Weather detection method and system based on meteorological rainfall radar, and storage medium

Abstract

The invention discloses a weather detection method and system based on a weather rainfall radar and a storage medium. The system comprises the weather rainfall radar, a weather data centralized control management center and a radar time sequence chart neural network detection algorithm arranged in the weather data centralized control management center. According to the method, different cloud layer characteristic weathers are collected through a meteorological rain detection radar, and obtained electromagnetic wave scattering parameters of the different cloud layer characteristic weathers are input into a radar time sequence chart neural network detection algorithm for analysis and calculation, so that polarization phase differences corresponding to the different cloud layer characteristic weathers are obtained to detect the height and vertical distribution of cloud layers; and when the height and the vertical distribution of the cloud layer exceed boundary threshold values, sending out risk early warning information of different weather types, and automatically detecting spectral widths of different cloud layer characteristic weather and corresponding polarization phase differences. And the weather detection precision and efficiency are improved.

Description

Technical Field

[0001] The present invention relates to the field of weather detection with different cloud layer characteristics, and in particular to a weather detection method, system and storage medium based on a meteorological rainfall radar. Background Art

[0002] Weather detection methods based on meteorological rainfall radar are an indispensable technical means in modern meteorological observation. By emitting electromagnetic waves and receiving reflected echoes from target objects, the reflectivity, Doppler velocity, polarization characteristics and other parameters of precipitation particles (such as raindrops, hail, and snowflakes) are analyzed to monitor weather conditions in real time. This technology has played an important role in short-term precipitation forecasting, rainstorm warnings, and extreme weather monitoring. However, with the increasing complexity of meteorological systems and the diversification of weather phenomena, the detection capabilities of a single radar system can no longer fully meet the needs. At this stage, weather detection methods rely more on a single radar or a single data source for analysis, and do not fully utilize the potential of multimodal data fusion and historical meteorological data to improve prediction accuracy, which has become a key shortcoming that restricts the further development of existing technologies.

[0003] The current meteorological rainfall radar technology has significant defects in analysis accuracy, mainly due to the insufficient integration of multimodal data. Meteorological phenomena themselves have complex multidimensional characteristics, and there is a strong correlation between different radar parameters (such as reflectivity, Doppler velocity, polarization characteristics) and other observation data (such as weather station records, satellite remote sensing data, wind profile radar data), but existing technologies usually process radar data separately, ignoring the necessity of multimodal data fusion. Due to the lack of effective multi-source information collaborative analysis, the limitations of single radar data in complex weather conditions are magnified. For example, in severe convective weather, the irregular shape and motion state of precipitation particles will cause inconsistencies in polarization parameters and velocity data in radar data, and single data analysis is difficult to accurately describe the overall characteristics of the weather system. In addition, existing analysis methods are more based on real-time data, and historical data are not fully utilized for comparative analysis, resulting in insufficient understanding of weather change trends and affecting the early warning capabilities of extreme weather.

[0004] Another major disadvantage is the low utilization rate of historical meteorological data, and the failure to combine historical weather patterns and trends to conduct a more comprehensive analysis of current weather data. Historical data contains a large number of statistical laws and characteristic patterns about weather evolution. If multimodal real-time data and historical data can be combined for correlation analysis, the accuracy of early warning will be significantly improved. However, most existing technologies lack the ability to mine historical data and cannot use past data patterns to correct errors in real-time observations. For example, the formation of certain local weather phenomena (such as tornadoes and hail) may be related to specific weather patterns, and these patterns can only be identified through the analysis of historical data. Without this related information, radar observations may underestimate the intensity or probability of certain extreme weather. Overall, the existing meteorological rainfall radar technology has failed to effectively integrate multimodal data and historical data, resulting in inaccurate radar data analysis, which limits its application effect under complex weather conditions. Summary of the invention

[0005] In order to overcome the above-mentioned deficiencies in the prior art, the present invention provides a weather detection method, system and storage medium based on a meteorological rainfall radar to solve the above-mentioned problems.

[0006] In order to achieve the above-mentioned purpose, the technical solution provided by the preferred embodiment of the present invention is as follows: In terms of the method, the preferred embodiment of the present invention provides a weather detection method based on a meteorological rain measuring radar, the method comprising a meteorological rain measuring radar, a weather data set control management center for data interaction with the meteorological rain measuring radar, and a radar timing graph neural network detection algorithm arranged in the weather data set control management center; the method comprises: using the meteorological rain measuring radar to collect electromagnetic wave scattering parameters of different cloud layer characteristic weather, which are used as electromagnetic wave scattering parameters of different cloud layer characteristic weather acquired by the different cloud layer characteristic weather detection method; inputting the acquired electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar timing graph neural network detection algorithm for analysis and calculation to obtain the spectral width and corresponding polarization phase difference of different cloud layer characteristic weather; combining the influencing factors of the electromagnetic wave scattering parameters and the judgment results of the parameter boundary conditions to perform polarization phase difference judgment in the big data model, and if the polarization phase difference fluctuation range is higher than the safety limit interval, issuing risk warning information of different weather types.

[0007] In the present invention, before the step of inputting the acquired electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation, the method includes: integrating electromagnetic wave scattering parameter sets of different cloud layer characteristic weather, including electromagnetic wave scattering parameters of different unit times, each of the electromagnetic wave scattering parameters per unit time includes a single electromagnetic wave scattering parameter and the corresponding cloud particle type of different cloud layer characteristic weather, and the cloud particle type corresponds to the echo intensity and polarization parameters of different cloud layer characteristic weather.

[0008] The electromagnetic wave scattering parameter set is used, a convolutional neural network model is adopted to optimize the radar time series graph neural network detection algorithm, and the model is used to analyze the influencing factors of electromagnetic wave scattering parameters and judge the boundary conditions of parameters.

[0009] In the present invention, the step of inputting the acquired electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation includes: matching the electromagnetic wave scattering parameter characteristics of the electromagnetic wave scattering parameters of different cloud layer characteristic weather with the electromagnetic wave scattering parameter characteristics of the electromagnetic wave scattering parameters per unit time in the electromagnetic wave scattering parameter set in the radar time series graph neural network detection algorithm to obtain the compliance rate of the electromagnetic wave scattering parameters of different cloud layer characteristic weather and the electromagnetic wave scattering parameters per unit time; selecting the electromagnetic wave scattering parameters per unit time with the largest compliance rate in the electromagnetic wave scattering parameter set, and using the cloud particle type of the selected electromagnetic wave scattering parameters per unit time as the actual monitoring value of the electromagnetic wave scattering parameters of different cloud layer characteristic weather, so as to obtain the spectral width of different cloud layer characteristic weather and the polarization parameters of different cloud layer characteristic weather.

[0010] In the present invention, the step of inputting the acquired electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation includes: statistically analyzing the echo ratios corresponding to the polarization parameters of different cloud layer characteristic weather collected by the meteorological rain measuring radar within a preset time period.

[0011] In the present invention, the above-mentioned step of judging the polarization parameters and issuing risk warning information of different weather types when the fluctuation range of the polarization phase difference is higher than the safety limit interval includes: the height of the cloud layer, vertical distribution data and structural changes and development data of the cloud layer at different terrains and altitudes, calculating the characteristic weather of different cloud layers in the area according to the regional information and polarization parameters, and issuing risk warning information of different weather types, wherein the risk warning information of different weather types includes rainfall probability, thunderstorm probability, climate temperature and humidity, haze probability and level, secondary comprehensive disaster probability and level, wherein different probabilities and levels correspond to different degrees of hazard of severe convective weather.

[0012] In the present invention, the above-mentioned method for detecting weather characteristics of different cloud layers also includes an edge computing gateway server, and the edge computing gateway server and the weather data control management center use 5G signals to exchange data. The edge computing gateway server is provided with the weather analysis and early warning model of the multimodal fusion weight analysis. The step of using meteorological rain measuring radar to collect electromagnetic wave scattering parameters of weather characteristics of different cloud layers in the monitoring area, which are used as electromagnetic wave scattering parameters of weather characteristics of different cloud layers obtained by the method for detecting weather characteristics of different cloud layers, includes: the meteorological rain measuring radar uploads the collected electromagnetic wave scattering parameters of weather characteristics of different cloud layers in the monitoring area to the edge computing gateway server.

[0013] Furthermore, the weather analysis and warning model of the multimodal fusion weight analysis is expressed as:

[0014] in, represents the weights of factors affecting different electromagnetic wave scattering parameters, Indicates the echo strength, represents the polarization parameter, Represents the differential reflectivity of different cloud characteristics, represents the parameter boundary conditions, Cloud particle characteristic parameter matrix, represents polarization error compensation, LN is layer normalization, Dropout( ) is the preset weight distribution calculation, represents the multi-head self-attention mechanism, , , They are weather embedding matrices of different cloud layer characteristics, corresponding to query, key and value, is the learnable matrix of multi-head self-attention in different unit times.

[0015] In the present invention, the above-mentioned step of inputting the acquired electromagnetic wave scattering parameters of weather with different cloud layer characteristics into the radar time series graph neural network detection algorithm for analysis and calculation includes: the edge computing gateway server inputs the acquired electromagnetic wave scattering parameters of weather with different cloud layer characteristics in the monitoring area into the radar time series graph neural network detection algorithm of the edge computing gateway server to detect and count the electromagnetic wave scattering parameters of weather with different cloud layer characteristics in the monitoring area.

[0016] As for the system, a preferred embodiment of the present invention provides a weather detection system based on a meteorological rain measuring radar, the system comprising a meteorological rain measuring radar, a weather data set control and management center for data interaction with the meteorological rain measuring radar, and different weather type risk warning modules connected to the weather data set control and management center, wherein the meteorological rain measuring radar is used to collect electromagnetic wave scattering parameters of different cloud layer characteristic weather in the monitoring area, which are used as electromagnetic wave scattering parameters of different cloud layer characteristic weather acquired by the risk warning method for detecting different weather types with different cloud layer characteristic weather; the weather data set control and management center comprises a radar time series graph neural network detection algorithm, and the different weather type risk warning modules comprise: an electromagnetic wave scattering parameter interface, which is used to input the acquired electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation, so as to obtain the spectral width of different cloud layer characteristic weather and the corresponding polarization phase difference; a different weather type risk warning judgment interface, which is used to judge the polarization parameter, and when the fluctuation range of the polarization phase difference is higher than the safety limit interval, different weather type risk warning information is issued.

[0017] In the present invention, the above-mentioned risk warning module for different weather types also includes: an electromagnetic wave scattering parameter acquisition interface, which is used to integrate the electromagnetic wave scattering parameter sets of different cloud layer characteristic weather, including electromagnetic wave scattering parameters of different unit times, each of the electromagnetic wave scattering parameters per unit time includes a single electromagnetic wave scattering parameter of a different cloud layer characteristic weather and the corresponding cloud particle type of the different cloud layer characteristic weather, and the cloud particle type corresponds to the echo intensity and polarization parameters of the different cloud layer characteristic weather.

[0018] The weather analysis and warning model establishment and judgment interface is used to use the electromagnetic wave scattering parameter set, adopt the convolutional neural network model to optimize the radar time series graph neural network detection algorithm, and use the model to analyze the influencing factors of electromagnetic wave scattering parameters and judge the parameter boundary conditions.

[0019] In the present invention, the above-mentioned risk warning judgment interface for different weather types is also used for: the height of the cloud layer, vertical distribution data, and structural changes and development data of the cloud layer in the workshop area at different terrains and altitudes, and the cumulative number and incremental number of different cloud characteristic weather in the area are calculated according to regional information and polarization parameters, and different weather type risk warning information is issued, wherein the risk warning information for different weather types includes rainfall probability, thunderstorm probability, climate temperature and humidity, haze probability and level, secondary comprehensive disaster probability and level, wherein different probabilities and levels correspond to different degrees of severe convective weather hazards.

[0020] The present invention also proposes a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the weather detection method based on the meteorological rainfall radar are executed.

[0021] Compared with the prior art, the present invention has at least one of the following beneficial technical effects:

[0022] 1. This method uses the electromagnetic wave scattering parameters of different cloud layer characteristic weather in the monitoring area collected by the meteorological rainfall radar as the electromagnetic wave scattering parameters of different cloud layer characteristic weather acquired by the weather detection method; then the acquired electromagnetic wave scattering parameters of different cloud layer characteristic weather are input into the radar time series graph neural network detection algorithm for analysis and calculation to obtain the spectrum width and corresponding polarization phase difference of different cloud layer characteristic weather; the weather data collection and control management center is used to detect the spectrum width and corresponding polarization phase difference of different cloud layer characteristic weather, which simplifies the operation of analyzing and calculating different cloud layer characteristic weather. This method is an efficient and accurate modern meteorological monitoring method with the significant advantages of multi-dimensional fusion, intelligent analysis, real-time response and wide adaptability.

[0023] 2. This method comprehensively reflects the physical properties and dynamic evolution process of different cloud characteristic weather by collecting electromagnetic wave scattering parameters, including echo intensity, polarization parameters, differential reflectivity and other data. Compared with traditional single parameter analysis, this method introduces multimodal data fusion technology, which combines cloud particle types, cloud fluctuation range and environmental boundary conditions to accurately capture the core characteristics of complex weather phenomena. Secondly, this method uses a multi-head self-attention mechanism to dynamically assign weights and perform deep modeling and correlation analysis on cloud characteristic data in the spatiotemporal dimension. This mechanism can not only highlight key weather features and capture subtle changes in time series, but also enhance the precise understanding of spatial distribution characteristics, providing a reliable basis for risk warnings of severe convective weather such as thunderstorms, hail, and heavy rains.

[0024] 3. By introducing layer normalization and polarization error compensation technology, the robustness to observation data noise and computational stability are greatly improved. Normalization reduces the interference of different data scales on model performance, while error compensation effectively eliminates the system error caused by equipment deviation, making the results more accurate and reliable. Combining the big data model of electromagnetic wave scattering parameter influencing factors and parameter boundary condition judgment results, this method can set a reasonable fluctuation threshold based on historical distribution parameters, and judge whether the current cloud characteristics exceed the safety range through deviation measurement. This big data-driven analysis framework not only enhances the timeliness and adaptability of the model, but also dynamically adjusts the warning level according to real-time observation data, significantly improving the accuracy of extreme weather forecasts. At the same time, this method has strong scalability and universality, and can adapt to the warning needs of different weather types (such as heavy rain, strong wind, typhoon, etc.).

[0025] 4. Through the flexible embedding matrix and weight distribution mechanism, the key dimensions of the model can be adjusted according to the characteristics of the target weather phenomenon, thereby improving the detection effect in a targeted manner. This method not only reaches a new level in the accuracy and efficiency of weather detection by integrating multi-parameter collaborative analysis, multimodal data modeling and real-time warning strategies, but also provides strong technical support for improving meteorological disaster monitoring capabilities and reducing disaster risks. This comprehensive advantage makes it an important tool for dealing with complex meteorological conditions and extreme weather, and provides broad application prospects in the fields of meteorological services, public safety and disaster prevention and mitigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. It should be understood that the following drawings only illustrate certain embodiments of the present invention and should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0027] Figure 1 is a flow chart of the steps of the method of the present invention;

[0028] Figure 2 It is a module composition diagram of the system of the present invention;

[0029] Figure 3 This is a composition diagram of the risk warning modules for different weather types of the present invention. DETAILED DESCRIPTION

[0030] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application may be combined with each other. The present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0031] like Figure 1 As shown, in this embodiment, the weather detection method based on the meteorological rain measuring radar includes the following steps:

[0032] Step S1, using a meteorological rainfall radar to collect electromagnetic wave scattering parameters of different cloud layer characteristic weather in the monitoring area, which are used as electromagnetic wave scattering parameters of different cloud layer characteristic weather acquired by different cloud layer characteristic weather detection;

[0033] Step S2, integrating electromagnetic wave scattering parameter sets of different cloud layer characteristic weather, including electromagnetic wave scattering parameters of different unit times, each of the electromagnetic wave scattering parameters of the unit time including a single electromagnetic wave scattering parameter of different cloud layer characteristic weather and corresponding cloud particle types of different cloud layer characteristic weather, the cloud particle types corresponding to the echo intensity and polarization parameters of different cloud layer characteristic weather, the polarization parameters corresponding to the differential reflectivity of different cloud layer characteristic weather;

[0034] Step S3, using the electromagnetic wave scattering parameter set to establish a weather analysis and warning model for multi-modal fusion weight analysis, and using the model to analyze the influencing factors of the electromagnetic wave scattering parameters and determine the boundary conditions of the parameters;

[0035] Step S4, inputting the obtained electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation, so as to obtain the spectrum width and corresponding polarization phase difference of different cloud layer characteristic weather;

[0036] Step S5, combining the influencing factors of electromagnetic wave scattering parameters and the judgment results of parameter boundary conditions to judge the polarization phase difference in the big data model, and if the polarization phase difference fluctuation range is higher than the safety limit range, issuing risk warning information of different weather types;

[0037] In this embodiment, the electromagnetic wave scattering parameter set of different cloud layer characteristic weather may include electromagnetic wave scattering parameters of different periods and seasons of different cloud layer characteristic weather in the above-mentioned monitoring area (that is, the electromagnetic wave scattering parameters per unit time). The cloud particle type may include the spectrum width and polarization parameters of the above-mentioned different cloud layer characteristic weather, wherein the number of electromagnetic wave scattering parameters per unit time in the electromagnetic wave scattering parameter set may be set according to specific circumstances, and is not specifically limited here.

[0038] In this embodiment, the electromagnetic wave scattering parameter set can use a weather analysis and warning model of multimodal fusion weight analysis to calculate the radar time series graph neural network detection algorithm, and use the model to analyze the influencing factors of the electromagnetic wave scattering parameters and judge the boundary conditions of the parameters. Understandably, the electromagnetic wave scattering parameter set is calculated to obtain the electromagnetic wave scattering parameter characteristics corresponding to different cloud layer characteristic weather under each polarization parameter.

[0039] The weather data set control management center or the edge computing gateway server can set up a radar time series graph neural network detection algorithm. Of course, the weather data set control management center and the edge computing gateway server can also set up a radar time series graph neural network detection algorithm to detect the electromagnetic wave scattering parameters of weather with different cloud layer characteristics. If the weather data set control management center is used to detect the electromagnetic wave scattering parameters of weather with different cloud layer characteristics, the computing power of the edge computing gateway server can be shared to reduce the load of the edge computing gateway server. In addition, the weather data set control management center directly detects the electromagnetic wave scattering parameters of weather with different cloud layer characteristics, which reduces the upload of electromagnetic wave scattering parameter data of weather with different cloud layer characteristics, and can improve the speed of analysis and calculation.

[0040] In this embodiment, when integrating electromagnetic wave scattering parameters of weather with different cloud layer characteristics, they can be directly obtained manually, or they can be obtained using equipment, tools, etc. Specifically, for example, relevant personnel go to the production workshop to use a meteorological rain measuring radar to shoot electromagnetic wave scattering parameters of weather with different cloud layer characteristics, and then the meteorological rain measuring radar uploads the obtained electromagnetic wave scattering parameters of weather with different cloud layer characteristics to the edge computing gateway server through the first data interaction interface. The method of obtaining electromagnetic wave scattering parameters is not specifically limited here. The obtained electromagnetic wave scattering parameters of weather with different cloud layer characteristics only include different cloud layer characteristic weather per unit time, so as to detect different weather type risk warnings for such different cloud layer characteristic weather. If the electromagnetic wave scattering parameters of weather with different cloud layer characteristics include different cloud layer characteristic weather in multiple time periods, the electromagnetic wave scattering parameters can be divided so that each electromagnetic wave scattering parameter of weather with different cloud layer characteristics includes different cloud layer characteristic weather per unit time.

[0041] In this embodiment, the weather analysis and warning model of the multimodal fusion weight analysis is expressed as:

[0042] in, represents the weights of factors affecting different electromagnetic wave scattering parameters, Indicates the echo strength, represents the polarization parameter, Represents the differential reflectivity of different cloud characteristics, represents the parameter boundary conditions, Cloud particle characteristic parameter matrix, represents polarization error compensation, LN is layer normalization, Dropout( ) is the preset weight distribution calculation, represents the multi-head self-attention mechanism, , , They are weather embedding matrices of different cloud layer characteristics, corresponding to query, key and value, is the learnable matrix of multi-head self-attention in different unit times.

[0043] In other embodiments, the collected electromagnetic wave scattering parameters of different cloud layer characteristic weather do not need to be uploaded to the edge computing gateway server. The collected electromagnetic wave scattering parameters can be directly input into the radar time series graph neural network detection algorithm in the weather data control management center for statistical detection, and then the occurrence period of different cloud layer characteristic weather is predicted, and different weather type risk warning information is issued according to the occurrence period.

[0044] The radar time series graph neural network detection algorithm can be used in parallel to perform statistical detection on electromagnetic wave scattering parameters of different cloud layer characteristic weather at the same time to improve the detection efficiency. The radar time series graph neural network detection algorithm can be used for the electromagnetic wave scattering parameter characteristics of electromagnetic wave scattering parameters of different cloud layer characteristic weather, and the extracted electromagnetic wave scattering parameter characteristics are matched with the electromagnetic wave scattering parameter characteristics obtained by the above calculation to realize the analysis and calculation of different cloud layer characteristic weather.

[0045] Furthermore, the electromagnetic wave scattering parameter characteristics of the electromagnetic wave scattering parameters of the different cloud layer characteristic weather are matched with the electromagnetic wave scattering parameter characteristics of the electromagnetic wave scattering parameters per unit time in the electromagnetic wave scattering parameter set in the radar time series graph neural network detection algorithm to obtain the compliance rate of the electromagnetic wave scattering parameters of the different cloud layer characteristic weather and the electromagnetic wave scattering parameters per unit time; the electromagnetic wave scattering parameters per unit time with the largest compliance rate are selected from the electromagnetic wave scattering parameter set, and the cloud particle type of the selected electromagnetic wave scattering parameters per unit time is used as the actual monitoring value of the electromagnetic wave scattering parameters of the different cloud layer characteristic weather to obtain the spectral width and polarization parameters of the different cloud layer characteristic weather.

[0046] Furthermore, the echo ratios corresponding to the polarization parameters of different cloud layer characteristic weather in the monitoring area collected by the meteorological rain measuring radar within a preset time are counted. Among them, the preset time can be set according to the specific situation. By inputting the electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm, efficient and accurate analysis and calculation are performed. This method makes full use of the multi-parameter observation capability of the meteorological rain measuring radar, especially the key indicators such as the polarization parameters and echo ratios of different cloud layer characteristic weather in the monitoring area within a predetermined time, and combines the time series graph neural network model to deeply mine and analyze the data. Through the input radar data, the model can capture the dynamic evolution law of cloud layer characteristic weather in time and space dimensions, and at the same time associate the interaction of different electromagnetic wave scattering parameters to generate more accurate weather analysis results.

[0047] The core steps include: First, based on the meteorological rain radar equipment, collect the electromagnetic wave scattering parameters within a specific time range in the monitoring area, including polarization parameters, differential reflectivity, echo intensity, Doppler velocity, etc. These parameters can characterize the characteristics of cloud particles (such as shape, size and number) and the distribution dynamics of precipitation particles. Then, by inputting these data into the radar time series graph neural network detection algorithm, the spatial and temporal modeling ability of the neural network is used to perform feature learning and comprehensive analysis on the radar time series data and graph structure data. In particular, the relationship between statistical polarization parameters and echo ratios provides a basis for the classification and dynamic monitoring of cloud weather systems. In addition, by correlating different time series data in the radar monitoring area, the model can achieve accurate prediction of the dynamic characteristics of the cloud layer.

[0048] The significant advantage of this method is that the radar time series graph neural network model can effectively fuse multimodal data and capture spatiotemporal features, solving the problem of insufficient capture of dynamic weather changes in traditional meteorological detection. At the same time, the introduction of polarization parameters and echo ratios makes the monitoring of complex cloud features (such as convective weather, thunderstorm clouds, hail, etc.) more accurate, providing scientific support for early warning of extreme weather. This method has shown a wide range of application value in heavy rain monitoring, hail warning, thunderstorm tracking, etc.

[0049] The graph neural network detection algorithm includes:

[0050] (1) Data input and preprocessing, assuming that radar observation data Include time points, the data at each time point is a feature vector, and the data can be constructed as a graph structure:

[0051] in, Indicates The feature vector at time, is the feature dimension. The radar data is constructed as a time series graph ,in represents a set of nodes (corresponding to time points), Represents the connection relationship between time points (adjacent relationship in time or space).

[0052] (2) Graph Neural Network Modeling, 2.1 Adjacency Matrix Definition, Constructing the Adjacency Matrix of Radar Observations :

[0053] 2.2 Graph Convolution Operation,The update rules of the graph convolution layer are as follows:

[0054] in: It is The node feature matrix of the layer; It is The weight matrix of the layer; is the bias term; is the activation function (such as ReLU). The initial input features is the time series characteristic of radar .

[0055] (3) Time series modeling, 3.1 Time series feature extraction (LSTM / GRU), perform time series modeling on the features output by the graph neural network, and use the LSTM network to extract time-dependent features:

[0056] in: It is Graph convolution features at each moment; is the hidden state output by the LSTM.

[0057] 3.2 Time series feature aggregation, aggregate the hidden state extracted by LSTM in the time direction:

[0058] in, It is the aggregated time series feature.

[0059] (4) Multimodal fusion and classification, 4.1 Feature fusion, the spatial features output by the graph neural network Time features with LSTM output Fusion:

[0060] Among them, Concat represents the feature concatenation operation.

[0061] 4.2 Classification and prediction, use the fully connected layer to classify the fusion features:

[0062] in, is the type of weather predicted (such as heavy rain, hail, etc.); and are the weights and biases of the fully connected layer.

[0063] (5) Risk assessment of polarization parameters. Combining historical data and the upper and lower bounds of polarization parameters, risk assessment is performed using the conditional discriminant function:

[0064] in, is the polarization parameter detected by the current radar; is the preset risk threshold, It is an indicator function used to determine whether it exceeds the safety range.

[0065] The present invention can simultaneously capture the temporal dynamics and spatial structural characteristics of radar data through the graph neural network detection algorithm, and improve the prediction accuracy of complex weather phenomena. The fusion of big data models and neural networks is suitable for meteorological forecasting tasks with high real-time and high precision requirements.

[0066] The polarization phase difference is judged in the big data model by combining the influencing factors of electromagnetic wave scattering parameters and the judgment results of parameter boundary conditions. If the fluctuation range of the polarization phase difference is higher than the safety limit range, different weather type risk warning information is issued; including:

[0067] (a) The historical statistical model of polarization parameters is established based on the influencing factors of electromagnetic wave scattering parameters and the judgment results of parameter boundary conditions. It is used to represent the parameter fluctuation range under a certain weather type.

[0068] Assume that the polarization parameter set is ,in For the The polarization parameter value of the observation.

[0069] Calculate historical mean of polarization parameters and standard deviation :

[0070] Determine the safe fluctuation range of a certain weather type based on historical data ,in It is a parameter that controls the safety range (usually 2 or 3 represents a 95% or 99% confidence interval).

[0071] (b) Real-time polarization phase difference fluctuation judgment, real-time polarization phase difference fluctuation range judgment formula:

[0072] Assume that the polarization parameter value at the current moment is , determine whether it exceeds the safety limit range:

[0073] in, Indicates the degree of deviation of polarization parameters.

[0074] Conditional judgment:

[0075] when , it means that the current polarization parameters are out of the safe range and an early warning needs to be triggered.

[0076] (c) Combined with multi-parameter big data model calculation, in practical applications, big data models will combine multiple polarization parameters (such as Assume that multiple polarization parameters constitute a feature vector ,in, For the polarization parameters.

[0077] Calculate the deviation of the real-time polarization parameter vector from the historical distribution:

[0078] in: and For the The historical mean and standard deviation of the polarization parameters. Represents the normalized Euclidean distance of the integrated polarization parameters.

[0079] Conditional judgment:

[0080] (d) Weather type risk warning formula: According to the fluctuation of polarization phase difference and historical weather data, the risk is classified into different levels and the corresponding warning is triggered: Let the polarization phase difference risk weight corresponding to each weather type be and impact factor ,in Indicates weather type (such as heavy rain, hail, etc.).

[0081] Calculating a composite risk score :

[0082] in: Indicates the impact weight of weather type (for example, if the risk of heavy rain is high, a higher value). Represents the relevance weight for a specific weather type.

[0083] Comprehensive judgment warning level:

[0084] in, and is the threshold of the warning level.

[0085] like Figure 2 and Figure 3 As shown, a weather detection system based on a meteorological rain measuring radar comprises: a meteorological rain measuring radar, a weather data set control and management center for data interaction with the meteorological rain measuring radar, and different weather type risk warning modules connected to the weather data set control and management center. The different weather type risk warning modules comprise: a radar electromagnetic wave scattering parameter analysis interface, an initial information comprehensive preprocessing interface, a different weather type risk warning judgment interface, an electromagnetic wave scattering parameter acquisition interface, and a weather analysis warning model establishment and judgment interface.

[0086] Weather comprehensive management personnel can query historical statistical detection records and risk warning information records of different weather types through the weather data control management center. In addition, weather comprehensive management personnel can query the current proportion of polarization parameters of different cloud layer characteristic weather in different regions and corresponding echo ratios and other risk warning information of different weather types per unit time through the weather data control management center. Based on the above design, the method provided by the present invention can replace the artificial detection of spectral width and polarization parameters of different cloud layer characteristic weather by the edge computing gateway server or the weather data control management center, simplifying the steps of detecting different cloud layer characteristic weather and improving the detection accuracy. The method collects and integrates electromagnetic wave scattering parameters, combines polarization parameters, differential reflectivity, cloud particle characteristic matrix and other multimodal information, and uses a multi-head self-attention mechanism to dynamically allocate weights, thereby realizing accurate modeling and analysis of spatiotemporal characteristics. In addition, the method introduces a historical big data model to determine the range of parameter fluctuations, improves the detection capability and warning accuracy of extreme weather, has the advantages of strong robustness, high adaptability and good scalability, and can be applied to real-time risk assessment and warning of various complex weather types.

[0087] In this embodiment, the method for risk warning of different weather types by detecting different cloud layer characteristic weather may include a meteorological rain measuring radar, a weather data monitoring and management center, and risk warning modules of different weather types. The meteorological rain measuring radar and the weather data monitoring and management center use 5G signals for data exchange, and are used to upload the collected electromagnetic wave scattering parameters of different cloud layer characteristic weather in the monitoring area to the weather data monitoring and management center, so as to analyze and calculate the electromagnetic wave scattering parameters of different cloud layer characteristic weather, and issue risk warning information of different weather types when the detection result exceeds the boundary threshold.

[0088] The different weather types risk warning module may include a radar electromagnetic wave scattering parameter analysis interface and a different weather type risk warning judgment interface. In other embodiments, the different weather types risk warning module may also include an electromagnetic wave scattering parameter acquisition interface and a weather analysis warning model establishment and judgment interface.

[0089] The radar electromagnetic wave scattering parameter analysis interface is used to input the electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation to obtain the spectral width of different cloud layer characteristic weather and the corresponding polarization phase difference.

[0090] The risk warning judgment interface of different weather types is used to judge the polarization parameter. When the fluctuation range of the polarization phase difference is higher than the safety limit range, different weather type risk warning information is issued.

[0091] Furthermore, the risk warning judgment interface for different weather types is also used to obtain the current local meteorological information, calculate the occurrence period of the different cloud characteristic weather according to the meteorological information and polarization parameters, and issue risk warning information for different weather types.

[0092] The electromagnetic wave scattering parameter acquisition interface is used to integrate the electromagnetic wave scattering parameter sets of different cloud layer characteristic weather conditions, including electromagnetic wave scattering parameters of different unit times, each of the electromagnetic wave scattering parameters of the unit time including a single electromagnetic wave scattering parameter of a different cloud layer characteristic weather condition and the corresponding cloud particle type of the different cloud layer characteristic weather condition, the cloud particle type corresponding to the echo intensity and polarization parameter of the different cloud layer characteristic weather condition, wherein the polarization parameter includes the differential reflectivity of the different cloud layer characteristic weather conditions.

[0093] The weather analysis and warning model establishment and judgment interface is used to use the electromagnetic wave scattering parameter set, adopt the convolutional neural network model to optimize the radar time series graph neural network detection algorithm, and use the model to analyze the influencing factors of electromagnetic wave scattering parameters and judge the parameter boundary conditions.

[0094] In an embodiment of the present invention, the method for detecting different cloud layer characteristic weather may include an edge computing gateway server, a meteorological rain radar, and a weather data monitoring and management center. The edge computing gateway server and the meteorological rain radar use 5G signals to exchange data through the network, and are used to perform statistical detection of the electromagnetic wave scattering parameters sent by the meteorological rain radar, and make judgments based on the detection results obtained. For example, when the polarization phase difference fluctuation range of different cloud layer characteristic weather is higher than the safety limit interval, different weather type risk warning information is issued. The weather data monitoring and management center can communicate with the edge computing gateway server through the network, and can also analyze and calculate the electromagnetic wave scattering parameters of different cloud layer characteristic weather. The weather comprehensive management personnel can understand the polarization parameters of different cloud layer characteristic weather in the monitoring area and the risk warning information of different weather types in the electromagnetic wave scattering parameters of different cloud layer characteristic weather collected by the meteorological rain radar through the weather data monitoring and management center. The method for detecting different cloud layer characteristic weather realizes the automatic detection of different cloud layer characteristic weather and the risk warning of different weather types, and improves the efficiency of detecting different cloud layer characteristic weather.

[0095] In this embodiment, the weather data collection and control management center is an integrated system platform, which is specially used to collect, store, process and manage weather data, and is a core hub for providing decision support for meteorological monitoring and forecasting. Its main function is to obtain large-scale meteorological data in real time through a variety of sensors (such as meteorological rainfall radars, satellites, ground meteorological stations, etc.), and to uniformly store, intelligently analyze and distribute these data to achieve efficient meteorological monitoring and forecasting capabilities.

[0096] In this embodiment, the meteorological rain radar may include a first processor, a first data interaction interface, a first memory, and an electromagnetic wave scattering parameter collection interface. The first processor, the first data interaction interface, the first memory, and the electromagnetic wave scattering parameter collection interface are electrically connected directly or indirectly to achieve data transmission or interaction. For example, these components can be electrically connected to each other through one or more communication buses or signal lines. Meteorological rain radar is a high-tech meteorological device for monitoring and analyzing the characteristics and distribution of precipitation in the atmosphere, and is widely used in weather forecasting and disaster warning. It captures the intensity, distribution, and dynamic changes of precipitation in real time by emitting high-frequency electromagnetic waves and using the reflection characteristics of precipitation particles (such as raindrops, snowflakes, or hail) to electromagnetic waves. According to the intensity, phase, and frequency offset of the echo signal, the radar can obtain key information such as the reflectivity, particle velocity, and particle distribution of precipitation, thereby analyzing the intensity, movement direction, and characteristics of the weather system of precipitation. Doppler radar further measures the speed of precipitation particles through the Doppler effect, while polarization radar distinguishes precipitation types (such as rain, snow, and hail) and particle shapes through the difference between horizontal and vertical polarization waves. Meteorological precipitation radar has the advantages of strong real-time performance, wide coverage, and high accuracy. It can quickly monitor precipitation distribution and wind field dynamics over a large area, and plays a key role in the early warning of extreme weather such as heavy rain, hail, and thunderstorms. Although there are blind spots in its close-range detection and it may be affected by signal attenuation at long distances, these limitations are gradually overcome through the integration of multi-source radar networks. As an important tool for modern meteorological observation, meteorological precipitation radar is widely used in precipitation monitoring, disaster prevention and mitigation, aviation safety, and water resources management, providing important technical support for improving the accuracy of weather forecasts.

[0097] The meteorological rainfall radar can also include a low-stray local oscillator generation interface composed of a constant temperature clock source, a low-stray phase-locked loop, an operational amplifier, etc.; a frequency synthesis interface composed of a low-stray phase-locked loop, a DAC, a digital DDS, etc.; a transceiver link interface; and a data processing interface. In specific implementation: a stable local oscillator signal is generated through the low-stray local oscillator generation interface; a high-purity intermediate frequency signal is generated through the frequency synthesis interface; the signal frequency is increased, amplified, and filtered through the transmission link interface, and the signal is output to the transmitting antenna; the received signal is amplified, filtered, frequency-reduced, and digitized through the receiving link interface; and the signal is collected, processed, and output through the data processing interface.

[0098] In this embodiment, the electromagnetic wave scattering parameter acquisition interface is a key device for capturing and analyzing the scattering characteristics of electromagnetic waves on target objects (such as precipitation particles, clouds or ground surfaces), and is widely used in meteorological detection, radar systems, environmental monitoring and other fields. The interface transmits high-frequency electromagnetic waves, receives echo signals reflected by the target, and extracts scattering characteristic parameters such as reflectivity, polarization characteristics, phase difference and Doppler frequency shift. These parameters can reveal the size, shape, material and motion state of the target, such as precipitation intensity, precipitation type (such as rain, snow or hail) and particle distribution and movement direction in meteorological applications. The acquisition interface usually includes a transmitter, a receiver and a signal processing module, and uses modern electronic technology and algorithms to filter, amplify and digitize the echo signal to achieve high-precision data acquisition. By combining with the radar system, the electromagnetic wave scattering parameter acquisition interface can be used to monitor the target characteristics in a complex environment in real time, support weather forecasting, disaster prevention and mitigation, and atmospheric science research. Its advantages include strong real-time performance, high accuracy, and wide adaptability, but it may also be affected by terrain shielding, environmental noise and multipath effects. With the advancement of technology, the interface has been continuously improved in detection range, sensitivity and data processing capabilities, becoming an indispensable core module in modern radar and meteorological monitoring systems.

[0099] A meteorological rainfall radar is established through the network and an edge computing gateway server uses 5G signals to exchange data, and sends and receives data through the network.

[0100] In this embodiment, the edge computing gateway server and the weather data set control management center use 5G signals to exchange data, and may have the same or similar structure as the weather data set control management center, which can be used to statistically detect electromagnetic wave scattering parameters of different cloud layer characteristic weather, and predict the occurrence period of different cloud layer characteristic weather, and send different weather type risk warning information to the weather data set control management center according to the occurrence period of polarization parameters corresponding to different cloud layer characteristic weather. Of course, in other embodiments, the different cloud layer characteristic weather detection method may not include an edge computing gateway server, and the functions performed by the edge computing gateway server may be replaced by the weather data set control management center, which will not be repeated here.

[0101] The present invention also proposes a storage medium, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the steps in the weather detection method based on the meteorological rainfall radar are executed.

[0102] In summary, the present invention provides a weather detection method, system and storage medium based on a meteorological rain measuring radar. The method uploads the collected electromagnetic wave scattering parameters of different cloud layer characteristic weather in the monitoring area to an edge computing gateway server through a meteorological rain measuring radar; the edge computing gateway server inputs the acquired electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar timing graph neural network detection algorithm for analysis and calculation to obtain the spectral width and corresponding polarization phase difference of different cloud layer characteristic weather; the polarization parameter is judged, and when the fluctuation range of the polarization phase difference is higher than the safety limit interval, different weather type risk warning information is issued. The method can automatically detect the spectral width and corresponding polarization phase difference of different cloud layer characteristic weather, thereby improving the accuracy and efficiency of detecting different cloud layer characteristic weather.

[0103] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A weather detection method based on a meteorological rainfall radar, characterized in that: The method includes: Step S1, using a meteorological rainfall radar to collect electromagnetic wave scattering parameters of different cloud layer characteristic weather in the monitoring area, which are used as electromagnetic wave scattering parameters of different cloud layer characteristic weather acquired by different cloud layer characteristic weather detection; Step S2, integrating electromagnetic wave scattering parameter sets of different cloud layer characteristic weather, including electromagnetic wave scattering parameters of different unit times, each of the electromagnetic wave scattering parameters of the unit time including a single electromagnetic wave scattering parameter of different cloud layer characteristic weather and corresponding cloud particle types of different cloud layer characteristic weather, the cloud particle types corresponding to the echo intensity and polarization parameters of different cloud layer characteristic weather, the polarization parameters corresponding to the differential reflectivity of different cloud layer characteristic weather; Step S3, using the electromagnetic wave scattering parameter set to establish a weather analysis and warning model for multi-modal fusion weight analysis, and using the model to analyze the influencing factors of the electromagnetic wave scattering parameters and determine the boundary conditions of the parameters; Step S4, inputting the obtained electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation, so as to obtain the spectrum width and corresponding polarization phase difference of different cloud layer characteristic weather; Step S5, combining the influencing factors of electromagnetic wave scattering parameters and the judgment results of parameter boundary conditions to judge the polarization phase difference in the big data model, and if the polarization phase difference fluctuation range is higher than the safety limit range, issuing risk warning information of different weather types; The weather analysis and warning model of the multi-modal fusion weight analysis is expressed as: ； in, represents the weights of factors affecting different electromagnetic wave scattering parameters, Indicates the echo strength, represents the polarization parameter, Represents the differential reflectivity of different cloud characteristics, represents the parameter boundary conditions, Cloud particle characteristic parameter matrix, represents polarization error compensation, LN is layer normalization, Dropout( ) is the preset weight distribution calculation, represents the multi-head self-attention mechanism, , , They are weather embedding matrices of different cloud layer characteristics, corresponding to query, key and value, is the learnable matrix of multi-head self-attention in different unit times.

2. The weather detection method based on meteorological rainfall radar according to claim 1 is characterized in that: The obtained electromagnetic wave scattering parameters of different cloud layer characteristic weather are input into the radar time series graph neural network detection algorithm for analysis and calculation, including statistically analyzing the echo ratios corresponding to the polarization parameters of different cloud layer characteristic weather in the monitoring area collected by the meteorological rain measuring radar within a predetermined time period.

3. The weather detection method based on meteorological rainfall radar according to claim 1 is characterized in that: When the polarization phase difference fluctuation range is higher than the safety limit interval, different weather type risk warning information is issued, including obtaining the cloud height, vertical distribution data, and cloud structure changes and development data for the weather in different terrains and altitudes, and calculating the alarm range of different cloud characteristic weather according to regional information and polarization parameters, and issuing different weather type risk warning information. The different weather type risk warning information includes rainfall probability, thunderstorm probability, climate temperature and humidity, haze probability and level, secondary comprehensive disaster probability and level, where different probabilities and levels correspond to different degrees of severe convective weather hazard.

4. The weather detection method based on meteorological rainfall radar according to claim 1 is characterized in that: The method uses 5G signals to exchange data between an edge computing gateway server and a weather data set control and management center. The edge computing gateway server is provided with a weather analysis and early warning model with multimodal fusion weight analysis. A meteorological rain measuring radar is used to collect electromagnetic wave scattering parameters of weather characteristics of different cloud layers in a monitored area. The steps of using the electromagnetic wave scattering parameters of weather characteristics of different cloud layers as obtained by a method for detecting weather characteristics of different cloud layers include: the meteorological rain measuring radar uploads the collected electromagnetic wave scattering parameters of weather characteristics of different cloud layers in a monitored area to the edge computing gateway server.

5. The weather detection method based on meteorological rainfall radar according to claim 1 is characterized in that: The step of inputting the obtained electromagnetic wave scattering parameters of weather with different cloud layer characteristics into the radar time series graph neural network detection algorithm for analysis and calculation includes: the edge computing gateway server inputs the obtained electromagnetic wave scattering parameters of weather with different cloud layer characteristics in the monitoring area into the radar time series graph neural network detection algorithm of the edge computing gateway server to analyze and calculate the electromagnetic wave scattering parameters of weather with different cloud layer characteristics in the monitoring area.

6. A weather detection system based on meteorological rain radar, characterized in that: The method according to any one of claims 1 to 5 is implemented by a weather detection system based on a meteorological rainfall radar, the system comprising: The meteorological rainfall radar, the weather data set control and management center that interacts with the meteorological rainfall radar, and the different weather type risk warning modules connected to the weather data set control and management center use the meteorological rainfall radar to collect the electromagnetic wave scattering parameters of different cloud layer characteristic weather in the monitoring area, which are used as different cloud layer characteristic weather detection and different weather type risk warning methods to obtain the electromagnetic wave scattering parameters of different cloud layer characteristic weather; the weather data set control and management center includes the radar time series graph neural network detection algorithm, and the different weather type risk warning modules include: The radar electromagnetic wave scattering parameter analysis interface is used to input the electromagnetic wave scattering parameters of different cloud layer characteristic weather into the radar time series graph neural network detection algorithm for analysis and calculation, so as to obtain the spectrum width and corresponding polarization phase difference of different cloud layer characteristic weather; The initial information comprehensive preprocessing interface is used to extract features from the unit time acquisition data corresponding to different cloud layer characteristic weather obtained from the radar electromagnetic wave scattering parameter analysis interface, denoise and cluster the extracted data, and transmit the clustered data to the risk warning judgment interface of different weather types; The risk warning judgment interface for different weather types is used to combine the influencing factors of electromagnetic wave scattering parameters and the judgment results of parameter boundary conditions to make polarization phase difference judgments in the big data model. If the polarization phase difference fluctuation range is higher than the safety limit range, risk warning information for different weather types will be issued.

7. The weather detection system based on meteorological rainfall radar according to claim 6 is characterized in that: The different weather type risk warning modules also include: The electromagnetic wave scattering parameter acquisition interface is used to integrate the electromagnetic wave scattering parameter sets of different cloud layer characteristic weather, including electromagnetic wave scattering parameters of different unit times, each of which includes a single electromagnetic wave scattering parameter of different cloud layer characteristic weather and the corresponding cloud particle type of different cloud layer characteristic weather, and the cloud particle type corresponds to the echo intensity and polarization parameter of different cloud layer characteristic weather; The weather analysis and early warning model establishment and judgment interface is used to use the electromagnetic wave scattering parameter set to establish a weather analysis and early warning model for multi-modal fusion weight analysis, and use the model to analyze the influencing factors of electromagnetic wave scattering parameters and judge the parameter boundary conditions.

8. The weather detection system based on meteorological rainfall radar according to claim 6 is characterized in that: The risk warning judgment interface for different weather types is also used for: the height of the cloud layer, vertical distribution data, structural changes and development data of the cloud layer in different terrains and altitudes, calculates the cumulative number and incremental number of different cloud characteristic weather in the area according to regional information and polarization parameters, and issues risk warning information for different weather types. The risk warning information for different weather types includes rainfall probability, thunderstorm probability, climate temperature and humidity, haze probability and level, secondary comprehensive disaster probability and level, where different probabilities and levels correspond to different degrees of severe convective weather hazards.

9. A storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, the steps of the weather detection method based on meteorological rainfall radar as described in any one of claims 1 to 5 are executed.

Images