CN108537368A - A lightning warning method, device and system - Google Patents

Abstract

The embodiment of the invention provides a lightning early warning method, a lightning early warning device and a lightning early warning system. The method comprises the following steps: acquiring satellite-borne lightning monitoring data and foundation lightning monitoring data of the current time period in the detection area and performing cluster analysis; acquiring the center position of a thunderstorm cloud, the range of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud in a detection area; and according to a thunderstorm cloud feature evolution sequence formed by the thunderstorm cloud center position, the thunderstorm cloud range and the lightning quantity in the thunderstorm cloud, predicting by using a lightning prediction model so as to realize tracking and extrapolation of the thunderstorm cloud and obtain lightning early warning information at the next moment. The apparatus is configured to perform the method. According to the embodiment of the invention, the thunderstorm cloud development characteristics are obtained by clustering analysis of the satellite-borne lightning monitoring data and the foundation lightning monitoring data, then the lightning prediction model is used for prediction to obtain the lightning early warning information of the next moment, the thunderstorm development and evolution can be predicted through machine learning, and the accuracy of lightning prediction is improved.

Description

A lightning warning method, device and system

technical field

Embodiments of the present invention relate to the technical field of weather forecasting, and in particular, to a lightning warning method, device and system.

Background technique

Lightning is one of the "Top Ten Natural Disasters" announced by the United Nations. Due to its huge destructive power, it causes a large number of casualties across the country every year. At the same time, because lightning releases a large amount of electric charges instantly, resulting in extremely strong electromagnetic pulses, it often poses a major threat to increasingly sophisticated electronic instruments. Therefore, the early warning and forecast of lightning is of great significance.

Thunderstorm activities are always accompanied by strong discharge phenomena, and lightning activity can well reflect changes in the intensity and movement trends of thunderstorm activities. In recent years, the national lightning monitoring network has become more and more perfect. The ground-based lightning location network can monitor cloud-ground lightning activities nationwide, and the Lightning Mapping Imager (LMI) on Fengyun Satellite can realize the observation of thunderstorm cloud top discharge, thus realizing All-round observation of lightning across the country.

Lightning observation data has the characteristics of high temporal and spatial resolution, low transmission delay, etc., which is of great significance for real-time monitoring of fast generation and disappearance of small and medium-scale convective systems. Therefore, it has certain advantages to directly use lightning data to identify, track and extrapolate thunderstorm clouds.

At present, the lightning near warning mainly relies on the near extrapolation of Doppler weather radar, meteorological satellite and other monitoring data. Extrapolation algorithms such as TITAN (Thunderstorm identification, tracking, analysis, and nowcasting) of Doppler radar, SCIT (Storm Cell Identification and Tracking), and optical flow method can extrapolate the convective system in the next 0-1 hours to a certain extent activity status. The traditional extrapolation is mostly linear extrapolation, which cannot realize the forecast of thunderstorm turning, and has no ability to predict the strengthening or weakening of thunderstorm, resulting in relatively low prediction accuracy.

Contents of the invention

Aiming at the problems existing in the prior art, the embodiments of the present invention provide a lightning warning method, device and system.

In the first aspect, an embodiment of the present invention provides a lightning early warning method, including:

Obtain the spaceborne lightning monitoring data and ground-based lightning monitoring data in the current period in the detection area;

According to the cluster analysis of the spaceborne lightning monitoring data and the ground-based lightning monitoring data, the development characteristics of the thunderstorm cloud in the detection area are obtained, and the development characteristics of the thunderstorm cloud include: the central position of the thunderstorm cloud, the range of the thunderstorm cloud and the the number of lightning bolts in the cloud;

According to the characteristic evolution sequence of the thunderstorm cloud formed by the central position of the thunderstorm cloud, the range of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud, the lightning prediction model is used to predict, so as to realize the tracking and extrapolation of the thunderstorm cloud, and obtain Lightning warning information at the next moment.

In a second aspect, an embodiment of the present invention provides a lightning warning device, including:

The obtaining module is used to obtain the spaceborne lightning monitoring data and the ground-based lightning monitoring data in the current period in the detection area;

The clustering module is used to perform cluster analysis according to the spaceborne lightning monitoring data and the ground-based lightning monitoring data to obtain the development characteristics of thunderstorm clouds in the detection area, and the development characteristics of thunderstorm clouds include: the center position of thunderstorm clouds , thunderstorm cloud range and the number of lightning in the thunderstorm cloud;

A prediction module, configured to use a lightning forecasting model to perform predictions based on the thunderstorm cloud characteristic evolution sequence formed by the thunderstorm cloud center position, the thunderstorm cloud range, and the number of lightning in the thunderstorm cloud, so as to realize the tracking of the thunderstorm cloud and extrapolation to obtain the lightning warning information at the next moment.

In a third aspect, an embodiment of the present invention provides a lightning early warning system, including: the lightning early warning device and terminal described in the second aspect; wherein,

The lightning warning device sends lightning warning information to the terminal;

The terminal receives the lightning warning information, and displays the lightning warning information.

In a fourth aspect, an embodiment of the present invention provides an electronic device, including: a processor, a memory, and a bus, wherein,

The processor and the memory communicate with each other through the bus;

The memory stores program instructions executable by the processor, and the processor invokes the program instructions to execute the method steps of the first aspect.

In a fifth aspect, an embodiment of the present invention provides a non-transitory computer-readable storage medium, including:

The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the method steps of the first aspect.

A lightning early warning method, device, and system provided by the embodiments of the present invention perform cluster analysis on spaceborne lightning monitoring data and ground-based lightning monitoring data to obtain the development characteristics of thunderstorm clouds, and then use lightning prediction models to predict and obtain the next moment The lightning warning information can predict the development and evolution of thunderstorms through machine learning, and improve the accuracy of lightning prediction.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description These are some embodiments of the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

Fig. 1 is a schematic flow chart of a lightning early warning method provided by an embodiment of the present invention;

Fig. 2 is a structural schematic diagram of a lightning warning device provided by an embodiment of the present invention;

Fig. 3 is a schematic structural diagram of a lightning early warning system provided by an embodiment of the present invention;

FIG. 4 is a schematic diagram of a physical structure of an electronic device provided by an embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments It is a part of embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

Fig. 1 is a schematic flow chart of a lightning early warning method provided by an embodiment of the present invention. As shown in Fig. 1, the method includes:

Step 101: Obtain the spaceborne lightning monitoring data and ground-based lightning monitoring data in the current period in the detection area;

Specifically, the device acquires spaceborne lightning monitoring data and ground-based lightning monitoring data in the detection area for the current period. It should be noted that the cloud flash detection data of the Fengyun No. Ground-based cloud-ground lightning detection data from monitoring networks.

Step 102: Perform cluster analysis according to the spaceborne lightning monitoring data and the ground-based lightning monitoring data to obtain the development characteristics of the thunderstorm cloud in the detection area, and the development characteristics of the thunderstorm cloud include: the center position of the thunderstorm cloud, the The extent and amount of lightning in a thunderstorm cloud;

Specifically, according to the acquired spaceborne lightning monitoring data and ground-based lightning monitoring data in the detection area for the current period, cluster analysis is performed, and the machine learning clustering algorithm is used to calculate the distribution of "lightning clusters", so as to realize the identification of thunderstorm clouds , to obtain the development characteristics of the thunderstorm cloud. It should be noted that the development characteristics of the thunderstorm cloud include: the central position of the thunderstorm cloud, the range of the thunderstorm cloud, and the number of lightning in the thunderstorm cloud, and may also include other information, which is not specifically limited in the embodiment of the present invention.

Step 103: According to the characteristic evolution sequence of the thunderstorm cloud formed by the central position of the thunderstorm cloud, the scope of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud, use the lightning prediction model to make predictions, so as to realize the tracking and appearance of the thunderstorm cloud. Push to get the lightning warning information at the next moment.

Specifically, in the cluster analysis, the spaceborne lightning monitoring data and the ground-based lightning monitoring data can be clustered and analyzed according to the preset times to obtain multiple sub-development characteristics corresponding to the thunderstorm cloud, or the spaceborne lightning monitoring data can be analyzed first. Cluster analysis with the ground-based lightning monitoring data, and then divide the thunderstorm cloud development characteristics obtained after clustering according to the preset time, and obtain multiple sub-development characteristics corresponding to the thunderstorm cloud. Each sub-development characteristic includes the center position of the thunderstorm cloud, The scope of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud, a plurality of sub-development characteristics corresponding to the thunderstorm cloud constitute a thunderstorm cloud feature evolution sequence as input, input into the lightning prediction model, and predict by the lightning prediction model, so as to realize the thunderstorm cloud According to the tracking and extrapolation, thunderstorm cloud is the main source of lightning. According to the sounding results of Winn et al (1974), when the local electric field in the cloud exceeds about 400kV/m, lightning discharge can occur. Obtain the lightning warning information of the next moment. The development of thunderstorms has continuity, and the occurrence and development of thunderstorms follow the stages of convective generation, development, maturity, and extinction. Therefore, the early development of thunderstorms can provide information for the next development and movement. For the time series prediction of thunderstorm development, it is difficult for traditional neural network models to maintain the memory of thunderstorm development status. Relatively speaking, the recurrent (or recursive) neural network has a memory function, and can effectively use the previous activity device of the thunderstorm to realize the prediction of the future development of the thunderstorm. It should be noted that the lightning prediction model constructed by the recurrent (or recursive) neural network is pre-established and trained.

The embodiments of the present invention perform cluster analysis on the spaceborne lightning monitoring data and the ground-based lightning monitoring data to obtain the development characteristics of thunderstorm clouds, and then use the lightning prediction model to predict and obtain the lightning early warning information at the next moment, and can predict the development of thunderstorms through machine learning. Forecasting with evolution improves the accuracy of lightning forecasts.

On the basis of the foregoing embodiments, the method further includes:

The space-borne lightning monitoring data and the ground-based lightning monitoring data are respectively normalized, and the normalized space-borne lightning monitoring data and the ground-based lightning monitoring data are fused.

Specifically, for spaceborne lightning monitoring data and ground-based lightning monitoring data, quality control methods are designed respectively. For the ground-based lightning monitoring data, according to the synoptic laws of thunderstorm activity, the isolated lightning data in the time range T and space range D are filtered out. Among them, according to the law of thunderstorm activity, it belongs to the small-medium scale system, so T is taken as 30min, and D is taken as 20km.

For the spaceborne lightning monitoring data, use the observation data of other channels of the FY4 satellite, such as the infrared channel, water vapor and visible light channels, to conduct comprehensive observations. Lightning observation records corresponding to temperatures higher than -40 degrees Celsius were removed as noise.

The temporal and spatial resolutions of ground-based lightning monitoring data and spaceborne lightning monitoring data are quite different, so the two need to be fused to achieve comprehensive monitoring of lightning. The time resolution of FY4 lightning observation data is 1 minute, and the spatial resolution is 7.8 km, while the temporal and spatial resolution of ground-based lightning location data has reached the order of seconds and meters. In order to facilitate cluster analysis, the time of lightning detection data was normalized to 10min, and the spatial resolution was normalized to 7.8km. The normalized spaceborne lightning monitoring data and the ground-based lightning monitoring data are fused.

In the embodiment of the present invention, the space-borne lightning monitoring data and the ground-based lightning monitoring data with different orders of magnitude are normalized, so as to improve the accuracy of subsequent clustering analysis.

On the basis of the foregoing embodiments, the method further includes:

Obtain the historical thunderstorm cloud development characteristics in the historical time period, construct a time-series thunderstorm cloud development sample set according to the historical thunderstorm cloud development characteristics, train the cyclic or recursive neural network model according to the thunderstorm cloud development sample set, and obtain the obtained The lightning forecast model described above.

Specifically, obtaining the historical thunderstorm cloud development characteristics in the historical time period can be the historical thunderstorm cloud development characteristics in the nearest week from the current time. The historical thunderstorm cloud development characteristics include the center position of the thunderstorm cloud, the range of the thunderstorm cloud and the number of lightning in the thunderstorm , can also include information such as the moving path of the thunderstorm cloud. According to the development characteristics of historical thunderstorm clouds, the thunderstorm development sample set is constructed according to the time series, that is, with the development of time, the development characteristics of historical thunderstorm clouds corresponding to each historical moment, and the long short term memory model (Long short term memory model) is analyzed by using the thunderstorm development sample set. , referred to as LSTM) for training to obtain the lightning prediction model.

LSTM controls discarding or adding information through "gates", so as to realize the function of forgetting or remembering. A "gate" is a structure that selectively passes information, consisting of a sigmoid function and a dot product operation. The output value of the sigmoid function is in the [0,1] interval, 0 means completely discarded, and 1 means completely passed. An LSTM unit has three such gates, which are the forget gate, the input gate, and the output gate.

At time t, there are three inputs to LSTM: the input value x _t of the network in the current period, the output value h _t-1 of the LSTM at the previous moment, and the cell state C _t-1 at the previous moment; there are two outputs of the LSTM: the current Period LSTM output value h _t , and cell state C _t for the current period.

Forget gate: The forget gate is a sigmoid function that takes the output h _t-1 of the previous unit and the input x _t of this unit as input, and generates a value in [0,1] for each item in C _t-1 The value in to control the degree to which the last cell state is forgotten.

f _t ＝σ(W _f ·[h _t-1 ,x _t ]+b _f ) (1)

Among them, W _f is the neuron weight, b _f is the bias.

Input gate: The input gate works with a tanh function to control which new information is added. The tanh function produces a new candidate vector The input gate is Each item in produces a value in [0,1], controlling how much new information is added. So far, we already have the output _ft of the forget gate, which is used to control the degree of forgetting of the previous unit, and the output _it of the input gate, which is used to control how much new information is added, and we can update this memory unit The unit state is up,

f _t ＝σ(W _f ·[h _t-1 ,x _t ]+b _i ) (2)

Output gate: The output gate is used to control how much of the current cell state is filtered out. First activate the cell state, and the output gate produces a value in [0,1] for each of them, controlling the degree to which the cell state is filtered.

f _t ＝σ(W _f ·[h _t-1 ,x _t ]+b _o ) (4)

h _t ＝o _t ·tanh(C _t ) (5)

In the embodiment of this paper, the LSTM model is constructed according to the weather rules of the occurrence and development of thunderstorms, and the model is composed of LSTM layers and fully connected layers. LSTM needs to be trained before it is used, and a thunderstorm development sample set needs to be constructed. Including the moving path of the thunderstorm, the change of the lightning frequency, and the area of the thunderstorm.

The embodiment of the present invention establishes the LSTM model, trains the LSTM model to obtain the lightning prediction model, uses the lightning prediction model to predict, and obtains the lightning early warning information at the next moment. Forecast of the future development of thunderstorms.

On the basis of the above-mentioned embodiments, the cluster analysis according to the spaceborne lightning monitoring data and the ground-based lightning monitoring data includes:

according to or Calculate lightning density;

in, n is the total number of lightnings clustered, d _ij is the distance between two lightnings i and j, d _c is the interval threshold, and the number of lightnings with d _ij < d _c accounts for 2% of the total number of lightnings;

Calculate the lightning distance according to δ _i = min(d _ij ), ρ _j > ρ _i ;

Wherein, δ _i is the minimum distance between lightning j and lightning i, and the lightning density of lightning i is greater than that of lightning j;

Determine the center position of the thunderstorm, the range of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud according to the lightning density and the lightning distance.

Specifically, the clustering analysis is performed through the fast search clustering algorithm of the density maximum value, and the specific algorithm is as follows:

1. Lightning density calculation

in, n is the total number of lightnings clustered, d _ij is the distance between two lightnings i and j; d _c is the interval threshold, and the value of d _c should make the number of lightnings with d _ij < d _c account for about 2% of the total .

Use the formula (6) to calculate the density of each lightning in the detection area, that is to calculate the number of lightning within the distance d _c of each lightning. The lightning density value is an important basis for identifying the center of a thunderstorm. The greater the lightning density, the denser the lightning and the more intense the discharge process.

In the actual clustering process, the Gaussian Kernel function can also be used to calculate the lightning density, as shown in formula (7):

The Gaussian kernel decays exponentially with distance from the center to the periphery, making it easier to identify a unique thunderstorm center point.

2. Lightning distance calculation

δ _i ＝min(d _ij ), ρ _j ＞ρ _i (8)

Using formula (8), for each lightning, calculate the minimum distance between all other lightnings with higher density than this lightning (for the lightning with the highest density, δ=max(d _ij )). For the lightning with larger δ, there are fewer scattered points around it, and the cluster independence in a certain area is higher.

3. Confirmation of thunderstorm cluster center

Arrange the lightning density values from high to low, and the confirmation of thunderstorm center lightning can be selected by given δ _min and ρ _min to select points that satisfy both (ρ>ρ _min ) and (δ>δ _min ) conditions as the distance center point.

Thunderstorms belong to small and medium-scale weather systems. Considering the scale of thunderstorms and the actual clustering effect comprehensively, ρ _min = 1.5, δ _min = 20km, which can better identify individual lightning clusters of various thunderstorms.

4. Allocation of the rest of Lightning

After the thunderstorm center lightning is determined, the category labels of the remaining lightning are assigned according to the following principle: the category label of the lightning is consistent with the category of the nearest lightning above the lightning density.

5. Determination of Thunderstorm Boundary

First define the thunderstorm boundary area: the thunderstorm boundary area of a certain thunderstorm cloud is composed of the lightning in this thunderstorm cloud and any lightning distance of other thunderstorm clouds is less than d _c . Then, look for the lightning with the highest density in the thunderstorm boundary area, and record its density as ρ _max . Finally, the lightning with ρ<ρ _max in the thunderstorm is removed as noise, so as to determine the boundary of the thunderstorm.

Therefore, the location of the center of the thunderstorm and the range of the thunderstorm cloud can be determined according to the lightning density and lightning distance. When the range of the thunderstorm cloud is determined, the number of lightning in the thunderstorm cloud can be obtained.

In the embodiment of the present invention, the clustering analysis is performed through the fast search clustering algorithm of the maximum density value, which can accurately obtain the center position of the thunderstorm, the range of the thunderstorm cloud, and the number of lightning in the thunderstorm cloud, and provide data basis for model prediction.

On the basis of the above-mentioned embodiments, said determining the central position of the thunderstorm according to the lightning density and the lightning distance includes:

The lightning whose lightning density is greater than a first preset threshold and whose lightning distance is greater than a second preset threshold is used as a thunderstorm center to obtain the thunderstorm center position.

Specifically, the lightning density values are arranged from high to low, and the confirmation of thunderstorm center lightning can be selected by given δ _min and ρ _min to select points that satisfy both (ρ>ρ _min ) and (δ>δ _min ) conditions as the distance from the center point.

Thunderstorms belong to small and medium-scale weather systems. Considering the scale of thunderstorms and the actual clustering effect comprehensively, ρ _min = 1.5, δ _min = 20km, which can better identify individual lightning clusters of various thunderstorms.

On the basis of the foregoing embodiments, the method further includes:

The thunderstorm cloud development characteristics corresponding to the current period are added to the thunderstorm cloud development sample set as a new sample to form a new thunderstorm cloud development sample set, and the lightning prediction model is trained and updated through the new thunderstorm cloud development sample set.

Specifically, after using the lightning prediction model to predict the development characteristics of thunderstorm clouds in the current period and obtain the prediction results, the development characteristics of thunderstorm clouds in the current period are used as new samples and added to the original sample set of thunderstorm cloud development to form a new thunderstorm cloud The development sample set makes the new thunderstorm cloud development sample set more abundant, and continues to train and update the lightning prediction model through the new thunderstorm cloud development sample set, so as to ensure the prediction accuracy of the lightning prediction model.

Fig. 2 is a schematic structural diagram of a lightning early warning device provided by an embodiment of the present invention. As shown in Fig. 2, the device includes: an acquisition module 201, a clustering module 201 and a prediction module 203, wherein:

The acquisition module 201 is used to acquire the spaceborne lightning monitoring data and the ground-based lightning monitoring data in the current period in the detection area; the clustering module 202 is used to perform cluster analysis according to the spaceborne lightning monitoring data and the ground-based lightning monitoring data to obtain The thunderstorm cloud development characteristics in the detection area, the thunderstorm cloud development characteristics include: thunderstorm cloud center position, thunderstorm cloud range and thunderstorm cloud lightning quantity; prediction module 203 is used for according to described thunderstorm cloud center position, described thunderstorm The characteristic evolution sequence of the thunderstorm cloud formed by the cloud range and the quantity of lightning in the thunderstorm cloud is predicted by using the lightning prediction model, so as to realize the tracking and extrapolation of the thunderstorm cloud, and obtain the lightning warning information at the next moment.

Specifically, the acquisition module 201 acquires the spaceborne lightning monitoring data and ground-based lightning monitoring data in the detection area for the current period. It should be noted that the cloud flash detection data of Fengyun No. 4 lightning detector (Lightning Mapping Imager, LMI for short) can be used and ground-based cloud-ground lightning detection data from the National Lightning Monitoring Network. The clustering module 202 performs clustering analysis according to the acquired spaceborne lightning monitoring data and ground-based lightning monitoring data in the detection area for the current period, and uses the machine learning clustering algorithm to calculate the distribution of "lightning clusters", so as to realize the distribution of thunderstorm clouds. It should be noted that the development characteristics of the thunderstorm cloud include: the central position of the thunderstorm cloud, the range of the thunderstorm cloud, and the number of lightning in the thunderstorm cloud, and may also include other information, which is not specifically limited in the embodiment of the present invention. The prediction module 203 takes the thunderstorm cloud center position, thunderstorm cloud range and lightning quantity in the thunderstorm cloud obtained by the cluster analysis as input, and inputs it into the lightning prediction model, and predicts through the lightning prediction model to realize the thunderstorm prediction. According to cloud tracking and extrapolation, thunderstorm clouds are the main source of lightning. According to the sounding results of Winn et al (1974), when the local electric field in the cloud exceeds about 400kV/m, lightning discharge can occur. Obtain the lightning warning information of the next moment. The development of thunderstorms has continuity, and the occurrence and development of thunderstorms follow the stages of convective generation, development, maturity, and extinction. Therefore, the early development of thunderstorms can provide information for the next development and movement. For the time series prediction of thunderstorm development, it is difficult for traditional neural network models to maintain the memory of thunderstorm development status. Relatively speaking, the recurrent (or recursive) neural network has a memory function, and can effectively use the previous activity device of the thunderstorm to realize the prediction of the future development of the thunderstorm. It should be noted that the lightning prediction model constructed by the recurrent (or recursive) neural network is pre-established and trained.

The embodiments of the apparatus provided by the present invention can be specifically used to execute the processing procedures of the above-mentioned method embodiments, and the functions thereof will not be repeated here, and reference can be made to the detailed description of the above-mentioned method embodiments.

The embodiments of the present invention perform cluster analysis on the spaceborne lightning monitoring data and the ground-based lightning monitoring data to obtain the development characteristics of thunderstorm clouds, and then use the lightning prediction model to predict and obtain the lightning early warning information at the next moment, and can predict the development of thunderstorms through machine learning. Forecasting with evolution improves the accuracy of lightning forecasts.

FIG. 3 is a schematic structural diagram of a lightning warning system provided by an embodiment of the present invention. As shown in FIG. 3 , the system includes: a lightning warning device 301 and a terminal 302, wherein:

The lightning warning device 301 sends lightning warning information to the terminal 302; the terminal 302 receives the lightning warning information and displays the lightning warning information.

Specifically, the lightning early warning device 301 is used to perform the functions of the above-mentioned method embodiments. The system is a visual lightning approach warning platform based on a geographic information system, implemented based on a B/S and C/S hybrid architecture, and a C/S structure program (Background program) runs in the lightning early warning device 301 and executes concurrently with multiple threads. After obtaining the spaceborne lightning monitoring data and the ground-based lightning monitoring data, after cluster analysis, input the lightning prediction model to obtain lightning early warning information. After that, the thunderstorm sequence information identified and tracked this time is normalized and added to the training sample set, and the LSTM model is trained to update the forecast model for the next use. It should be noted that the lightning warning device 301 may be a server.

The client 302, as a B/S structure program (foreground program), is a display platform for displaying the lightning warning result of the lightning warning device 301, displaying information such as thunderstorm warning position, range, and lightning activity intensity. Design a database storage system with reasonable storage and a good data interface. The design mainly includes the design and construction of table structure (data fields, data types, notes), relationships between tables, stored procedures, and application layers (such as indexes). Build a background program system, run on the server side, which is responsible for multi-threaded concurrent execution. In view of the timeliness of the data, the background program is multi-threaded and integrated into the background program system. The running status of the background program can be monitored and displayed on the foreground.

The entire system is designed and implemented based on the Python language, and the program runs on a Windows PC server. The server contains at least one NVidia GPU, which is used for training and forecasting operations of the cyclic (or recursive) neural network model. The test results show that the system can complete the generation of lightning warning products within 5 minutes after receiving the lightning data. Therefore, the algorithm can be used as a real-time forecasting tool in terms of time.

The lightning early warning system provided by the embodiment of the present invention performs clustering analysis on the spaceborne lightning monitoring data and ground-based lightning monitoring data to obtain the development characteristics of thunderstorm clouds, and then uses the lightning prediction model to make predictions to obtain the lightning early warning information at the next moment. Learning can forecast the development and evolution of thunderstorms and improve the accuracy of lightning forecasts.

FIG. 4 is a schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention. As shown in FIG. 4, the electronic device includes: a processor (processor) 401, a memory (memory) 402, and a bus 403; wherein,

The processor 401 and the memory 402 communicate with each other through the bus 403;

The processor 401 is used to call the program instructions in the memory 402 to execute the methods provided by the above-mentioned method embodiments, for example, including: acquiring spaceborne lightning monitoring data and ground-based lightning monitoring data in the current period in the detection area; According to the cluster analysis of the spaceborne lightning monitoring data and the ground-based lightning monitoring data, the development characteristics of the thunderstorm cloud in the detection area are obtained, and the development characteristics of the thunderstorm cloud include: the central position of the thunderstorm cloud, the range of the thunderstorm cloud and the The quantity of lightning in the cloud; according to the characteristic evolution sequence of the thunderstorm cloud formed by the center position of the thunderstorm cloud, the scope of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud, the lightning prediction model is used to predict, so as to realize the tracking of the thunderstorm cloud and extrapolation to obtain the lightning warning information at the next moment.

This embodiment discloses a computer program product, the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by the computer, the computer The methods provided by the above-mentioned method embodiments can be executed, for example, including: acquiring spaceborne lightning monitoring data and ground-based lightning monitoring data in the current period in the detection area; performing aggregation according to the spaceborne lightning monitoring data and the ground-based lightning monitoring data class analysis, obtain the thunderstorm cloud development characteristics in the detection area, the thunderstorm cloud development characteristics include: thunderstorm cloud center position, thunderstorm cloud range and thunderstorm cloud lightning quantity; according to the thunderstorm cloud center position, the thunderstorm cloud The characteristic evolution sequence of the thunderstorm cloud formed by the scope and the quantity of lightning in the thunderstorm cloud is predicted by the lightning prediction model, so as to realize the tracking and extrapolation of the thunderstorm cloud, and obtain the lightning warning information at the next moment.

This embodiment provides a non-transitory computer-readable storage medium, the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions cause the computer to execute the methods provided in the above method embodiments, for example, including : Obtain the spaceborne lightning monitoring data and the ground-based lightning monitoring data of the current period in the detection area; perform cluster analysis according to the space-borne lightning monitoring data and the ground-based lightning monitoring data, and obtain the development characteristics of thunderstorm clouds in the detection area , the development characteristics of the thunderstorm cloud include: the central position of the thunderstorm cloud, the scope of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud; The evolution sequence is predicted by using the lightning prediction model to realize the tracking and extrapolation of the thunderstorm cloud and obtain the lightning warning information at the next moment.

Those of ordinary skill in the art can understand that all or part of the steps for realizing the above-mentioned method embodiments can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the It includes the steps of the above method embodiments; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other various media that can store program codes.

The devices and other embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may Located in one place, or can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. It can be understood and implemented by those skilled in the art without any creative efforts.

Through the above description of the implementations, those skilled in the art can clearly understand that each implementation can be implemented by means of software plus a necessary general hardware platform, and of course also by hardware. Based on this understanding, the essence of the above technical solution or the part that contributes to the prior art can be embodied in the form of software products, and the computer software products can be stored in computer-readable storage media, such as ROM/RAM, magnetic discs, optical discs, etc., including several instructions to make a computer device (which may be a personal computer, server, or network device, etc.) execute the methods described in various embodiments or some parts of the embodiments.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims (10)

1. A lightning alerting method, comprising:

acquiring satellite-borne lightning monitoring data and foundation lightning monitoring data of a current time period in a detection area;

performing cluster analysis according to the satellite-borne lightning monitoring data and the foundation lightning monitoring data to obtain thunderstorm cloud development characteristics in the detection area, wherein the thunderstorm cloud development characteristics comprise: the central position of the thunderstorm cloud, the range of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud;

and predicting by using a lightning prediction model according to a thunderstorm cloud feature evolution sequence formed by the thunderstorm cloud center position, the thunderstorm cloud range and the lightning quantity in the thunderstorm cloud so as to realize tracking and extrapolation of the thunderstorm cloud and obtain lightning early warning information at the next moment.

2. The method of claim 1, further comprising:

and respectively carrying out normalization processing on the satellite-borne lightning monitoring data and the foundation lightning monitoring data, and fusing the normalized satellite-borne lightning monitoring data and the normalized foundation lightning monitoring data.

3. The method of claim 1, further comprising:

acquiring historical thunderstorm cloud development characteristics in a historical time period, constructing a time-series thunderstorm cloud development sample set according to the historical thunderstorm cloud development characteristics, and training a circulating or recurrent neural network model according to the thunderstorm cloud development sample set to obtain the lightning prediction model.

4. The method according to claim 1, wherein said performing a cluster analysis based on said lightning on board lightning monitoring data and said lightning on ground monitoring data comprises:

according toOrCalculating the lightning density;

wherein,n is the total number of lightning of the cluster, d_ijIs the distance between two lightning rays i and j, d_cIs a spacing threshold, d_ij＜d_cThe number of lightning of (a) is 2% of the total number of lightning;

according to delta_i＝min(d_ij),ρ_j＞ρ_iCalculating a lightning distance;

wherein, delta_iThe minimum distance between a lightning j and a lightning i is larger than the lightning density of the lightning j;

and determining the thunderstorm center position, the thunderstorm cloud range and the lightning quantity in the thunderstorm cloud according to the lightning density and the lightning distance.

5. The method of claim 4, wherein said determining a thunderstorm center location from said lightning density and said lightning distance comprises:

and taking the lightning with the lightning density larger than a first preset threshold value and the lightning distance larger than a second preset threshold value as a thunderstorm center, and obtaining the thunderstorm center position.

6. The method of claim 3, further comprising:

and adding the thunderstorm cloud development characteristics corresponding to the current time period as new samples into the thunderstorm cloud development sample set to form a new thunderstorm cloud development sample set, and training and updating the lightning prediction model through the new thunderstorm cloud development sample set.

7. A lightning early warning device, comprising:

the acquisition module is used for acquiring satellite-borne lightning monitoring data and foundation lightning monitoring data of the current time period in the detection area;

a clustering module, configured to perform clustering analysis according to the satellite-borne lightning monitoring data and the foundation lightning monitoring data to obtain thunderstorm cloud development characteristics in the detection area, where the thunderstorm cloud development characteristics include: the central position of the thunderstorm cloud, the range of the thunderstorm cloud and the quantity of lightning in the thunderstorm cloud;

and the prediction module is used for predicting by using a lightning prediction model according to a thunderstorm cloud feature evolution sequence formed by the thunderstorm cloud center position, the thunderstorm cloud range and the lightning quantity in the thunderstorm cloud so as to realize tracking and extrapolation of the thunderstorm cloud and obtain lightning early warning information at the next moment.

8. A lightning alerting system comprising the lightning alerting device of claim 7 and a terminal; wherein,

the lightning early warning device sends lightning early warning information to the terminal;

and the terminal receives the lightning early warning information and displays the lightning early warning information.

9. An electronic device, comprising: a processor, a memory, and a bus, wherein,

the processor and the memory are communicated with each other through the bus;

the memory stores program instructions executable by the processor, the processor invoking the program instructions to perform the method of any of claims 1-6.

10. A non-transitory computer-readable storage medium storing computer instructions that cause a computer to perform the method of any one of claims 1-6.