the original meteorological data, and then process the original meteorological data to obtain the required meteorological sample set. The target area can be a specific geographical space, that is, the geographical area to be measured. The various meteorological data can include any combination of satellite data, ocean temperature data, ERA5 multiple meteorological element data (including wind speed, sea level pressure, etc.) and other meteorological product data. The time resolution of the original meteorological data can be hours.
[0057] S102, inputting the meteorological data sample set into the spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model; the multi-source meteorological prediction model is the first part of the typhoon generation prediction model.
[0058] It should be noted that the spatiotemporal prediction neural network can use a variety of deep learning models such as SwinLSTM network and PredRNN network.
[0059] S103: Acquire and process historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with typhoon locations.
[0060] In implementation, the historical typhoon meteorological data corresponding to the same target area as step S101 is first obtained, and then the historical typhoon meteorological data is processed to obtain a typhoon data sample set marked with the typhoon position.
[0061] S104, inputting the typhoon data training set into the target detection neural network for training and testing to obtain a typhoon recognition model; the typhoon recognition model is the second part of the typhoon generation prediction model.
[0062] It should be noted that the target detection neural network can use a variety of deep learning models such as DETR network, YOLO network, Fast R-CNN network, etc.
[0063] S105, inputting the meteorological data to be measured in the target area into a typhoon generation prediction model, and outputting the time and location of future typhoon generation.
[0064] In the above-mentioned typhoon generation prediction method provided by the embodiment of the present invention, the typhoon generation prediction problem is divided into two problems, namely meteorological prediction and typhoon identification, which are solved in sequence. The spatiotemporal prediction neural network and the target detection neural network are trained and tested respectively to obtain a typhoon generation prediction model including a multi-source meteorological prediction model and a typhoon identification model. The meteorological data to be measured in the target area is input into the typhoon generation prediction model, so that the time and location of future typhoon generation can be predicted with high efficiency and accuracy, thereby meeting the needs of efficiently and accurately predicting the time and location of typhoon generation.
[0065] Furthermore, in a specific implementation, in the above-mentioned typhoon formation prediction method provided in an embodiment of the present invention, step S101 obtains and processes a variety of original meteorological data of the target area at different historical moments to obtain a meteorological sample set, which may specifically include: obtaining multi-source meteorological data including satellite data, ocean temperature data, and ERA5 meteorological element data at different historical moments in the target area to obtain original meteorological data; deleting noise data in the original meteorological data, and supplementing missing data to obtain preprocessed data; arranging and grouping the preprocessed data in chronological order, each group having data of a set number of consecutive moments, to obtain a total of multiple groups of sample data; the data of the set number of consecutive moments includes input feature data and prediction label data; and dividing the multiple groups of sample data into training sets and test sets according to a set ratio to obtain a meteorological sample set.
[0066] It should be noted that the original meteorological data is used to train the meteorological prediction model. The noise data in the original meteorological data is first deleted, and the missing data is supplemented to obtain the preprocessed data. Then the preprocessed data is arranged and grouped in chronological order. Each group has data from s consecutive moments, of which the first i data are input feature data and the last o data are prediction label data, resulting in a total of n groups of sample data; where s is the sum of i and o; then the n groups of sample data are divided into training sets and test sets according to a certain ratio to obtain the meteorological sample set required for the training of the spatiotemporal prediction model.
[0067] During implementation, it is assumed that the target area is 100 to 150° longitude and -10° to 40° latitude, and the data history length is 5 years from 2019 to 2023. Finally, the historical satellite data, ocean temperature data and ERA5 multi-meteorological element data (wind speed, sea level pressure) of the target area are obtained to obtain the original meteorological data. The noise data in the original meteorological data is deleted, and the missing data is supplemented to obtain the preprocessed data; wherein the noise data is the outlier data (such as 9999, NAN, etc.), and the method of supplementing the missing data includes but is not limited to bilinear interpolation, Newton interpolation and other methods; the preprocessed data is arranged and grouped in chronological order, the starting time of each group is any hour, each group has 16 consecutive moments of data, and each moment is 6 hours apart, wherein the first 4 data are input feature data, and the last 12 data are prediction label data, and the 5-year historical data has a total of 43,824 hours of data, so 43,824 groups of sample data can be obtained; the 43,824 groups of sample data are divided into training set and test set according to 8:2 to obtain the meteorological sample set required for the training of the spatiotemporal prediction model.
[0068] Furthermore, in a specific implementation, in the above-mentioned typhoon formation prediction method provided in an embodiment of the present invention, step S102 inputs the meteorological data sample set into the spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model, which may specifically include: inputting the training set in the meteorological data sample set into the spatiotemporal prediction neural network for training to obtain a trained first model; inputting the test set in the meteorological data sample set into the first model for testing to obtain the test results and judging whether the first model is trained according to the specified threshold parameters; if so, obtaining the multi-source meteorological prediction model; if not, retraining.
[0069] Figure 2 The following is a flowchart of obtaining a multi-source weather forecast model provided by an embodiment of the present invention. Figure 2 As shown, in the implementation, the training set in the meteorological data sample set is input into the spatiotemporal prediction neural network for training to obtain a trained model; wherein the spatiotemporal prediction neural network can use the SwinLSTM deep learning model. The test set in the meteorological data sample set is input into the trained model for testing, and the test results are obtained and the model training is judged according to whether the SSIM similarity is greater than the threshold parameter of 0.9. If the training is completed, the multi-source meteorological prediction model is obtained, otherwise it is retrained; wherein SSIM is the structural similarity index, and the formula is:
[0070] ; (1)
[0071] in, and are the means of the predicted and true values, respectively. and is the standard deviation, is the covariance, and is a constant.
[0072] Furthermore, in a specific implementation, in the above-mentioned typhoon formation prediction method provided in an embodiment of the present invention, step S103 obtains and processes historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with the typhoon position, which may specifically include: obtaining satellite image data of the early typhoon generation in the history of the target area to obtain historical typhoon meteorological data; processing the typhoon meteorological data to obtain a typhoon data sample set marked with the typhoon position.
[0073] In implementation, the target area here is the same as the target area in step S101. It is assumed that the target area is longitude 100 to 150° and latitude -10° to 40°. Since there is less typhoon data each year, the data history length is determined to be 20 years from 2004 to 2023. Finally, the historical satellite image data of the area is obtained to obtain typhoon meteorological data for training the typhoon recognition model.
[0074] Furthermore, in the specific implementation, the typhoon meteorological data is processed in the above steps to obtain a typhoon data sample set marked with the typhoon position, which may specifically include: performing feature engineering on the typhoon meteorological data, marking the typhoon position as a data label on the typhoon meteorological data, and obtaining an initial typhoon sample set; dividing the initial typhoon sample set into a training set and a test set according to a specified ratio, and obtaining a typhoon data sample set.
[0075] During implementation, feature engineering is performed on typhoon meteorological data, and professional meteorological workers mark the locations of typhoons as data labels to obtain an initial typhoon sample set. The initial typhoon sample set can then be divided into a training set and a test set in a ratio of 8:2 to obtain the typhoon data sample set required for target detection model training.
[0076] Furthermore, in a specific implementation, in the above-mentioned typhoon generation prediction method provided in an embodiment of the present invention, step S104 inputs the typhoon data training set into the target detection neural network for training and testing to obtain a typhoon recognition model, which may specifically include: inputting the training set in the typhoon data sample set into the target detection neural network for training to obtain a trained second model; inputting the test set in the typhoon data sample set into the second model for testing, obtaining the test results and judging whether the second model is trained according to the specified threshold parameters; if so, obtaining the typhoon recognition model; if not, retraining.
[0077] Figure 3 The following is a flow chart of obtaining a typhoon identification model provided by an embodiment of the present invention. Figure 3As shown, in the implementation, the training set in the typhoon data sample set is input into the target detection neural network for training to obtain a trained model; wherein the target detection neural network uses the DETR deep learning model; the test set in the typhoon data sample set is input into the trained model for testing, and the test results are obtained and the threshold parameter with an accuracy greater than 0.9 is used to determine whether the model training is completed. If the training is completed, the typhoon recognition model is obtained, otherwise it is retrained; wherein the accuracy represents the ratio of the number of correctly detected targets to the total number of targets, and the formula is:
[0078] ; (2)
[0079] Among them, TP is a true positive example (both the prediction and the actual are true), TN is a true negative example (both the prediction and the actual are false), FP is a false positive example (the prediction is true, the actual is false), and FN is a false negative example (the prediction is false, the actual is true).
[0080] Furthermore, in a specific implementation, in the above-mentioned typhoon generation prediction method provided in an embodiment of the present invention, step S105 inputs the meteorological data to be measured in the target area into the typhoon generation prediction model, and outputs the time and location of future typhoon generation, which may specifically include: inputting the meteorological data to be measured in the target area into the multi-source meteorological prediction model of the typhoon generation prediction model to obtain meteorological forecast results for multiple moments in the future; inputting the satellite image data in the meteorological forecast results into the typhoon identification model of the typhoon generation prediction model in chronological order to obtain the time and location of future typhoon generation.
[0081] Figure 4 The overall framework diagram of the typhoon generation prediction model based on multi-source meteorological data provided by the embodiment of the present invention. Figure 4 As shown, in implementation, various meteorological data of the target area are input into the multi-source meteorological prediction model of the typhoon generation prediction model to obtain meteorological forecast results for multiple moments in the future; the satellite image data in the meteorological forecast results are input into the typhoon identification model in chronological order to obtain the time and location of future typhoon generation. Figure 5 A schematic diagram of prediction results of a typhoon formation prediction model provided in an embodiment of the present invention.
[0082] In the above embodiments, the typhoon generation prediction method is described in detail, and the present invention also provides embodiments corresponding to the typhoon generation prediction device and the typhoon generation prediction equipment. It should be noted that the present invention describes the embodiments of the device part from two perspectives, one is based on the functional module perspective, and the other is based on the hardware perspective.
[0083] Figure 6 Schematic diagram of the structure of the typhoon generation prediction device provided by the embodiment of the present invention. This embodiment is based on the perspective of functional modules, such as Figure 6 As shown, the device comprises:
[0084] The meteorological sample set acquisition module 10 is used to acquire and process a variety of original meteorological data of the target area at different historical moments to obtain a meteorological sample set;
[0085] The multi-source meteorological prediction model training module 11 is used to input the meteorological data sample set into the spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model; the multi-source meteorological prediction model is the first part of the typhoon generation prediction model;
[0086] A typhoon data sample set acquisition module 12 is used to acquire and process the historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with the typhoon position;
[0087] The typhoon identification model training module 13 is used to input the typhoon data training set into the target detection neural network for training and testing to obtain a typhoon identification model; the typhoon identification model is the second part of the typhoon generation prediction model;
[0088] The typhoon generation prediction model inference module 14 is used to input the meteorological data to be measured in the target area into the typhoon generation prediction model, and output the time and location of future typhoon generation.
[0089] In the above-mentioned typhoon generation prediction device provided by the embodiment of the present invention, the typhoon generation prediction problem can be divided into two problems of meteorological prediction and typhoon identification through the interaction of the above-mentioned five modules, and the two problems are solved in sequence. The spatiotemporal prediction neural network and the target detection neural network are trained and tested respectively to obtain a typhoon generation prediction model including a multi-source meteorological prediction model and a typhoon identification model. The meteorological data to be measured in the target area is input into the typhoon generation prediction model to predict the time and location of future typhoon generation with high efficiency and accuracy, thereby simultaneously meeting the needs of efficiently and accurately predicting the time and location of typhoon generation.
[0090] Since the embodiments of the device part correspond to the embodiments of the method part, the embodiments of the device part refer to the description of the embodiments of the method part, which will not be described here. And it has the same beneficial effects as the above-mentioned typhoon generation prediction method.
[0091] Figure 7 A schematic diagram of the structure of a typhoon generation prediction device provided by an embodiment of the present invention. This embodiment is based on the hardware perspective, such as Figure 7 As shown, the typhoon generation prediction equipment includes:
[0092] A memory 20, used for storing computer programs;
[0093] The processor 21 is used to implement the steps of the typhoon generation prediction method mentioned in the above embodiment when executing the computer program.
[0094] Among them, the processor 21 may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 21 can be implemented in at least one hardware form of a digital signal processor (DSP), a field-programmable gate array (FPGA), and a programmable logic array (PLA). The processor 21 may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU; the coprocessor is a low-power processor for processing data in the standby state. In some embodiments, the processor 21 may be integrated with a graphics processing unit (GPU), which is responsible for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 21 may also include an artificial intelligence (AI) processor, which is used to process computing operations related to machine learning.
[0095] The memory 20 may include one or more computer-readable storage media, which may be non-transitory. The memory 20 may also include a high-speed random access memory, and a non-volatile memory, such as one or more disk storage devices, flash memory storage devices. In this embodiment, the memory 20 is at least used to store the following computer program 201, wherein, after the computer program is loaded and executed by the processor 21, it can implement the relevant steps of the typhoon generation prediction method disclosed in any of the aforementioned embodiments. In addition, the resources stored in the memory 20 may also include an operating system 202 and data 203, etc., and the storage method may be temporary storage or permanent storage. Among them, the operating system 202 may include Windows, Unix, Linux, etc. The data 203 may include, but is not limited to, the data involved in the typhoon generation prediction method mentioned above.
[0096] In some embodiments, the typhoon generation prediction device may further include a display screen 22, an input and output interface 23, a communication interface 24, a power supply 25, and a communication bus 26. Those skilled in the art will appreciate that Figure 7The structure shown in the figure does not constitute a limitation on the typhoon generation prediction device, and may include more or fewer components than shown in the figure. The typhoon generation prediction device provided by the embodiment of the present invention includes a memory and a processor. When the processor executes the program stored in the memory, it can implement the typhoon generation prediction method mentioned above, and the effect is the same as above.
[0097] Finally, the present invention also provides an embodiment corresponding to a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps recorded in the above method embodiment are implemented.
[0098] It is understandable that if the method in the above embodiment is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and executes all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc. Various media that can store program codes. The computer-readable storage medium provided by the present invention can implement the above-mentioned typhoon generation prediction method, and the effect is the same as above.
[0099] Finally, the present invention also provides an embodiment corresponding to a computer program product. The computer program product includes a computer program / instruction, and when the computer program / instruction is executed by a processor, the steps described in the above-mentioned typhoon generation prediction method embodiment are implemented. The computer program product provided by the present invention can implement the above-mentioned typhoon generation prediction method, and the effect is the same as above.
[0100] It should also be noted that, in this specification, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the statement "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device including the element.
[0101] The typhoon generation prediction method, device, equipment and medium provided by the present invention are introduced in detail above. The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description. It should be pointed out that for ordinary technicians in this technical field, without departing from the principle of the present invention, the present invention can also be improved and modified in a number of ways, and these improvements and modifications also fall within the scope of protection of the present invention.
Claims
1. A typhoon generation prediction method, characterized in that: The method comprises: Obtain and process a variety of original meteorological data at different historical moments in the target area to obtain a meteorological sample set; Inputting the meteorological data sample set into a spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model; the multi-source meteorological prediction model is the first part of the typhoon generation prediction model; Acquire and process historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with typhoon locations; The typhoon data training set is input into the target detection neural network for training and testing to obtain a typhoon recognition model; the typhoon recognition model is the second part of the typhoon generation prediction model; The meteorological data to be measured in the target area is input into the typhoon generation prediction model, and the time and location of future typhoon generation are output.
2. The typhoon generation prediction method according to claim 1, characterized in that: Obtain and process a variety of original meteorological data at different historical moments in the target area to obtain a meteorological sample set, including: Acquire multi-source meteorological data including satellite data, ocean temperature data, and ERA5 multiple meteorological element data at different historical moments of the target area to obtain the original meteorological data; Deleting noise data in the original meteorological data and supplementing missing data to obtain preprocessed data; Arrange and group the preprocessed data in chronological order, each group having data of a set number of consecutive moments, and obtain a total of multiple groups of sample data; the data of the set number of consecutive moments includes input feature data and predicted label data; The multiple groups of sample data are divided into a training set and a test set according to a set ratio to obtain the meteorological sample set.
3. The typhoon generation prediction method according to claim 2, characterized in that: The meteorological data sample set is input into the spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model, including: Inputting the training set in the meteorological data sample set into the spatiotemporal prediction neural network for training to obtain a trained first model; The test set in the meteorological data sample set is input into the first model for testing, the test result is obtained and it is judged whether the training of the first model is completed according to the specified threshold parameters; if so, the multi-source meteorological prediction model is obtained; if not, retraining is performed.
4. The typhoon generation prediction method according to claim 1, characterized in that: The historical typhoon meteorological data of the target area is obtained and processed to obtain a typhoon data sample set marked with typhoon positions, including: Acquire satellite image data of the target area during the early stages of typhoon formation to obtain historical typhoon meteorological data; The typhoon meteorological data are processed to obtain a typhoon data sample set marked with typhoon positions.
5. The typhoon generation prediction method according to claim 4, characterized in that: The typhoon meteorological data is processed to obtain a typhoon data sample set marked with the typhoon position, including: Performing feature engineering on the typhoon meteorological data, marking the location of the typhoon on the typhoon meteorological data as a data label, and obtaining an initial typhoon sample set; The initial typhoon sample set is divided into a training set and a test set according to a specified ratio to obtain the typhoon data sample set.
6. The typhoon generation prediction method according to claim 5, characterized in that: The typhoon data training set is input into the target detection neural network for training and testing to obtain a typhoon recognition model, including: Inputting the training set in the typhoon data sample set into the target detection neural network for training to obtain a trained second model; The test set in the typhoon data sample set is input into the second model for testing, the test result is obtained and it is judged whether the training of the second model is completed according to the specified threshold parameters; if so, the typhoon recognition model is obtained; if not, retraining is performed.
7. The typhoon generation prediction method according to claim 1, characterized in that: Inputting the meteorological data to be measured in the target area into the typhoon generation prediction model, and outputting the time and location of future typhoon generation, including: Inputting the meteorological data to be measured in the target area into the multi-source meteorological prediction model of the typhoon generation prediction model to obtain meteorological prediction results at multiple moments in the future; The satellite image data in the meteorological forecast results are sequentially input into the typhoon identification model of the typhoon generation prediction model in chronological order to obtain the time and location of future typhoon generation.
8. A typhoon generation prediction device, characterized in that: The device comprises: A meteorological sample set acquisition module is used to acquire and process a variety of original meteorological data at different historical moments in the target area to obtain a meteorological sample set; A multi-source meteorological prediction model training module is used to input the meteorological data sample set into the spatiotemporal prediction neural network for training and testing to obtain a multi-source meteorological prediction model; the multi-source meteorological prediction model is the first part of the typhoon generation prediction model; A typhoon data sample set acquisition module is used to acquire and process the historical typhoon meteorological data of the target area to obtain a typhoon data sample set marked with the typhoon position; A typhoon identification model training module is used to input the typhoon data training set into the target detection neural network for training and testing to obtain a typhoon identification model; the typhoon identification model is the second part of the typhoon generation prediction model; The typhoon generation prediction model inference module is used to input the meteorological data to be measured in the target area into the typhoon generation prediction model and output the time and location of future typhoon generation.
9. A typhoon generation prediction device, characterized in that: The device comprises: Memory for storing computer programs; A processor, configured to implement the steps of the typhoon generation prediction method according to any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the typhoon generation prediction method according to any one of claims 1 to 7 are implemented.
