Typhoon dynamic data prediction method and device, electronic equipment and storage medium

Abstract

This application provides a method, apparatus, electronic device, and storage medium for predicting typhoon dynamic data, belonging to the field of typhoon prediction technology. The method includes: acquiring typhoon meteorological time-series data, wherein the typhoon meteorological time-series data includes multiple variable dimensions and multiple time nodes; based on the time nodes, performing time-series feature encoding on the typhoon meteorological time-series data to obtain typhoon meteorological time features; based on the variable dimensions, performing variable feature encoding on the typhoon meteorological time features to obtain typhoon meteorological fusion features; based on the typhoon meteorological fusion features, performing typhoon path prediction to obtain typhoon path prediction data, and based on the typhoon meteorological fusion features, performing typhoon intensity prediction to obtain typhoon intensity prediction data; and merging the typhoon path prediction data and the typhoon intensity prediction data to obtain typhoon dynamic data. This application embodiment can improve the prediction accuracy of typhoon dynamic data.

Description

Technical Field

[0001] This application relates to the field of typhoon forecasting technology, and in particular to a method and apparatus for forecasting dynamic typhoon data, electronic equipment and storage medium. Background Technology

[0002] Typhoon dynamic data forecasting is used to obtain dynamic information about future typhoons in advance, such as the location of the typhoon center and the intensity of the typhoon. By forecasting typhoon dynamic data, it can provide key information for disaster prevention and mitigation decisions.

[0003] Currently, the most common method for predicting typhoon dynamic data is deep learning, which analyzes information about typhoons over a period of time to predict future typhoon dynamic data. However, due to the limited data dimensions processed, the accuracy of the predicted typhoon data is relatively low. Therefore, how to improve the accuracy of typhoon dynamic data prediction has become an urgent technical problem to be solved. Summary of the Invention

[0004] The main objective of this application is to provide a method, apparatus, electronic device, and storage medium for predicting typhoon dynamic data, aiming to improve the accuracy of typhoon dynamic data prediction.

[0005] To achieve the above objectives, a first aspect of this application proposes a method for predicting typhoon dynamic data, the method comprising:

[0006] Acquire typhoon meteorological time-series data, wherein the typhoon meteorological time-series data includes multiple variable dimensions and multiple time nodes;

[0007] Based on the aforementioned time points, the typhoon meteorological time series data are encoded with time series features to obtain typhoon meteorological time features;

[0008] Based on the aforementioned variable dimensions, the typhoon meteorological time features are encoded to obtain typhoon meteorological fusion features;

[0009] Based on the aforementioned typhoon meteorological fusion characteristics, typhoon path prediction is performed to obtain typhoon path prediction data, and typhoon intensity prediction is performed based on the aforementioned typhoon meteorological fusion characteristics to obtain typhoon intensity prediction data.

[0010] The typhoon path prediction data and the typhoon intensity prediction data are merged to obtain typhoon dynamic data.

[0011] In some embodiments, the step of encoding the typhoon meteorological time-series data based on the time node to obtain typhoon meteorological time features includes:

[0012] Based on the aforementioned time points, determine the data segmentation time step;

[0013] Based on the data segmentation time step, the typhoon meteorological time series data is spatiotemporally embedded to obtain the typhoon meteorological spatiotemporal embedding vector.

[0014] Based on the typhoon meteorological spatiotemporal embedding vector, vector attention weighting processing is performed to obtain a weighted spatiotemporal embedding vector.

[0015] Based on the typhoon meteorological spatiotemporal embedding vector, the weighted spatiotemporal embedding vector is optimized to obtain the typhoon meteorological time features.

[0016] In some embodiments, the step of performing spatiotemporal embedding processing on the typhoon meteorological time-series data based on the data segmentation time step to obtain a typhoon meteorological spatiotemporal embedding vector includes:

[0017] Based on the data segmentation time step, the typhoon meteorological time series data is divided into time series to obtain typhoon meteorological time series fragment data.

[0018] The typhoon meteorological time-series data is embedded temporally to obtain a typhoon meteorological time embedding vector;

[0019] The typhoon meteorological time embedding vector is augmented with positional encoding to obtain the typhoon meteorological spatiotemporal embedding vector.

[0020] In some embodiments, the step of performing vector optimization on the weighted spatiotemporal embedding vector based on the typhoon meteorological spatiotemporal embedding vector to obtain the typhoon meteorological time features includes:

[0021] By performing a residual connection on the typhoon meteorological spatiotemporal embedding vector and the weighted spatiotemporal embedding vector, a typhoon meteorological spatiotemporal optimized vector is obtained.

[0022] The typhoon meteorological optimization vector is regularized to obtain the typhoon meteorological time characteristics.

[0023] In some embodiments, the step of encoding the typhoon meteorological time features based on the variable dimensions to obtain typhoon meteorological fusion features includes:

[0024] Based on the variable dimensions, feature mapping is performed on the typhoon meteorological time characteristics to obtain the original dimension typhoon meteorological matrix;

[0025] The original typhoon meteorological matrix is ​​subjected to variable embedding processing to obtain the typhoon meteorological variable embedding vector;

[0026] Based on the typhoon meteorological variable embedding vector, feature extraction is performed to obtain the typhoon meteorological fusion feature.

[0027] In some embodiments, the step of predicting typhoon paths based on the typhoon meteorological fusion features to obtain typhoon path prediction data includes:

[0028] Based on the aforementioned typhoon meteorological fusion characteristics, typhoon latitude and longitude prediction is performed to obtain typhoon latitude and longitude prediction data;

[0029] Based on the predicted latitude and longitude data of the typhoon, trajectory mapping is performed to obtain the predicted typhoon path data.

[0030] In some embodiments, the step of merging the typhoon path prediction data and the typhoon intensity prediction data to obtain typhoon dynamic data includes:

[0031] The typhoon path prediction data and the typhoon intensity prediction data are stitched together to obtain typhoon prediction stitched data.

[0032] The typhoon forecast splicing data is sorted by time to obtain the typhoon dynamic data.

[0033] To achieve the above objectives, a second aspect of this application provides a typhoon dynamic data prediction device, the device comprising:

[0034] The meteorological data acquisition module is used to acquire typhoon meteorological time-series data, wherein the typhoon meteorological time-series data includes multiple variable dimensions and multiple time nodes;

[0035] The time-series feature encoding module is used to perform time-series feature encoding on the typhoon meteorological time-series data based on the time nodes to obtain typhoon meteorological time features;

[0036] The variable feature encoding module is used to encode the typhoon meteorological time features based on the variable dimensions to obtain typhoon meteorological fusion features;

[0037] The typhoon information prediction module is used to predict the typhoon path based on the typhoon meteorological fusion features to obtain typhoon path prediction data, and to predict the typhoon intensity based on the typhoon meteorological fusion features to obtain typhoon intensity prediction data.

[0038] The typhoon information merging module is used to merge the typhoon path prediction data and the typhoon intensity prediction data to obtain dynamic typhoon data.

[0039] To achieve the above objectives, a third aspect of the present application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method of the first aspect described above.

[0040] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method of the first aspect described above.

[0041] The typhoon dynamic data prediction method, apparatus, electronic device, and storage medium proposed in this application acquire typhoon meteorological time-series data and encode the time-series features of the typhoon meteorological time-series data based on the time nodes in the data to obtain typhoon meteorological time features. This accurately captures the temporal evolution of typhoon meteorological data, thereby improving the prediction accuracy of typhoon dynamic information. Furthermore, based on the variable dimensions in the typhoon meteorological time-series data, variable feature encoding of the typhoon meteorological time features is performed to obtain typhoon meteorological fusion features. These fusion features can represent both the correlation between meteorological features and time, as well as the coupling relationship between various variables in the meteorological data. Further, based on the aforementioned typhoon meteorological fusion features, typhoon path prediction and typhoon intensity prediction are performed separately to obtain typhoon path prediction data and typhoon intensity prediction data. Finally, the typhoon path prediction data and typhoon intensity prediction data are merged to obtain typhoon dynamic data, significantly improving the prediction accuracy of typhoon dynamic information. In addition, it greatly shortens the calculation time and improves the prediction timeliness of typhoon dynamic information. Attached Figure Description

[0042] Figure 1 This is a flowchart of the typhoon dynamic data prediction method provided in the embodiments of this application;

[0043] Figure 2 yes Figure 1 The flowchart of step S102 in the document;

[0044] Figure 3 yes Figure 2 The flowchart of step S202 in the document;

[0045] Figure 4 yes Figure 2 The flowchart of step S204 in the process;

[0046] Figure 5 yes Figure 1 The flowchart of step S103 in the process;

[0047] Figure 6 yes Figure 1 The flowchart of step S104 in the process;

[0048] Figure 7 yes Figure 1 The flowchart of step S105 in the process;

[0049] Figure 8This is a schematic diagram of the typhoon dynamic data prediction device provided in the embodiments of this application;

[0050] Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0052] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0054] First, let's analyze some of the terms used in this application:

[0055] Typhoon Dynamic Data Prediction System: The Typhoon Dynamic Data Prediction System is a set of intelligent prediction frameworks that organize multi-source observations such as satellites, radar, and buoys into continuous time-series data. The system first captures the movement rhythm of typhoons through time coding, then uses variable coding to characterize the pressure-wind speed coupling, and then outputs the latitude, longitude and intensity changes for the next tens of hours at once. Finally, the system stitches together the path and intensity results into a complete spatiotemporal sequence that can be directly used for disaster prevention command.

[0056] Variable dimension: Variable dimension refers to treating different meteorological elements such as longitude, latitude, air pressure, and wind speed as independent channels in typhoon data. For example, in a typhoon record, "121.3°E, 25.1°N, 970hPa, 40m / s" are placed on different dimensions at the same time, so that the model can process and capture the coupling relationship and difference weights between the elements along this axis.

[0057] Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0058] Typhoon dynamic data forecasting is used to obtain dynamic information about future typhoons in advance, such as the location of the typhoon center and the intensity of the typhoon. By forecasting typhoon dynamic data, it can provide key information for disaster prevention and mitigation decisions.

[0059] Currently, common methods for predicting typhoon dynamic data include numerical weather prediction (NMR) and deep learning. NMR predicts weather by solving atmospheric dynamics and thermodynamics equations, but due to the large amount of data and complex equations required, the computation time for numerical typhoon data prediction is long, resulting in poor timeliness of the predicted typhoon data. Deep learning methods predict future typhoon dynamic data by analyzing information from past typhoon periods, but the accuracy of the predicted typhoon data is low due to the limited data dimensions processed. Therefore, improving the accuracy of typhoon dynamic data prediction has become an urgent technical problem to be solved.

[0060] Based on this, embodiments of this application provide a method and apparatus for predicting typhoon dynamic data, an electronic device and a storage medium, aiming to improve the accuracy of typhoon dynamic data prediction.

[0061] The typhoon dynamic data prediction method, apparatus, electronic device, and storage medium provided in this application are specifically described through the following embodiments. First, the typhoon dynamic data prediction method in this application is described.

[0062] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0063] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0064] The typhoon dynamic data prediction method provided in this application relates to the field of typhoon prediction technology. The typhoon dynamic data prediction method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the typhoon dynamic data prediction method, but is not limited to the above forms.

[0065] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0066] Figure 1 This is an optional flowchart of the typhoon dynamic data prediction method provided in the embodiments of this application. The typhoon dynamic data prediction method can be applied to a typhoon dynamic data prediction system. Figure 1 The method may include, but is not limited to, steps S101 to S105.

[0067] Step S101: Obtain typhoon meteorological time series data, which includes multiple variable dimensions and multiple time nodes;

[0068] Step S102: Based on time nodes, perform time series feature encoding on typhoon meteorological time series data to obtain typhoon meteorological time features;

[0069] Step S103: Based on the variable dimension, the typhoon meteorological time characteristics are encoded to obtain the typhoon meteorological fusion characteristics;

[0070] Step S104: Based on the typhoon meteorological fusion characteristics, perform typhoon path prediction to obtain typhoon path prediction data, and based on the typhoon meteorological fusion characteristics, perform typhoon intensity prediction to obtain typhoon intensity prediction data.

[0071] Step S105: Merge the typhoon path prediction data and typhoon intensity prediction data to obtain typhoon dynamic data.

[0072] Steps S101 to S105 of this embodiment involve acquiring typhoon meteorological time-series data and encoding the typhoon meteorological time-series data based on time nodes to obtain typhoon meteorological time features. This accurately captures the temporal evolution of typhoon meteorological data, thereby improving the accuracy of typhoon dynamic information prediction. Furthermore, based on the variable dimensions in the typhoon meteorological time-series data, variable feature encoding is performed on the typhoon meteorological time features to obtain typhoon meteorological fusion features. These fusion features can represent both the correlation between meteorological features and time, as well as the coupling relationship between various variables in the meteorological data. Further, based on the aforementioned typhoon meteorological fusion features, typhoon path prediction and typhoon intensity prediction are performed separately to obtain typhoon path prediction data and typhoon intensity prediction data. These data are then merged to obtain typhoon dynamic data, significantly improving the accuracy of typhoon dynamic information prediction. In addition, the computation time is greatly shortened, improving the timeliness of typhoon dynamic information prediction.

[0073] In step S101 of some embodiments, typhoon meteorological time-series data refers to a typhoon observation sequence arranged chronologically and containing multiple meteorological variables, such as the typhoon center longitude, latitude, minimum air pressure, and maximum wind speed recorded every 6 hours over the past 72 hours. Variable dimension refers to the channel dimension corresponding to each different meteorological variable in the typhoon meteorological time-series data, such as longitude, latitude, air pressure, and wind speed. Time node refers to the data sampling time of each meteorological variable in the typhoon meteorological time-series data, such as 00:00, 06:00, 12:00, etc.

[0074] This application embodiment can collect data on predetermined typhoon meteorological data targets based on a pre-set typhoon meteorological data collection time interval, including multiple variable dimensions such as latitude and longitude, air pressure, and wind speed. The collected scattered typhoon meteorological data is then transformed into a unified, continuous, and traceable sequence to obtain typhoon meteorological time-series data. It should be noted that if the collected typhoon meteorological data contains interfering or inaccurate data, noise reduction or data cleaning processing is required to ensure the accuracy and reliability of the typhoon meteorological time-series data.

[0075] In step S102 of some embodiments, typhoon meteorological time characteristics refer to a high-dimensional representation that carries the pattern of typhoon changes over time.

[0076] In this embodiment, the time step can be set according to the time node, and then the time feature of the typhoon meteorological time series data can be extracted according to the time step to obtain the typhoon meteorological time feature that can be used to predict typhoon dynamic data.

[0077] For details, please refer to Figure 2 In some embodiments, step S102 may include, but is not limited to, steps S201 to S204:

[0078] Step S201: Determine the data segmentation time step based on the time node;

[0079] Step S202: Based on the data segmentation time step, perform spatiotemporal embedding processing on the typhoon meteorological time series data to obtain the typhoon meteorological spatiotemporal embedding vector.

[0080] Step S203: Based on the typhoon meteorological spatiotemporal embedding vector, perform vector attention weighting processing to obtain a weighted spatiotemporal embedding vector;

[0081] Step S204: Based on the typhoon meteorological spatiotemporal embedding vector, perform vector optimization on the weighted spatiotemporal embedding vector to obtain the typhoon meteorological time features.

[0082] In step S201 of some embodiments, the data segmentation time step refers to the time span between two adjacent time nodes. For example, in the typhoon path forecast scenario, a data segmentation time step of 6 hours can be taken.

[0083] In this embodiment of the application, it is important to understand that the selection of the data segmentation time step is crucial in typhoon forecasting, as it directly affects the model's ability to capture the dynamic changes of the typhoon. For example, if the selected data segmentation time step is too long, it may miss rapid changes in the typhoon's path or intensity; if the data segmentation time step is too short, it will increase the computational burden. Typically, the selection of the data segmentation time step is based on the acquisition frequency of typhoon meteorological time-series data. For example, for typhoon meteorological time-series data acquired more frequently, the time step can be set to 1 hour or less, while for typhoon meteorological time-series data acquired more frequently, the time step can be set to 3 hours or 6 hours.

[0084] In step S202 of some embodiments, the typhoon meteorological spatiotemporal embedding vector refers to the high-dimensional representation vector obtained after the typhoon meteorological time series data has undergone spatiotemporal embedding processing.

[0085] This application embodiment can utilize a pre-built typhoon dynamic data prediction model to perform spatiotemporal embedding processing on typhoon meteorological time series data. It should be noted that the typhoon dynamic data prediction model can be a TVFormer (temporal-variate Transformer) model. It should also be noted that the typhoon dynamic data prediction model includes a time feature extraction sub-model, a variable feature extraction sub-model, and a feature decoding sub-model. Among them, the time feature extraction sub-model is used to perform spatiotemporal embedding processing on typhoon meteorological time series data.

[0086] In this embodiment of the application, the time feature extraction sub-model realizes spatiotemporal embedding processing by constructing an embedding function. Through this embedding function, the multi-dimensional typhoon meteorological time series data of each data segmentation time step can be converted into a high-dimensional vector.

[0087] For details, please refer to Figure 3 In some embodiments, step S202 may include, but is not limited to, steps S301 to S303:

[0088] Step S301: Based on the data segmentation time step, the typhoon meteorological time series data is divided into time series segments to obtain typhoon meteorological time series fragment data.

[0089] Step S302: Perform time embedding on the typhoon meteorological time series data to obtain the typhoon meteorological time embedding vector;

[0090] Step S303: Add positional encoding to the typhoon meteorological time embedding vector to obtain the typhoon meteorological spatiotemporal embedding vector.

[0091] In step S301 of some embodiments, typhoon meteorological time series data refers to the data obtained by dividing typhoon meteorological time series data into multiple segments according to the data segmentation time step. It should be noted that each typhoon meteorological time series data contains relevant meteorological information of the typhoon within a certain time range, such as longitude, latitude, air pressure, wind speed, etc. For example, dividing 24 hours of typhoon meteorological data into segments of 6 hours results in 4 typhoon meteorological time series data. Each typhoon meteorological time series data records the longitude, latitude, air pressure, wind speed, and other information of the typhoon within that 6-hour period.

[0092] This application embodiment can start from the beginning of typhoon meteorological time series data and sequentially extract data segments according to the data segmentation time step to obtain typhoon meteorological time series segment data.

[0093] It is important to understand that during the segmentation of typhoon meteorological time-series data, it is crucial to ensure the integrity and consistency of each segment to avoid data loss or anomalies that could degrade the data quality. Simultaneously, attention must be paid to the continuity between the typhoon meteorological time-series data segments to ensure accurate reconstruction of the typhoon's evolution.

[0094] In step S302 of some embodiments, the typhoon meteorological time embedding vector refers to a vector that can characterize the changing patterns and characteristics of the typhoon in the time dimension. It should be noted that the typhoon meteorological time embedding vector contains information about the typhoon meteorological time series data in the time series, such as the changing trend of the typhoon path and the evolution of its intensity.

[0095] In this application embodiment, a recurrent neural network or a long short-term memory network in the time feature extraction sub-model can be used to embed typhoon meteorological time series data into time, thereby mapping the typhoon meteorological time series data from the original observation space to a high-dimensional embedding space, and transforming it into a fixed-dimensional vector representation, namely the typhoon meteorological time embedding vector.

[0096] In step S303 of some embodiments, the time feature extraction sub-model typically includes a position encoding module for adding position information to the typhoon meteorological time embedding vector. Specifically, in the embodiments of this application, position encoding can be added using absolute position encoding or relative position encoding. Absolute position encoding directly encodes the position index of each data segmentation time step into a fixed vector, while relative position encoding considers the relative distance between data segmentation time steps and emphasizes the relative positional relationship between elements in the typhoon meteorological time series data.

[0097] Steps S301 to S303 as illustrated in this embodiment of the application, by reasonably determining the data segmentation time step and dividing the typhoon meteorological time series data into time series, can accurately transform continuous typhoon meteorological time series data into multiple typhoon meteorological time series segments, improving the timeliness and precision of typhoon dynamic data prediction. Secondly, by embedding the typhoon meteorological time series segments into time, the temporal dependencies in the typhoon meteorological time series segments can be fully explored, resulting in a typhoon meteorological time embedding vector that can characterize the dynamic changes of the typhoon, thereby enhancing the ability of the time feature extraction sub-model to understand the complex behavior of the typhoon. Finally, by adding position encoding to the typhoon meteorological time embedding vector through the time feature extraction sub-model, the time embedding vector is transformed into a typhoon meteorological spatiotemporal embedding vector that integrates time and spatial features, enabling the typhoon dynamic data prediction model to better understand and utilize spatiotemporal features to predict typhoon dynamic data.

[0098] In step S203 of some embodiments, the weighted spatiotemporal embedding vector refers to a high-dimensional representation vector after assigning different weights to each dimension of the spatiotemporal embedding vector and adding them together.

[0099] In this embodiment, time attention calculation is performed on the typhoon meteorological spatiotemporal embedding vector to obtain vector attention weights, and then vector weighting is performed on the typhoon meteorological spatiotemporal embedding vector according to the vector attention weights to obtain a weighted spatiotemporal embedding vector.

[0100] In step S204 of some embodiments, the weighted spatiotemporal embedding vector can be optimized by performing residual connection and regularization on the weighted spatiotemporal embedding vector, thereby improving the accuracy and reliability of typhoon meteorological time characteristics.

[0101] For details, please refer to Figure 4 In some embodiments, step S204 may include, but is not limited to, steps S401 to S402:

[0102] Step S401: Perform residual connection on the typhoon meteorological spatiotemporal embedding vector and the weighted spatiotemporal embedding vector to obtain the typhoon meteorological spatiotemporal optimized vector;

[0103] Step S402: Regularize the typhoon meteorological optimization vector to obtain the typhoon meteorological time characteristics.

[0104] In step S401 of some embodiments, the typhoon meteorological spatiotemporal optimization vector refers to a vector that retains information about the typhoon meteorological spatiotemporal embedding vector.

[0105] This embodiment of the application achieves residual connection between the typhoon meteorological spatiotemporal embedding vector and the weighted spatiotemporal embedding vector by adding them element-wise. For example, in the time feature extraction sub-model, the dimension of the typhoon meteorological spatiotemporal embedding vector obtained after time embedding and location encoding is [batch_size, sequence_length, embed_dim]. The dimension of the weighted spatiotemporal embedding vector after attention weighting is also [batch_size, sequence_length, embed_dim]. Furthermore, by adding these two typhoon meteorological spatiotemporal embedding vectors with equal dimensions and the weighted spatiotemporal embedding vector element-wise, the typhoon meteorological spatiotemporal optimization vector can be obtained, which also has the dimension [batch_size, sequence_length, embed_dim].

[0106] In step S402 of some embodiments, a regularization term can be added during the training of the time feature extraction sub-model to achieve regularization of the typhoon meteorological optimization vector, thereby limiting the complexity of the model parameters, preventing overfitting of the time feature extraction sub-model, and improving the accuracy of typhoon meteorological time features.

[0107] Steps S401 to S402, as illustrated in the embodiments of this application, integrate the typhoon meteorological spatiotemporal embedding vector and the weighted spatiotemporal embedding vector through residual connection. This can retain and enhance the key spatiotemporal features in the typhoon meteorological spatiotemporal embedding vector, generate a more expressive spatiotemporal optimized vector, and improve the ability of the time feature extraction sub-model to capture typhoon dynamic events. Furthermore, regularization processing of the typhoon meteorological optimized vector can further refine the features, remove redundant and noise information, enhance the generalization ability of the time feature extraction sub-model, and ensure stable output of high-precision time features under different typhoon scenarios, thereby improving the prediction accuracy of typhoon dynamic data.

[0108] Steps S201 to S204, as illustrated in this embodiment, determine the data segmentation time step based on the time node. This accurately divides continuous typhoon meteorological time series data into multiple segments, thereby improving the timeliness of typhoon forecasting. Based on the data segmentation time step, spatiotemporal embedding processing is performed on the typhoon meteorological time series data. This fully mines the temporal and spatial features in the typhoon meteorological time series data, obtaining a typhoon meteorological spatiotemporal embedding vector that can characterize the dynamic changes of the typhoon. Furthermore, vector attention weighting processing is applied to the typhoon meteorological spatiotemporal embedding vector, allowing the model to focus on the key features of the typhoon meteorological spatiotemporal embedding vector and suppress irrelevant features, thereby improving the accuracy of typhoon dynamic data forecasting. Finally, by refining the weighted spatiotemporal embedding vector through vector optimization, redundant information can be removed, improving the representational ability and generalization performance of the weighted spatiotemporal embedding vector. This results in more efficient and representative typhoon meteorological time features, providing strong support for accurate forecasting of typhoon dynamic data and enhancing the reliability and practicality of typhoon dynamic data forecasting.

[0109] In step S103 of some embodiments, the typhoon meteorological fusion feature refers to the unified representation after fusing the typhoon meteorological time features with the variable feature encoding results.

[0110] In this embodiment, the typhoon meteorological time features can be mapped according to the variable dimension to generate a matrix consistent with the original variable dimension. Then, the matrix is ​​subjected to variable embedding processing to obtain the typhoon meteorological variable embedding vector. Finally, the typhoon meteorological fusion features are obtained by feature extraction based on the typhoon meteorological variable embedding vector.

[0111] For details, please refer to Figure 5 In some embodiments, step S103 may include, but is not limited to, steps S501 to S503:

[0112] Step S501: Based on the variable dimension, perform feature mapping on the typhoon meteorological time characteristics to obtain the original dimension typhoon meteorological matrix;

[0113] Step S502: Perform variable embedding processing on the original typhoon meteorological matrix to obtain the typhoon meteorological variable embedding vector;

[0114] Step S503: Based on the typhoon meteorological variable embedding vector, feature extraction is performed to obtain typhoon meteorological fusion features.

[0115] In step S501 of some embodiments, the original dimension typhoon meteorological matrix refers to the matrix formed after mapping the typhoon meteorological time characteristics back to the variable dimensions of typhoon meteorological time series data. The original dimension typhoon meteorological matrix contains meteorological data such as longitude, latitude, air pressure, and wind speed of the typhoon at different time points.

[0116] In this embodiment, a pre-trained variable feature extraction sub-model can be used to perform feature mapping on typhoon meteorological time features. Specifically, the variable feature extraction sub-model can implement feature mapping by constructing a mapping function, which can be linear or non-linear, depending on the characteristics of the typhoon meteorological time features. For example, assuming that the typhoon meteorological time features are a 64-dimensional vector, and the variable dimensions are 4 (longitude, latitude, air pressure, wind speed), a mapping function can be used to map the 64-dimensional feature vector back to the original 4-dimensional typhoon meteorological matrix.

[0117] In step S502 of some embodiments, the typhoon meteorological variable embedding vector refers to the vector representation that can characterize the meteorological features of a typhoon after mapping each variable in the original dimensional typhoon meteorological matrix to a high-dimensional space.

[0118] In this embodiment of the application, the method for variable embedding processing of the original dimensional typhoon meteorological matrix is ​​similar to the method in step S202 above. Each variable feature in the original dimensional typhoon meteorological matrix can be embedded by the embedding function constructed during the model training process. Specifically, the embedding function can map the value of each variable feature to a fixed-dimensional vector space, thereby forming a typhoon meteorological variable embedding vector.

[0119] In step S503 of some embodiments, the typhoon meteorological variable embedding vector can be used to extract features from the feedforward neural network pre-built in the variable feature extraction sub-model to obtain the initial typhoon meteorological fusion features. Furthermore, the initial typhoon meteorological fusion features are then subjected to residual connection and regularization processing to obtain the typhoon meteorological fusion features.

[0120] Steps S501 to S503 of this embodiment, as shown in the present application, lay an interpretable foundation for variable embedding by mapping the typhoon meteorological time features back to the original typhoon meteorological matrix according to the variable dimensions. Secondly, variable embedding is performed on the original typhoon meteorological matrix, which maps variables such as longitude, latitude, air pressure, and wind speed in the original typhoon meteorological matrix to a high-dimensional space, thereby clarifying the coupling relationship between variables and generating a typhoon meteorological variable embedding vector with richer semantics. Finally, feature extraction is performed on the typhoon meteorological variable embedding vector to remove redundant information and strengthen key information, thereby outputting typhoon meteorological fusion features. This clarifies the impact of multiple dimensions on typhoon dynamic data, making the typhoon dynamic data prediction model more accurate and robust in predicting typhoon dynamic data.

[0121] In step S104 of some embodiments, the typhoon path prediction data refers to the sequence of the center latitude and longitude of the typhoon at various future times, such as the location points at longitude 121.3°E and latitude 25.1°N. The typhoon intensity prediction data refers to the sequence of air pressure and wind speed of the typhoon at various future times, such as the values ​​of air pressure 970hPa and wind speed 40m / s.

[0122] In this embodiment, the future latitude and longitude of the typhoon are directly regressed using the meteorological fusion characteristics of the typhoon, and then the latitude and longitude are mapped into a continuous trajectory point by point to obtain the typhoon path prediction data. At the same time, the future air pressure, wind speed and other data of the typhoon are directly regressed using the meteorological fusion characteristics of the typhoon, and then the air pressure, wind speed and other data are mapped into continuous intensity information point by point to obtain the typhoon intensity prediction data.

[0123] For details, please refer to Figure 6 In some embodiments, step S104, based on typhoon meteorological fusion features, performs typhoon path prediction to obtain typhoon path prediction data, which may include, but is not limited to, steps S601 to S602:

[0124] Step S601: Based on the typhoon meteorological fusion characteristics, perform typhoon latitude and longitude prediction to obtain typhoon latitude and longitude prediction data;

[0125] Step S602: Based on the typhoon latitude and longitude prediction data, perform trajectory mapping to obtain typhoon path prediction data.

[0126] In step S601 of some embodiments, the typhoon latitude and longitude prediction data refers to the longitude and latitude values ​​of the typhoon center location at a future time, such as coordinate points like 121.3°E and 25.1°N.

[0127] The embodiments of this application can use affine transformation and activation functions to map the typhoon meteorological fusion features to a low-dimensional space to obtain a low-dimensional feature vector. It should be noted that this low-dimensional space corresponds to the latitude and longitude information in the future. Therefore, the feature decoding sub-model of the typhoon dynamic data prediction model can predict the latitude and longitude information of the typhoon based on the low-dimensional feature vector, that is, the typhoon latitude and longitude prediction data.

[0128] In step S602 of some embodiments, the above-mentioned multiple discrete typhoon latitude and longitude prediction data can be regarded as a time-stamped observation sequence. First, spline interpolation is used to refine the time interval between the typhoon latitude and longitude prediction data into smaller intervals to maintain the curvature continuity of the trajectory. Then, piecewise circular arc fitting is performed on the interpolated typhoon latitude and longitude prediction data so that the entire path satisfies the Earth's great circle route constraint and avoids the physical irrationality of straight lines crossing land or mountains, thus obtaining typhoon path prediction data.

[0129] Steps S601 to S602, as shown in the embodiments of this application, involve predicting the latitude and longitude of a typhoon based on its meteorological fusion characteristics. This ensures that each point in the obtained typhoon latitude and longitude prediction data carries complete spatiotemporal information. The typhoon latitude and longitude prediction data is then smoothed into a continuous typhoon path through trajectory mapping, i.e., typhoon path prediction data. This approach enables high-precision, high-resolution, and computationally lightweight typhoon path prediction, thereby improving the accuracy of typhoon dynamic data prediction.

[0130] It should be noted that the method for predicting typhoon intensity based on typhoon meteorological fusion characteristics is similar to the method for predicting typhoon path based on typhoon meteorological fusion characteristics, and therefore will not be elaborated upon here.

[0131] In step S105 of some embodiments, typhoon dynamic data refers to the complete spatiotemporal state sequence after merging path prediction data and intensity prediction data by time step, for example, a 24×4 matrix, where each row contains the longitude, latitude, air pressure, and wind speed of the typhoon at a certain future time.

[0132] In this embodiment of the application, typhoon path prediction data and typhoon intensity prediction data can be spliced ​​along the variable dimension axis into a matrix with the number of prediction time nodes × the number of variable dimensions. Furthermore, the matrix can be sorted according to the time axis to obtain complete typhoon dynamic data.

[0133] For details, please refer to Figure 7 In some embodiments, step S105, based on typhoon meteorological fusion features, performs typhoon path prediction to obtain typhoon path prediction data, which may include, but is not limited to, steps S701 to S702:

[0134] Step S701: The typhoon path prediction data and typhoon intensity prediction data are stitched together to obtain the typhoon prediction stitched data.

[0135] Step S702: Sort the typhoon forecast splicing data by time to obtain typhoon dynamic data.

[0136] In step S701 of some embodiments, the typhoon prediction splicing data refers to a two-dimensional matrix formed by splicing the typhoon path prediction data and the typhoon intensity prediction data along the variable dimension. For example, it is a merged matrix with 4 columns and 24 rows, each row containing [longitude, latitude, air pressure, wind speed].

[0137] This application embodiment can form a two-dimensional matrix data, namely typhoon prediction splicing data, by parallel processing of data columns such as longitude and latitude in typhoon path prediction data and data columns such as air pressure and wind speed in typhoon intensity prediction data.

[0138] In step S702 of some embodiments, the typhoon dynamic data can be obtained by sorting the data in ascending order according to the timestamps of each row or column in the above-mentioned typhoon prediction splicing data.

[0139] Steps S701 to S702, as shown in the embodiments of this application, involve stitching together typhoon path prediction data to form complete typhoon prediction stitched data. This ensures that each row of data in the typhoon prediction stitched data carries both path location and intensity information. Furthermore, by sorting the typhoon prediction stitched data by time, high-resolution, non-abrupt typhoon dynamic data is obtained. This enables the progressive construction from discrete points to continuous spatiotemporal trajectories, thereby reducing information loss in the typhoon dynamic data prediction process and improving the intuitiveness of typhoon dynamic data prediction.

[0140] In one embodiment of this application, taking Typhoon Haikui as an example, firstly, 72 hours of continuous observation data of typhoon Haikui's latitude and longitude, central pressure, and maximum wind speed per hour are collected as typhoon meteorological time series data. Then, the typhoon meteorological time series data is encoded into typhoon meteorological time features according to a pre-set 6-hour step size. Next, according to the four variable dimensions of longitude, latitude, pressure, and wind speed, the typhoon meteorological time features are embedded with variables of the four dimensions respectively, and a high-dimensional typhoon meteorological fusion feature is output. Subsequently, the typhoon meteorological fusion feature is used to regress the latitude and longitude of the next 24 hours at once to obtain typhoon path prediction data, and the minimum pressure and maximum wind speed at the corresponding time are output simultaneously, i.e., typhoon intensity prediction data. Finally, the above two sets of data, typhoon path prediction data and typhoon intensity prediction data, are merged to generate a complete typhoon dynamic data covering the typhoon center location and typhoon intensity evolution. This typhoon dynamic data can then be directly used for typhoon wind force warning decision-making.

[0141] This application acquires typhoon meteorological time-series data and encodes its temporal features based on time nodes within the data to obtain typhoon meteorological time features. This accurately captures the temporal evolution of typhoon meteorological data, thereby improving the accuracy of typhoon dynamic information prediction. Furthermore, based on the variable dimensions within the typhoon meteorological time-series data, variable feature encoding is performed on the typhoon meteorological time features to obtain typhoon meteorological fusion features. These fusion features represent both the correlation between meteorological features and time, as well as the coupling relationships between various variables in the meteorological data. Further, based on these typhoon meteorological fusion features, typhoon path prediction and typhoon intensity prediction are performed separately to obtain typhoon path prediction data and typhoon intensity prediction data. These data are then merged to obtain typhoon dynamic data, significantly improving the accuracy of typhoon dynamic information prediction. In addition, the computation time is greatly shortened, enhancing the timeliness of typhoon dynamic information prediction.

[0142] Please see Figure 8 This application also provides a typhoon dynamic data prediction device, which can implement the above-mentioned typhoon dynamic data prediction method. The device includes:

[0143] The meteorological data acquisition module 801 is used to acquire typhoon meteorological time series data, which includes multiple variable dimensions and multiple time nodes.

[0144] The time-series feature encoding module 802 is used to encode the time-series meteorological data based on time nodes to obtain the time features of typhoon meteorology.

[0145] The variable feature encoding module 803 is used to encode the typhoon meteorological time features based on the variable dimension to obtain the typhoon meteorological fusion features;

[0146] The typhoon information prediction module 804 is used to predict the typhoon path based on the typhoon meteorological fusion characteristics, and to predict the typhoon intensity based on the typhoon meteorological fusion characteristics, and to obtain typhoon intensity prediction data.

[0147] The typhoon information merging module 805 is used to merge typhoon path prediction data and typhoon intensity prediction data to obtain dynamic typhoon data.

[0148] The specific implementation of this typhoon dynamic data prediction device is basically the same as the specific implementation of the typhoon dynamic data prediction method described above, and will not be repeated here.

[0149] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described typhoon dynamic data prediction method. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0150] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0151] The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0152] The memory 902 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 using the typhoon dynamic data prediction method of the embodiments of this application.

[0153] The input / output interface 903 is used to implement information input and output;

[0154] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0155] Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904);

[0156] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0157] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described typhoon dynamic data prediction method.

[0158] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0159] The typhoon dynamic data prediction method, typhoon dynamic data prediction device, electronic device, and storage medium provided in this application embodiment acquire typhoon meteorological time-series data, wherein the typhoon meteorological time-series data includes multiple variable dimensions and multiple time nodes. Based on the time nodes, the typhoon meteorological time-series data is encoded with time-series features to obtain typhoon meteorological time features. Based on the variable dimensions, the typhoon meteorological time features are encoded with variable features to obtain typhoon meteorological fusion features. Based on the typhoon meteorological fusion features, typhoon path prediction is performed to obtain typhoon path prediction data. Based on the typhoon meteorological fusion features, typhoon intensity prediction is performed to obtain typhoon intensity prediction data. The typhoon path prediction data and typhoon intensity prediction data are merged to obtain typhoon dynamic data.

[0160] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0161] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0162] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0163] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0164] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0165] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0166] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. The coupling or direct coupling or communication connection between the shown or discussed units may be through some interfaces, or indirect coupling or communication connection between the apparatus or units, and may be electrical, mechanical, or other forms.

[0167] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0168] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0169] If the integrated unit is implemented as 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 this application, in essence, 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. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0170] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for predicting typhoon dynamic data, characterized in that, The method includes: Acquire typhoon meteorological time-series data, wherein the typhoon meteorological time-series data includes multiple variable dimensions and multiple time nodes; Based on the aforementioned time points, determine the data segmentation time step; Based on the data segmentation time step, the typhoon meteorological time series data is divided into time series to obtain typhoon meteorological time series fragment data. The typhoon meteorological time-series data is embedded temporally to obtain a typhoon meteorological time embedding vector; The typhoon meteorological time embedding vector is augmented with positional encoding to obtain the typhoon meteorological spatiotemporal embedding vector; Based on the typhoon meteorological spatiotemporal embedding vector, vector attention weighting processing is performed to obtain a weighted spatiotemporal embedding vector. Based on the typhoon meteorological spatiotemporal embedding vector, the weighted spatiotemporal embedding vector is optimized to obtain the typhoon meteorological time features; Based on the variable dimensions, feature mapping is performed on the typhoon meteorological time characteristics to obtain the original dimension typhoon meteorological matrix; The original typhoon meteorological matrix is ​​subjected to variable embedding processing to obtain the typhoon meteorological variable embedding vector; Based on the typhoon meteorological variable embedding vector, feature extraction is performed to obtain typhoon meteorological fusion features; Based on the aforementioned typhoon meteorological fusion characteristics, typhoon path prediction is performed to obtain typhoon path prediction data, and typhoon intensity prediction is performed based on the aforementioned typhoon meteorological fusion characteristics to obtain typhoon intensity prediction data. The typhoon path prediction data and the typhoon intensity prediction data are merged to obtain typhoon dynamic data.

2. The method according to claim 1, characterized in that, The step of optimizing the weighted spatiotemporal embedding vector based on the typhoon meteorological spatiotemporal embedding vector to obtain the typhoon meteorological time features includes: By performing a residual connection on the typhoon meteorological spatiotemporal embedding vector and the weighted spatiotemporal embedding vector, a typhoon meteorological spatiotemporal optimized vector is obtained. The typhoon meteorological spatiotemporal optimization vector is regularized to obtain the typhoon meteorological time features.

3. The method according to any one of claims 1-2, characterized in that, The process of predicting typhoon paths based on the aforementioned typhoon meteorological fusion features, resulting in typhoon path prediction data, includes: Based on the aforementioned typhoon meteorological fusion characteristics, typhoon latitude and longitude prediction is performed to obtain typhoon latitude and longitude prediction data; Based on the predicted latitude and longitude data of the typhoon, trajectory mapping is performed to obtain the predicted typhoon path data.

4. The method according to any one of claims 1-2, characterized in that, The process of merging the typhoon path prediction data and the typhoon intensity prediction data to obtain dynamic typhoon data includes: The typhoon path prediction data and the typhoon intensity prediction data are stitched together to obtain typhoon prediction stitched data. The typhoon forecast stitched data is sorted by time to obtain the typhoon dynamic data.

5. A typhoon dynamic data prediction device, characterized in that, The device includes: The meteorological data acquisition module is used to acquire typhoon meteorological time-series data, wherein the typhoon meteorological time-series data includes multiple variable dimensions and multiple time nodes; The temporal feature encoding module is used to determine the data segmentation time step based on the time node, perform temporal segmentation on the typhoon meteorological time series data based on the data segmentation time step, obtain typhoon meteorological time series segment data, perform time embedding on the typhoon meteorological time series segment data to obtain typhoon meteorological time embedding vector, add position encoding to the typhoon meteorological time embedding vector to obtain typhoon meteorological spatiotemporal embedding vector, perform vector attention weighting processing on the typhoon meteorological spatiotemporal embedding vector to obtain weighted spatiotemporal embedding vector, and perform vector optimization on the weighted spatiotemporal embedding vector based on the typhoon meteorological spatiotemporal embedding vector to obtain typhoon meteorological time features; The variable feature encoding module is used to perform feature mapping on the typhoon meteorological time features based on the variable dimension to obtain the original dimension typhoon meteorological matrix, perform variable embedding processing on the original dimension typhoon meteorological matrix to obtain the typhoon meteorological variable embedding vector, and perform feature extraction based on the typhoon meteorological variable embedding vector to obtain the typhoon meteorological fusion feature. The typhoon information prediction module is used to predict the typhoon path based on the typhoon meteorological fusion features to obtain typhoon path prediction data, and to predict the typhoon intensity based on the typhoon meteorological fusion features to obtain typhoon intensity prediction data. The typhoon information merging module is used to merge the typhoon path prediction data and the typhoon intensity prediction data to obtain dynamic typhoon data.

6. An electronic device, characterized in that, The electronic device includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the typhoon dynamic data prediction method according to any one of claims 1 to 4.

7. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the typhoon dynamic data prediction method according to any one of claims 1 to 4.

Images