Lightning interceptor for lightning-struck area of forest and lightning current amplitude prediction method

Abstract

The present application relates to the technical field of lightning protection, in particular to a lightning interceptor for forest lightning stroke area and a lightning current amplitude prediction method. The sensor built-in in the lightning interceptor collects various historical time series data of each monitoring point in each lightning stroke event; based on the fluctuation and difference characteristics of the historical electric field time series data, the field strength weight of each monitoring point is dynamically calculated, and the prominent point position significantly affected by lightning stroke is screened out; further combining the historical electric field and historical meteorological data to generate the predicted electric field time series data of the monitoring point, by analyzing the similarity, spatial position and field strength weight of the historical electric field of the prominent point position, the electric field similarity between the monitoring point and each prominent point position is obtained. Finally, based on the historical meteorological, electric field and lightning current data of the prominent point position, the network model is trained, the data contribution degree is adjusted by using the electric field similarity, the influence of different prominent point positions is learned by the model, and the stability and precision of the lightning current prediction are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of lightning protection, in particular to a lightning interceptor for a forest lightning stroke area and a lightning current amplitude prediction method. BACKGROUND

[0002] With the change of climate in recent years, the forests in the north and south often have strong convective weather or dry weather, which can easily cause thunderstorms, leading to lightning striking trees and causing forest fires. In order to avoid the occurrence of these forest fires, lightning interceptors such as lightning protection belts are often installed in areas with high incidence of thunderstorms. The lightning interceptors break through the limitation of traditional lightning protection facilities that can only "passively wait" for lightning strikes. The lightning interceptors deploy high-conductivity lightning rod arrays and use the "upward leader" technology to actively release ion flow opposite in polarity to the charge of the thundercloud, forming a discharge channel between the cloud and the ground, thereby guiding lightning to strike the interceptor and avoiding direct strikes on trees. The lightning current amplitude is a core parameter for measuring the damage of lightning, and accurate prediction of the lightning current amplitude is of great significance for forest lightning protection, disaster reduction, and ecological protection.

[0003] In the prior art, when predicting the lightning current amplitude, the lightning conditions of monitoring points with similar thunderstorm characteristics and lightning events in the forest lightning stroke area are usually used as references to train a network model for predicting the lightning current amplitude that may occur in the future period of the monitoring position. However, in training the network model, the lightning conditions of all the referenced monitoring points are usually assigned the same attention weight, and this static weight allocation method often fails to focus on key influencing points, resulting in inaccurate final prediction results. SUMMARY

[0004] In order to solve the technical problem that in training the network model, the lightning conditions of all the referenced monitoring points are usually assigned the same attention weight, and this static weight allocation method often fails to focus on key influencing points, resulting in inaccurate final prediction results, the purpose of the present application is to provide a lightning interceptor for a forest lightning stroke area and a lightning current amplitude prediction method, and the technical solution adopted is as follows:

[0005] A lightning current amplitude prediction method for a forest lightning stroke area, comprising:

[0006] In the forest lightning stroke area to be measured, historical meteorological time series data, historical lightning current time series data, and historical electric field time series data of each monitoring point in each lightning stroke event are obtained;

[0007] In each lightning stroke event, the fluctuation characteristics and difference characteristics of the historical electric field time series data between the monitoring points are analyzed to determine the field strength weight of each monitoring point for screening out prominent points in each lightning stroke event;

[0008] Optionally, one monitoring point is taken as a to-be-monitored point, and predicted electric field time series data at the to-be-monitored point are obtained based on historical electric field time series data and historical meteorological time series data of the to-be-monitored point; similarity analysis is performed on the predicted electric field time series data of the to-be-monitored point and historical electric field time series data of a highlight point in each lightning stroke event, and the similarity analysis is combined with position distribution characteristics and a field strength weight of the highlight point to determine electric field similarity between the to-be-monitored point and each highlight point.

[0009] A network model is trained based on the electric field similarity between the to-be-monitored point and the highlight point, meteorological time series data of the highlight point, historical electric field time series data, and historical lightning current time series data, and is used for lightning current prediction of the to-be-monitored point.

[0010] Further, the field strength weight acquisition method comprises:

[0011] In each lightning stroke event, a value obtained by performing negative correlation mapping on a standard deviation of historical electric field strength values of each monitoring point is taken as an electric field strength stability factor of each monitoring point.

[0012] In each lightning stroke event, difference characteristics of historical electric field strength values of all monitoring points at the same time are analyzed to determine an electric field strength highlight factor of each monitoring point at each time.

[0013] In each lightning stroke event, a value obtained by normalizing a product of the electric field strength highlight factor and the electric field strength stability factor of each monitoring point at all times is taken as a field strength weight of each monitoring point in each lightning stroke event.

[0014] Further, the electric field strength highlight factor acquisition method comprises:

[0015] In each lightning stroke event, a mean value of historical electric field strength values of all monitoring points at the same time is taken as an electric field mean value characteristic value, and a value obtained by normalizing a difference between a historical electric field strength value of each monitoring point at each time and the corresponding electric field mean value characteristic value at each time is taken as an electric field strength highlight factor of each monitoring point at each time.

[0016] Further, the highlight point acquisition method comprises:

[0017] In each lightning stroke event, a monitoring point with a field strength weight greater than a preset field strength highlight threshold is taken as a highlight point in each historical lightning stroke event.

[0018] Further, the predicted electric field time series data acquisition method comprises:

[0019] In all lightning stroke events, the historical electric field time series data and the historical meteorological time series data of the to-be-monitored point are taken as the input of the SARIMA network model, so as to determine the model parameters;

[0020] Based on the historical electric field time series data of the to-be-monitored point in all lightning stroke events and the model parameters, the SARIMA network model is trained to obtain a SARIMA model for predicting the lightning electric field;

[0021] The current electric field time series data of the to-be-monitored point is obtained, and the current electric field time series data is taken as the input of the SARIMA model for predicting the lightning electric field, so as to obtain the predicted electric field time series data of the to-be-monitored point.

[0022] Further, the electric field similarity acquisition method comprises:

[0023] The predicted electric field time series data of the to-be-monitored point and the historical electric field time series data of the prominent point in each lightning stroke event are subjected to similarity analysis, so as to determine the electric field similarity parameters between the to-be-monitored point and each prominent point in each lightning stroke event;

[0024] The positional relationship between the to-be-monitored point and the prominent point in each lightning stroke event is analyzed, so as to determine the positional similarity parameters between the to-be-monitored point and each prominent point in each lightning stroke event;

[0025] The product of the electric field similarity parameters and the positional similarity parameters between the to-be-monitored point and each prominent point in each lightning stroke event is subjected to normalization, and the normalized value is taken as the electric field similarity factor between the to-be-monitored point and each prominent point in each lightning stroke event;

[0026] In each lightning stroke event, the product of the field intensity weight of each prominent point and the electric field similarity factor between each prominent point and the to-be-monitored point is subjected to normalization, and the normalized value is taken as the electric field similarity between the to-be-monitored point and each prominent point in each lightning stroke event.

[0027] Further, the electric field similarity parameter acquisition method comprises:

[0028] The DTW value of the predicted electric field time series data of the to-be-monitored point and the historical electric field time series data of each prominent point in each lightning stroke event is subjected to negative correlation mapping and normalization, and the normalized value is taken as the electric field similarity parameter between the to-be-monitored point and each prominent point in each lightning stroke event.

[0029] Further, the positional similarity parameter acquisition method comprises:

[0030] The negative correlation mapping value of the Euclidean distance between the to-be-monitored point and each prominent point in each lightning stroke event is taken as the positional similarity parameter.

[0031] Further, the network model is trained based on the electric field similarity between the to-be-monitored point and the highlight point, the weather time series data of the highlight point, the historical electric field time series data and the historical lightning current time series data, and is used for lightning current prediction of the to-be-monitored point, comprising:

[0032] The historical weather time series data, the historical electric field time series data of each highlight point corresponding to the to-be-monitored point in each lightning stroke event are taken as inputs, the electric field similarity between the to-be-monitored point and each highlight point corresponding to each lightning stroke event is taken as an attention weight, and the historical lightning current time series data of each highlight point corresponding to the to-be-monitored point in each lightning stroke event is taken as an output, so as to train the ConvLSTM network model and obtain the trained network model.

[0033] The current weather time series data and the current electric field time series data of the to-be-monitored point in a current period are obtained and taken as inputs of the trained network model, so that the lightning current time series data of the to-be-monitored point in a future period can be output.

[0034] A lightning interceptor for a forest lightning stroke area comprises a lightning interceptor body, and a lightning current prediction module is built in the lightning interceptor body, which is used to realize the steps in a lightning current amplitude prediction method for a forest lightning stroke area.

[0035] The present application has the following beneficial effects:

[0036] In the forest lightning strike area to be measured, the historical meteorological time series data, the historical lightning current time series data and the historical electric field time series data of each monitoring point in each lightning strike event are obtained, covering the multi-dimensional data set of the whole chain of lightning strike, providing more comprehensive data support for subsequent model training. Since lightning is discharged through the leader channel, the electric field intensity at the leader electrification site is more indicative of the lightning intensity during discharge lightning than other monitoring sites. Therefore, in the subsequent calculation and analysis of the similarity between monitoring sites, these leader electrification sites should be more referential for subsequent lightning prediction. Therefore, in each lightning strike event, based on the fluctuation and difference characteristics of the historical electric field time series data, the field intensity weight of each monitoring site is dynamically calculated, and the prominent sites significantly affected by lightning strike events are screened out, so that the model can focus on the key areas with true thunderstorm prominent features. Further, by combining the historical electric field and meteorological time series data to generate predicted electric field time series data of the monitoring site to be monitored, the change of the electric field before lightning strike is simulated in advance, and then the similarity, spatial position distribution characteristics of the predicted electric field time series data and the historical electric field of the prominent site, and the field intensity weight of the prominent site are analyzed, a time and space comprehensive similarity evaluation system is constructed, and the electric field similarity between the monitoring site to be monitored and each prominent site is obtained. This index provides an adaptive contribution weight for different prominent sites. Finally, the network model is trained based on the meteorological, electric field and lightning current data of the prominent sites, and the electric field similarity between the monitoring site to be monitored and the prominent sites is used to adjust the data contribution, so that the network model can adaptively learn the influence intensity of different prominent sites on the monitoring site to be monitored, and predict the lightning current of the monitoring site to be monitored, effectively improving the stability and accuracy of the prediction result. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, and the advantages thereof, a brief introduction will be given to the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0038] Figure 1 A module structure diagram of a lightning current prediction module in a lightning current interceptor for a forest lightning strike area provided by an embodiment of the present application;

[0039] Figure 2 A method flowchart of a lightning current amplitude prediction method for a forest lightning strike area provided by an embodiment of the present application;

[0040] Figure 3 A method flowchart of a prominent site acquisition method provided by an embodiment of the present application;

[0041] Figure 4 A method flowchart of an electric field similarity acquisition method provided by an embodiment of the present application. DETAILED DESCRIPTION

[0042] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined object of the present application, the following describes in detail the specific embodiments, structures, features and effects of a lightning interceptor for a lightning-struck forest area and a lightning current amplitude prediction method according to the present application, with reference to the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0044] The following describes in detail the specific scheme of a lightning interceptor for a lightning-struck forest area and a lightning current amplitude prediction method provided by the present application, with reference to the accompanying drawings.

[0045] With the intensification of global climate change, extreme weather events occur frequently, and the forest ecosystems in the north and south are facing increasingly severe lightning fire threats. The frequent occurrence of strong convective weather (such as thunderstorm clouds) leads to a significant increase in lightning activities in southern forests, while the northern forests are more prone to large-scale fires due to drought, dry vegetation and thunderstorms. In order to avoid the occurrence of these forest fires, lightning interceptors are often installed in areas with high thunderstorm frequency. The core principle of lightning interceptors is to actively intervene in the lightning discharge process, build a safe channel between the cloud and the ground, and guide the lightning current to the preset path to be discharged into the ground, thereby guiding the lightning to hit the interceptor and avoiding direct lightning hitting the trees.

[0046] The lightning interceptor deploys an array of highly conductive lightning rods (combination of vertical and horizontal needles), uses the sharp end discharge effect to release ion flow opposite in polarity to the thundercloud charge, and forms an "upward leader". When the intensity of the thundercloud reaches a threshold value, the upward leader couples with the thundercloud leader, actively intercepts the lightning discharge channel, and makes the lightning preferentially hit the interceptor rather than the protected target. Therefore, during the operation of the lightning interceptor, the lightning current amplitude is a core parameter for measuring the lightning damage, and accurate prediction of the lightning current amplitude is of great significance for forest lightning protection and disaster reduction, ecological protection, etc.

[0047] The embodiment of the present application provides a lightning interceptor for a lightning-struck forest area, which comprises a lightning interceptor body, and a lightning current prediction module is built in the lightning interceptor, please refer to Figure 1Fig. 1 is a schematic diagram of a module structure of a lightning current prediction module according to an embodiment of the present application, which shows a module structure of a lightning current prediction module provided by an embodiment of the present application, including a processor 100, a memory 101, a bus 102, and a communication interface 103, wherein the processor 100, the communication interface 103, and the memory 101 are connected through the bus 102; the memory 101 can include a high-speed random access memory, the bus 102 can be an ISA bus, a PCI bus, or an EISA bus, etc., the processor 100 can be an integrated circuit chip with signal processing capability; the memory 101 stores at least one instruction, at least one program, a code set, or an instruction set, which are loaded and executed by the processor to implement a step in a lightning current amplitude prediction method for a forest lightning stroke area.

[0048] Referring to Fig. 2, Figure 2 Fig. 2 is a flow chart of a lightning current amplitude prediction method for a forest lightning stroke area according to an embodiment of the present application, which shows a flow chart of a lightning current amplitude prediction method for a forest lightning stroke area provided by an embodiment of the present application, and the method includes the following steps:

[0049] Step S1: In a forest lightning stroke area to be measured, historical meteorological time series data, historical lightning current time series data, and historical electric field time series data of each monitoring point in each lightning stroke event are obtained.

[0050] First, in the lightning stroke area to be measured, the position of each lightning interceptor is taken as a monitoring point, and in the period of each lightning stroke event in the past historical period, the historical meteorological time series data (in the embodiment of the present application, the types of meteorological time series data are set as basic meteorological data such as temperature, humidity, and wind speed) of each monitoring point are obtained through a plurality of meteorological monitoring sensors built in the lightning interceptor, the historical lightning current time series data are obtained through a current meter built in the lightning interceptor, and the historical electric field time series data are obtained through an electric field meter built in the lightning interceptor. Among them, various time series data are collected synchronously and have the same collection frequency.

[0051] It should be noted that in the embodiment of the present application, the historical period is set as the past one year, and the collection frequency of various time series data is set as 10 times per second, and the specific values can be adjusted according to the implementation scene, which is not limited herein.

[0052] Step S2: In each lightning stroke event, the fluctuation characteristics and difference characteristics of the historical electric field time series data between the monitoring points are analyzed to determine the field intensity weight of each monitoring point for screening out the prominent points in each lightning stroke event.

[0053] When a lightning stroke occurs, a large amount of negative charge is accumulated on the side close to the ground of a thundercloud, and a large amount of positive charge is generated on the side close to the thundercloud on the trees and lightning protection measures on the ground under the action of electromagnetic induction, so that an electric field is formed at the lower part of the thundercloud and the top of the trees and lightning protection measures. Since the conductivity of the trees is poorer than that of the lightning protection measures, the thundercloud will preferentially strike the lightning protection measures. Since lightning discharge is carried out through a leader channel, the electric field intensity at the leader channel is more indicative of the lightning intensity during lightning discharge than other monitoring points, and the leader channel is more referential because it is formed by the charges of the thundercloud and the trees or lightning intercepters on the ground, which results in a stronger electric field than the electric field formed by the thundercloud and the ground in other regions, and the corresponding electric field gradually increases with the increase of the charge at the top of the trees or lightning intercepters.

[0054] Therefore, in this step, the fluctuation characteristics and difference characteristics of the historical electric field time series data between the monitoring points can be compared in each lightning stroke event, the field intensity weight of each monitoring point is determined, and the prominent point in each lightning stroke event is selected based on the field intensity weight. The prominent point is a monitoring point that has a greater response intensity to the lightning stroke event and is directly affected by the lightning stroke in each lightning stroke event.

[0055] Preferably, the method for obtaining the field prominent point in an embodiment of the present application comprises:

[0056] Please refer to Figure 3 which shows a method flowchart of the method for obtaining the prominent point in an embodiment of the present application. The method comprises the following steps:

[0057] Step S201: In each lightning stroke event, the fluctuation characteristics of the historical electric field time series data of each monitoring point are analyzed, and the electric field intensity stability factor of each monitoring point is determined.

[0058] In each lightning stroke event, the standard deviation of the historical electric field intensity value of each monitoring point is calculated. The greater the standard deviation, the greater the fluctuation degree of the historical electric field intensity value of the monitoring point, and the smaller the stability. Therefore, the value obtained by negatively correlating the standard deviation is used as the electric field intensity stability factor of each monitoring point. The greater the electric field intensity stability factor, the higher the stability of the electric field at the monitoring point. The negative correlation mapping can be performed by the formula wherein, represents an exponential function with the natural constant e as the base, and x represents the independent variable.

[0059] Step S202: In each lightning stroke event, the difference characteristics of the historical electric field intensity values of all monitoring points at the same time are analyzed, and the electric field intensity prominent factor of each monitoring point at each time is determined.

[0060] In each lightning stroke event, the average value of the historical electric field intensity values of all monitoring points at the same time is taken as the electric field average characteristic value, which represents the average level of the electric field intensity of all monitoring points at the same time, and provides a reference benchmark.

[0061] Then, the difference between the historical electric field intensity value of each monitoring point at each time and the corresponding electric field average characteristic value at each time is calculated. If the difference is positive and larger, it means that the historical electric field intensity of the monitoring point at the time is higher than the average level of all monitoring points at the same time, so the corresponding electric field will be stronger and more likely to be the leading power supply, and thus the reference will be higher. Therefore, the difference is normalized to obtain the electric field intensity highlight factor of each monitoring point at each time. Since the difference here can be positive or negative, the normalization method adopts function.

[0062] Step S203: In each lightning stroke event, the electric field intensity highlight factor and the electric field intensity stability factor corresponding to each monitoring point are fused to determine the field strength weight of each monitoring point.

[0063] Based on the analysis in the foregoing steps S201 and S202, the greater the electric field intensity stability factor of the monitoring point, the higher the electric field stability at the monitoring point, and the greater the electric field intensity highlight factor, the stronger the corresponding electric field, and the more likely it is to be the leading power supply, and thus the reference will be higher. Therefore, the two factors are positively correlated with the field strength weight of the monitoring point, so in each lightning stroke event, the product of the electric field intensity highlight factor and the electric field intensity stability factor of each monitoring point at all times is normalized to obtain the field strength weight of each monitoring point in each lightning stroke event. At this time, the greater the field strength weight, the greater the field strength highlight degree of the monitoring point in the lightning stroke event, the higher the response degree to the lightning stroke event, and the greater the reference value for the lightning current amplitude prediction of the subsequent process. The normalization is a technology known to those skilled in the art, and the selection of the normalization function can be linear normalization or standard normalization, and the specific normalization method is not limited here.

[0064] Step S204: In each lightning stroke event, the prominent point is selected from all monitoring points according to the field strength weight of the monitoring point.

[0065] In each lightning stroke event, the monitoring point with a field strength weight greater than a preset field strength highlight threshold is taken as the prominent point in each historical lightning stroke event, and the selected prominent point is a monitoring point significantly affected by the lightning stroke event, i.e., a key point with a thunderstorm highlight feature.

[0066] It should be noted that the preset field strength highlight threshold in the embodiment of the present application is 0.6, and the specific value can be adjusted according to the implementation scene, which is not limited herein.

[0067] Step S3: optionally selecting one monitoring point as a to-be-monitored point, obtaining predicted electric field time series data at the to-be-monitored point based on historical electric field time series data and historical meteorological time series data of the to-be-monitored point; performing similarity analysis on the predicted electric field time series data of the to-be-monitored point and historical electric field time series data of the highlight point in each lightning stroke event, combining with the position distribution characteristics and the field strength weight of the highlight point, and determining the electric field similarity between the to-be-monitored point and each highlight point.

[0068] When lightning strikes, a large amount of negative charge is gathered in the lower part of the thundercloud, and then the trees and lightning intercepters on the ground gather a large amount of positive charge at their top ends under the action of the electric field caused by the thundercloud. The positive charge at the top ends of the trees and lightning intercepters and the negative charge in the lower part of the thundercloud form an electric field, and when the intensity of the electric field reaches a certain level, a leader channel is formed, thereby generating lightning current when discharging lightning. Therefore, when the electric field intensity between two monitoring points is similar, the lightning current amplitude when discharging lightning will also be basically similar, so it is necessary to analyze the similarity of the electric field intensity between the monitoring points in this step.

[0069] In this step, first, one monitoring point is selected as a to-be-monitored point, and the electric field intensity of the to-be-monitored point in a future period of time is predicted based on historical electric field time series data and historical meteorological time series data of each monitoring point, thereby obtaining predicted electric field time series data of the to-be-monitored point.

[0070] Preferably, the method for obtaining the predicted electric field time series data in one embodiment of the present application comprises:

[0071] For thunderstorms, the speed of various conditions in the thundercloud to accumulate electric charge and form a leader channel is roughly the same under the influence of the meteorological environment, so the electric field situation between the thundercloud and the ground in the past numerous lightning stroke events and the meteorological characteristics can be used to preliminarily predict the lightning electric field situation in a future period of time through a seasonal autoregressive moving average model SARIMA model.

[0072] In all lightning stroke events, the historical electric field time series data and the historical meteorological time series data of the to-be-monitored point are used as inputs of the SARIMA network model, so as to determine the model parameters, and then the SARIMA network model is trained based on the historical electric field time series data of the to-be-monitored point in all lightning stroke events and the model parameters, thereby obtaining the SARIMA model for predicting lightning electric field; finally, the current electric field time series data of the to-be-monitored point is obtained, and the current electric field time series data is used as the input of the SARIMA model for predicting lightning electric field, thereby obtaining the predicted electric field time series data at the to-be-monitored point.

[0073] It should be noted that the SARIMA network model is a known technology, and the specific process is not described here; the length of the predicted electric field time series data in this embodiment of the application is set to 1 hour, and the specific length can be adjusted according to the implementation scene, which is not limited here.

[0074] The predicted electric field time series data of the to-be-monitored point realizes the simulation of the change of the electric field before lightning, so after obtaining the predicted electric field time series data of the to-be-monitored point, the similarity of the predicted electric field time series data of the to-be-monitored point and the historical electric field time series data of the prominent point in each lightning event can be analyzed, and the position distribution characteristics between the monitoring points and the field intensity weight of the prominent point are combined, so as to obtain the electric field similarity between the to-be-monitored point and each prominent point. The electric field similarity represents the similarity between the electric field characteristics of the to-be-monitored point in a future period of time and the electric field characteristics of the monitoring point significantly affected by the lightning event, so as to provide an adaptive contribution weight for each prominent point in the subsequent process.

[0075] Preferably, in an embodiment of the application, the method for obtaining the electric field similarity comprises:

[0076] Please refer to Figure 4 , which shows the method flowchart of the method for obtaining the electric field similarity in an embodiment of the application, and the method comprises the following steps:

[0077] Step S301: Perform similarity analysis on the predicted electric field time series data of the to-be-monitored point and the historical electric field time series data of the prominent point in each lightning event, and determine the electric field similarity parameter between the to-be-monitored point and each prominent point in each lightning event.

[0078] The DTW value can capture the shape matching degree of two time series data on the time axis, and can be used to reflect the similarity, so the DTW value of the predicted electric field time series data at the to-be-monitored point and the historical electric field time series data of each prominent point in each lightning event is calculated. The smaller the DTW value, the higher the similarity between the two time series data, so the DTW value is negatively correlated and normalized to correct the logical relationship, so as to obtain the electric field similarity parameter between the to-be-monitored point and each prominent point in each lightning event. The greater the electric field similarity parameter here, the higher the electric field similarity between the to-be-monitored point and each prominent point in each lightning event. The negative correlation and normalization processing here can use the formula , wherein, represents the exponential function with the natural constant e as the base, and x represents the independent variable.

[0079] It should be noted that the method for obtaining the DTW value is a known technology, and the specific process is not described here.

[0080] Step S302: Analyze the positional relationship between the to-be-monitored point and the prominent point in each lightning stroke event, and determine the position similarity parameter between the to-be-monitored point and each prominent point in each lightning stroke event.

[0081] In a lightning stroke event, the electric field change is affected by local meteorological conditions, and the electric field mode of the adjacent monitoring point can be more similar, so the Euclidean distance between the to-be-monitored point and each prominent point in each lightning stroke event is calculated. The smaller the Euclidean distance, the closer the distance between the two points, and the more likely it is to have a similar electric field mode. Therefore, the value after the negative correlation mapping of the Euclidean distance is taken as the position similarity parameter. Based on the foregoing logic, the greater the position similarity parameter, the closer the positions, and the more likely it is to have consistent electric field characteristics. Similarly, the negative correlation mapping method here can use the formula wherein, represents an exponential function with a natural constant e as the base, and x represents the independent variable.

[0082] Step S303: Determine the electric field similarity factor between the to-be-monitored point and each prominent point in each lightning stroke event according to the electric field similarity parameter and the position similarity parameter between the to-be-monitored point and each prominent point in each lightning stroke event.

[0083] Based on the analysis in steps S301 and S302, the electric field similarity parameter and the position similarity parameter between the to-be-monitored point and each prominent point in each lightning stroke event are positively correlated with the electric field similarity. Therefore, in this embodiment of the present application, the product of the electric field similarity parameter and the position similarity parameter between the to-be-monitored point and each prominent point in each lightning stroke event is normalized, and the value is taken as the electric field similarity factor between the to-be-monitored point and each prominent point in each lightning stroke event. The greater the electric field similarity factor obtained by fusing the similar characteristics of the electric field intensity and the similar characteristics of the spatial position, the higher the electric field similarity characteristics between the two points. The normalization is a well-known technical means to those skilled in the art, and the selection of the normalization function can be linear normalization or standard normalization, and the specific normalization method is not limited herein.

[0084] Step S304: Fuse the electric field similarity factor between the to-be-monitored point and each prominent point in each lightning stroke event with the field intensity weight of each prominent point, to obtain the electric field similarity between the to-be-monitored point and each prominent point in each lightning stroke event.

[0085] Based on the record in step S203, in a lightning stroke event, the greater the field strength weight of the prominent point, the greater the field strength prominence of the prominent point in the lightning stroke event, the higher the response degree to the lightning stroke event, and the greater the reference value of the lightning current amplitude prediction to the subsequent process, so in this step, in each lightning stroke event, the product of the field strength weight of each prominent point and the electric field similarity factor between each prominent point and the to-be-monitored point is normalized as the electric field similarity between the to-be-monitored point and each prominent point in each lightning stroke event. The electric field similarity can provide an adaptive contribution weight for the prominent point in each lightning stroke event. The normalization is a technical means familiar to those skilled in the art, and the selection of the normalization function can be linear normalization or standard normalization, and the specific normalization method is not limited here.

[0086] S4: training a network model based on the electric field similarity between the to-be-monitored point and the prominent point, the meteorological time series data of the prominent point, the historical electric field time series data, and the historical lightning current time series data, for lightning current prediction of the to-be-monitored point.

[0087] Since the prominent point in each lightning stroke event is a monitoring point significantly affected by the lightning stroke event, that is, a key point with a thunderstorm prominent feature, in this step, a network model can be trained based on the meteorological time series data of the prominent point, the historical electric field time series data, and the historical lightning current time series data. In view of the electric field similarity between the to-be-monitored point and each prominent point in each lightning stroke event obtained in the above step, a contribution weight is provided for each prominent point, so the electric field similarity is involved in the training process of the network model, thereby obtaining a trained network model for lightning current prediction of the to-be-monitored point.

[0088] Preferably, in an embodiment of the present application, a network model is trained based on the electric field similarity between the to-be-monitored point and the prominent point, the meteorological time series data of the prominent point, the historical electric field time series data, and the historical lightning current time series data, for lightning current prediction of the to-be-monitored point, comprising:

[0089] Setting data input channels: including meteorological parameter channels and electric field intensity channels, then inputting the historical meteorological time series data, the historical electric field time series data of the to-be-monitored point corresponding to each prominent point in each lightning stroke event, taking the electric field similarity between the to-be-monitored point and each prominent point in each lightning stroke event as the attention weight, and taking the historical lightning current time series data of the to-be-monitored point corresponding to each prominent point in each lightning stroke event as the output, thereby guiding the model to learn the mapping relationship from the input to the output flow, for training the ConvLSTM network model, to obtain a trained network model.

[0090] Finally, the current weather time series data and the current electric field time series data (obtained based on various sensors built in the lightning interceptor at the to-be-monitored point) in the current period of the to-be-monitored point are obtained and used as the input of the trained network model, so that the lightning current time series data of the to-be-monitored point in the future period can be output.

[0091] It should be noted that the ConvLSTM network model is a known technology, and the training process is also a known technology, so the specific process is not described here.

[0092] In summary, in the to-be-measured forest lightning stroke area, the historical weather time series data, the historical lightning current time series data and the historical electric field time series data of each monitoring point in each lightning stroke event are obtained, covering a multi-dimensional data set of the whole lightning stroke chain, providing more comprehensive data support for subsequent model training. Since lightning discharge is carried out through a leader channel, the electric field strength at the leader conduction point can better indicate the lightning intensity during discharge lightning than other monitoring points. Therefore, when calculating and analyzing the similarity between monitoring points, these leader conduction points should be more referential for subsequent lightning prediction. Therefore, in each lightning stroke event, the field strength weight of each monitoring point is dynamically calculated based on the fluctuation and difference characteristics of the historical electric field time series data, and the prominent points significantly affected by lightning stroke events are selected, so that the model can focus on the key areas with true thunderstorm prominent features. Further, the predicted electric field time series data of the to-be-monitored point is generated by combining the historical electric field and weather time series data, the change of the electric field before lightning is simulated in advance, and then the similarity, spatial position distribution characteristics of the predicted electric field time series data and the historical electric field of the prominent point, and the field strength weight of the prominent point are analyzed, a time and space comprehensive similarity evaluation system is constructed, and the electric field similarity between the to-be-monitored point and each prominent point is obtained. This index provides an adaptive contribution weight for different prominent points. Finally, the network model is trained based on the weather, electric field and lightning current data of the prominent points, and the data contribution is adjusted based on the electric field similarity between the to-be-monitored point and the prominent points, so that the network model can adaptively learn the influence intensity of different prominent points on the to-be-monitored point, and predict the lightning current of the to-be-monitored point, effectively improving the stability and accuracy of the prediction result.

[0093] It should be noted that the above-mentioned embodiment sequence of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous.

[0094] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the differences from other embodiments.

Claims

1. A lightning current amplitude prediction method for a forest lightning stroke area, characterized by, The method comprises: In the forest lightning stroke area to be tested, historical meteorological time series data, historical lightning current time series data and historical electric field time series data of each monitoring point in each lightning stroke event are acquired; In each lightning stroke event, fluctuation characteristics and difference characteristics of the historical electric field time series data between monitoring points are analyzed to determine the field strength weight of each monitoring point for screening out prominent points in each lightning stroke event; An optional monitoring point is taken as a monitoring point to be monitored, and the predicted electric field time series data of the monitoring point to be monitored is acquired based on the historical electric field time series data and the historical meteorological time series data of the monitoring point to be monitored; similarity analysis is performed on the predicted electric field time series data of the monitoring point to be monitored and the historical electric field time series data of the prominent points in each lightning stroke event, and the position distribution characteristics and the field strength weight of the prominent points are combined to determine the electric field similarity between the monitoring point to be monitored and each prominent point; A network model is trained based on the electric field similarity between the monitoring point to be monitored and the prominent points, the meteorological time series data, the historical electric field time series data and the historical lightning current time series data of the prominent points, and is used for lightning current prediction of the monitoring point to be monitored; The method for acquiring the electric field similarity comprises: Similarity analysis is performed on the predicted electric field time series data of the monitoring point to be monitored and the historical electric field time series data of the prominent points in each lightning stroke event to determine the electric field similarity parameter between the monitoring point to be monitored and each prominent point in each lightning stroke event; The position relationship between the monitoring point to be monitored and the prominent points in each lightning stroke event is analyzed to determine the position similarity parameter between the monitoring point to be monitored and each prominent point in each lightning stroke event; The product of the electric field similarity parameter and the position similarity parameter between the monitoring point to be monitored and each prominent point in each lightning stroke event is normalized to obtain the electric field similarity factor between the monitoring point to be monitored and each prominent point in each lightning stroke event; In each lightning stroke event, the product of the field strength weight of each prominent point and the electric field similarity factor between each prominent point and the monitoring point to be monitored is normalized to obtain the electric field similarity between the monitoring point to be monitored and each prominent point in each lightning stroke event.

2. The lightning current amplitude prediction method for a lightning stroke area of a forest according to claim 1, wherein The method for acquiring the field strength weight comprises: In each lightning stroke event, the standard deviation of the historical electric field intensity value of each monitoring point is negatively correlated to obtain the electric field intensity stability factor of each monitoring point; In each lightning stroke event, the difference characteristics of the historical electric field intensity values of all monitoring points at the same time are analyzed to determine the electric field intensity prominent factor of each monitoring point at each time; In each lightning stroke event, the product of the electric field intensity prominent factor and the electric field intensity stability factor of each monitoring point at all times is normalized to obtain the field strength weight of each monitoring point in each lightning stroke event.

3. The lightning current amplitude prediction method for a lightning stroke area of a forest according to claim 2, wherein The method for acquiring the electric field intensity prominent factor comprises: In each lightning stroke event, the mean value of the historical electric field intensity values of all monitoring points at the same time is taken as the electric field mean value characteristic value, and the normalized value of the difference between the historical electric field intensity value of each monitoring point at each time and the corresponding electric field mean value characteristic value at each time is taken as the electric field intensity highlight factor of each monitoring point at each time.

4. The lightning current amplitude prediction method for a lightning stroke area of a forest according to claim 1, wherein The highlight point acquisition method comprises: In each lightning stroke event, the monitoring point with a field intensity greater than a preset field intensity highlight threshold is taken as the highlight point in each historical lightning stroke event.

5. The lightning current amplitude prediction method for a lightning stroke area of a forest according to claim 1, wherein The prediction electric field time series data acquisition method comprises: In all lightning stroke events, the historical electric field time series data and historical meteorological time series data of the to-be-monitored point are taken as the input of the SARIMA network model, so as to determine the model parameters; Based on the historical electric field time series data of the to-be-monitored point in all lightning stroke events and the model parameters, a SARIMA network model is trained to obtain a SARIMA model for predicting lightning electric field; The current electric field time series data of the to-be-measured monitoring point is obtained, and the current electric field time series data is taken as the input of the SARIMA model for predicting lightning electric field, so as to obtain the prediction electric field time series data of the to-be-measured monitoring point.

6. The lightning current amplitude prediction method for a lightning stroke area of a forest according to claim 1, wherein The electric field similarity parameter acquisition method comprises: The negative correlation mapping and normalized value of the DTW value between the prediction electric field time series data of the to-be-measured monitoring point and the historical electric field time series data of each highlight point in each lightning stroke event is taken as the electric field similarity parameter between the to-be-monitored point and each highlight point in each lightning stroke event.

7. The lightning current amplitude prediction method for a lightning stroke area of a forest according to claim 1, wherein The position similarity parameter acquisition method comprises: The negative correlation mapping value of the Euclidean distance between the to-be-monitored point and the position of each highlight point in each lightning stroke event is taken as the position similarity parameter.

8. The lightning current amplitude prediction method for a lightning stroke area of a forest according to claim 1, wherein The network model is trained based on the electric field similarity between the to-be-monitored point and the highlight point, the meteorological time series data, the historical electric field time series data and the historical lightning current time series data of the highlight point, and is used for lightning current prediction of the to-be-monitored point, comprising: The historical meteorological time series data and historical electric field time series data of each highlight point corresponding to the to-be-measured monitoring point in each lightning stroke event are taken as the input, the electric field similarity between the to-be-monitored point and each highlight point corresponding to each lightning stroke event is taken as the attention weight, and the historical lightning current time series data of each highlight point corresponding to the to-be-measured point in each lightning stroke event is taken as the output, so as to train the ConvLSTM network model and obtain the trained network model; The current meteorological time series data and current electric field time series data in the current period of the to-be-monitored point are obtained and taken as the input of the trained network model, so as to output the lightning current time series data of the to-be-monitored point in the future period.

9. A lightning interceptor for lightning struck areas of a forest comprising a lightning interceptor body, characterized in that, The lightning interceptor body is built-in with a lightning current prediction module, and the lightning current prediction module is used to realize the steps in the lightning current amplitude prediction method for forest lightning stroke area according to any one of claims 1-8.

Images