Lightning ground lightning trajectory prediction method and device, storage medium and computer equipment
The method combines interpolation fitting and neural networks to enhance the accuracy and robustness of lightning trajectory predictions by using centroid points and correcting deviations, addressing the limitations of single prediction methods.
Patent Information
- Application Number
- CN202411211212.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-30
- Publication Date
- 2024-12-06
AI Technical Summary
Existing lightning trajectory prediction methods used by power systems are single and lack robustness, leading to low accuracy in predicting the movement trajectory of lightning-to-ground lightning.
A method combining interpolation fitting algorithms and neural network models to predict lightning ground lightning trajectories, using centroid points at multiple historical moments, and correcting predictions if they fall outside the observation area.
Improves the accuracy of lightning trajectory predictions by leveraging the strengths of both methods, enhancing robustness and overcoming the limitations of single prediction approaches.
Smart Images

Figure CN119089228A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of lightning activity analysis, and in particular to a method, device, storage medium and computer equipment for predicting lightning ground lightning trajectories. Background Art
[0002] In recent years, lightning monitoring technology has developed rapidly, and technical means such as lightning location systems have been widely used in lightning observation. The longitude and latitude coordinates of the lightning centroid obtained by the power system using these technologies are usually recorded in the lightning location system, including the time of occurrence of ground lightning, longitude and latitude coordinates, lightning current amplitude and polarity, number of lightning return strokes, etc., so as to carry out research on the laws related to lightning ground lightning activities and provide a certain reference basis for lightning protection of electrical equipment.
[0003] At present, the lightning location system has been developed to a mature level. The location accuracy of the arrival time difference location method (TDOA) can reach 300-1000 meters, and it is widely used in the field of lightning-to-ground lightning observation and law research. In detail, the existing technology can use the corresponding prediction method to predict the longitude and latitude coordinates of the lightning centroid point, which can grasp the movement trend of the lightning-to-ground lightning trajectory to a certain extent. However, in the existing research, the prediction method used by the power system in predicting the movement trajectory of lightning-to-ground lightning is single, resulting in low robustness, which in turn affects the accuracy of the prediction results. Summary of the invention
[0004] The purpose of the present application is to solve at least one of the above-mentioned technical defects, especially the technical defect that in the prior art, when the power system predicts the movement trajectory of lightning ground lightning, the prediction method adopted is single, resulting in low robustness, which in turn affects the accuracy of the prediction results.
[0005] The present application provides a method for predicting lightning ground lightning trajectory, the method comprising:
[0006] Acquire observation data of lightning to ground flashes in an observation area, and determine the centroid points of the lightning to ground flashes at multiple consecutive historical moments according to the observation data;
[0007] An interpolation fitting algorithm is used to perform interpolation fitting prediction on each centroid point to obtain a first prediction result, and a first predicted centroid point of the lightning to ground flash at the next moment is determined according to the first prediction result;
[0008] Predicting the position of each centroid point based on the neural network model to obtain a second prediction result, and determining a second predicted centroid point of the lightning to ground flash at a next moment according to the second prediction result;
[0009] The midpoint of the first predicted centroid point and the second predicted centroid point is obtained to obtain a target predicted centroid point, and the target predicted trajectory of the lightning ground flash is determined based on the target predicted centroid point and the centroid points of each historical moment.
[0010] Optionally, determining the centroid of the lightning to ground flash at multiple consecutive historical moments according to the observation data includes:
[0011] A spatiotemporal clustering algorithm is used to perform cluster analysis on the observation data to obtain lightning to ground flash clusters at multiple consecutive historical moments;
[0012] The longitude and latitude coordinates of each data point in each lightning-to-ground lightning cluster are determined, and the centroid point of each lightning-to-ground lightning cluster is calculated based on the longitude and latitude coordinates of each data point.
[0013] Optionally, the adopting an interpolation fitting algorithm to perform interpolation fitting prediction on each centroid point to obtain a first prediction result includes:
[0014] According to the latitude and longitude coordinates of each centroid point, a centroid point fitting line of the lightning ground lightning is generated by connecting them in chronological order;
[0015] Using an interpolation fitting algorithm to perform interpolation fitting on the centroid point fitting polyline to generate a centroid point fitting curve;
[0016] The moving speed of each centroid point is obtained, and a first prediction result of the lightning to ground flash at the next moment is determined from the centroid point fitting curve according to each moving speed.
[0017] Optionally, determining a first predicted centroid of the lightning to ground flash at a next moment according to the first prediction result includes:
[0018] Determine whether the longitude and latitude coordinates in the first prediction result are within the observation area;
[0019] If yes, taking the first prediction result as the first predicted centroid point of the lightning to ground flash at the next moment;
[0020] If not, construct the speed coverage range of the center of mass point at the current moment according to the center of mass point at the current moment and the first prediction result, and correct the longitude and latitude coordinates of the first prediction result based on the speed coverage range, the center of mass point fitting curve and the boundary line of the observation area to obtain the first predicted center of mass point of the lightning ground flash at the next moment.
[0021] Optionally, the performing latitude and longitude coordinate correction on the first prediction result based on the speed coverage range, the centroid fitting curve and the boundary line of the observation area to obtain a first predicted centroid of the lightning to ground flash at the next moment includes:
[0022] Respectively determining the velocity coverage range, the intersection points of the centroid fitting curve and the boundary line of the observation area, and the velocity vector of each intersection point;
[0023] A reference velocity vector of the lightning to ground flash is determined based on the centroid point at the first moment of the lightning to ground flash and the centroid point at the current moment, and the intersection point at which the angle between the velocity vector and the reference velocity vector is the smallest among each intersection point is used as the first predicted centroid point of the lightning to ground flash at the next moment.
[0024] Optionally, the performing position prediction on each centroid point based on the neural network model to obtain a second prediction result includes:
[0025] Determine a centroid prediction model, wherein the centroid prediction model is obtained by training with a preset neural network model;
[0026] Determine the longitude and latitude coordinates of each centroid point, and input each data into the centroid point prediction model to obtain a second prediction result output by the centroid point prediction model.
[0027] Optionally, determining a second predicted centroid of the lightning to ground flash at a next moment according to the second prediction result includes:
[0028] Determine whether the latitude and longitude coordinates in the second prediction result are within the observation area;
[0029] If yes, taking the second prediction result as the second predicted centroid point of the lightning to ground flash at the next moment;
[0030] If not, construct a target auxiliary line of the second prediction result in the observation area, determine at least one intersection of the target auxiliary line and the boundary line of the observation area, and correct the longitude and latitude coordinates of the second prediction result according to the number of intersections to obtain the second predicted centroid point of the lightning flash at the next moment.
[0031] Optionally, the step of correcting the latitude and longitude coordinates of the second prediction result according to the number of intersections to obtain a second predicted centroid point of the lightning to ground flash at the next moment includes:
[0032] When the number of intersection points is one, the intersection point is used as the second predicted centroid point of the lightning to ground flash at the next moment;
[0033] When there are two intersection points, the velocity vectors of the two intersection points are determined, and the intersection line segment between the two intersection points is connected, and the second predicted centroid point of the lightning to ground flash at the next moment is determined on the intersection line segment according to the velocity vectors of the two intersection points.
[0034] The present application also provides a lightning ground lightning trajectory prediction device, comprising:
[0035] A centroid determination module is used to obtain observation data of lightning to ground flashes in an observation area, and determine the centroid of the lightning to ground flashes at multiple consecutive historical moments according to the observation data;
[0036] A first prediction module is used to perform interpolation fitting prediction on each centroid point by using an interpolation fitting algorithm to obtain a first prediction result, and determine a first predicted centroid point of the lightning to ground flash at the next moment according to the first prediction result;
[0037] A second prediction module is used to predict the position of each centroid point based on the neural network model to obtain a second prediction result, and determine a second predicted centroid point of the lightning to ground flash at the next moment according to the second prediction result;
[0038] The trajectory determination module is used to obtain the midpoint of the first predicted centroid point and the second predicted centroid point to obtain a target predicted centroid point, and determine the target predicted trajectory of the lightning ground flash based on the target predicted centroid point and the centroid points of each historical moment.
[0039] The present application also provides a storage medium, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the lightning ground lightning trajectory prediction method as described in any of the above embodiments.
[0040] The present application also provides a computer device, comprising: one or more processors, and a memory;
[0041] The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the steps of the lightning ground lightning trajectory prediction method as described in any one of the above embodiments are performed.
[0042] It can be seen from the above technical solutions that this application has the following advantages:
[0043] The lightning-to-ground lightning trajectory prediction method, device, storage medium and computer equipment provided in the present application can first obtain the observation data of the lightning-to-ground lightning in the observation area when predicting the movement trajectory of the lightning-to-ground lightning in the observation area, and determine the centroid points of the lightning-to-ground lightning at multiple consecutive historical moments based on the observation data, so as to serve as the basic data for the lightning-to-ground lightning trajectory prediction. Then, based on each centroid point, different prediction methods can be used to predict the centroid point at the next moment, such as using an interpolation fitting algorithm to predict a first predicted centroid point and using a neural network model to predict a second predicted centroid point; wherein, in the centroid point prediction process, if the prediction result deviates from the observation area, the prediction result can be corrected to improve the accuracy of the first predicted centroid point and the second predicted centroid point. Finally, the present application can find the midpoint of the first predicted centroid point and the second predicted centroid point to obtain the target predicted centroid point, and determine the target predicted trajectory of lightning ground flashes based on the target predicted centroid point and the centroid points of each historical moment. In this way, the present application can combine the advantages of different prediction methods, learn from each other's strengths and overcome their weaknesses, and avoid the low robustness caused by adopting a single prediction method, which affects the accuracy of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0045] Figure 1 A schematic diagram of a flow chart of a lightning ground lightning trajectory prediction method provided in an embodiment of the present application;
[0046] Figure 2 A schematic diagram of the result of an interpolation fitting prediction provided in an embodiment of the present application;
[0047] Figure 3 A schematic diagram of a result of a first predicted centroid correction provided in an embodiment of the present application;
[0048] Figure 4 A schematic diagram of a neural network prediction result provided in an embodiment of the present application;
[0049] Figure 5 A schematic diagram of a result of a second predicted centroid correction provided in an embodiment of the present application;
[0050] Figure 6 A schematic diagram of the structure of a lightning ground lightning trajectory prediction device provided in an embodiment of the present application;
[0051] Figure 7A schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0052] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0053] At present, the lightning location system has been developed to a mature level. The location accuracy of the arrival time difference location method (TDOA) can reach 300-1000 meters, and it is widely used in the field of lightning-to-ground lightning observation and law research. In detail, the existing technology can use the corresponding prediction method to predict the longitude and latitude coordinates of the lightning centroid point, which can grasp the movement trend of the lightning-to-ground lightning trajectory to a certain extent. However, in the existing research, the prediction method used by the power system in predicting the movement trajectory of lightning-to-ground lightning is single, resulting in low robustness, which in turn affects the accuracy of the prediction results.
[0054] Based on this, this application proposes the following technical solutions, see below for details:
[0055] In one embodiment, Figure 1 As shown, Figure 1 A schematic diagram of a flow chart of a lightning ground lightning trajectory prediction method provided in an embodiment of the present application; the present application provides a lightning ground lightning trajectory prediction method, which specifically includes the following:
[0056] S110: Obtain observation data of lightning to ground flashes in an observation area, and determine centroid points of lightning to ground flashes at multiple consecutive historical moments based on the observation data.
[0057] In this embodiment, when the user needs to predict the movement trajectory of lightning flashes to ground in the observation area, the computer device can directly obtain the observation data of lightning flashes to ground in the observation area from the lightning locating system, and then determine the center of mass points of lightning flashes to ground at multiple consecutive historical moments based on the observation data, as the basic data for predicting the trajectory of lightning flashes to ground.
[0058] Among them, the centroid point represents the central position of all points in a data set or cluster, which can usually be determined by calculating the average value of all point coordinates. In this application, the centroid point can refer to the central position of a lightning-to-ground lightning event in a certain area during a certain period of time. By monitoring and analyzing the changes in the centroid point of lightning-to-ground lightning, computer equipment can predict the concentrated area and change trend of lightning-to-ground lightning events, providing important data support for lightning protection and disaster reduction.
[0059] It is understandable that when a lightning strike to the ground occurs in the observation area, the power system can conduct lightning observations in the observation area through lightning monitoring technology, and record the relevant data of the observed lightning strike to the ground into the lightning locating system. In this way, when the computer equipment predicts the movement trajectory of the lightning strike to the ground, it can directly obtain the observation data of the lightning strike to the ground in the corresponding observation area from the lightning locating system.
[0060] Furthermore, after the computer device obtains the observation data of lightning flashes in the observation area, it can also perform preprocessing operations on the observation data to improve the accuracy of the observation data. The preprocessing operations include but are not limited to denoising, data cleaning, feature extraction and dimensionality reduction, and data normalization.
[0061] To elaborate, denoising refers to removing noise and interference from observed data to improve data quality and availability; data cleaning refers to removing outliers, supplementing missing values, and smoothing observed data to improve data accuracy and completeness; feature extraction and dimensionality reduction refer to extracting effective information from observed data and converting the extracted effective information into a low-dimensional vectorized form to facilitate subsequent mathematical operations and analysis; and data normalization refers to normalizing observed data to ensure that data with different features have the same weights to avoid data bias and errors.
[0062] S120: using an interpolation fitting algorithm to perform interpolation fitting prediction on each centroid point to obtain a first prediction result, and determining a first predicted centroid point of the lightning to ground flash at the next moment according to the first prediction result.
[0063] In this embodiment, after obtaining the observation data of the lightning-to-ground flash in step S120, the computer device can use a corresponding prediction method to predict the centroid point of the lightning-to-ground flash at the next moment. For example, the computer device can use an interpolation fitting algorithm to perform interpolation fitting prediction on each centroid point to obtain a first prediction result, and then determine the first predicted centroid point of the lightning-to-ground flash at the next moment according to the first prediction result.
[0064] The interpolation fitting algorithm refers to a type of mathematical method used to estimate unknown values between or outside given data points. Therefore, in this application, the computer device can use the interpolation fitting algorithm to generate a smooth curve, and continue to search for the possible position of the centroid point at the next moment on the curve based on the centroid point at each historical moment.
[0065] It is understandable that in the prediction process of the centroid of lightning ground lightning, the interpolation fitting algorithm can reduce the impact of noise, consider the temporal and spatial changes, and improve the prediction accuracy, thereby providing a continuous and smooth centroid prediction result, which has strong visualization and interpretability. Of course, different types of interpolation fitting algorithms have different characteristics in the centroid prediction process, so this application can select the corresponding interpolation fitting algorithm according to actual needs, and there is no limitation here.
[0066] For example, interpolation fitting algorithms include but are not limited to Nearest, Cubic, and Spline. Among them, the Nearest method has the largest error and relatively poor smoothness; the Cubic method is relatively smooth and fast, but occupies the most memory; the Spline method is slow, but the smoothest and occupies less memory.
[0067] S130: Predict the position of each centroid point based on the neural network model to obtain a second prediction result, and determine a second predicted centroid point of the lightning to ground flash at the next moment according to the second prediction result.
[0068] In this embodiment, after obtaining the observation data of the lightning-to-ground flash in step S120, the computer device can use a corresponding prediction method to predict the centroid of the lightning-to-ground flash at the next moment. For example, the computer device can perform position prediction on each centroid based on a neural network model to obtain a second prediction result, and determine a second predicted centroid of the lightning-to-ground flash at the next moment according to the second prediction result.
[0069] Among them, the neural network model can use LSTM (Long Short-Term Memory) neural network, which is a special recurrent neural network (RNN) architecture used to process sequence data with long-term dependencies. Therefore, this application can process the relevant data of the centroid of each historical moment through the pre-trained LSTM neural network model, and then predict the possible location of the centroid at the next moment.
[0070] It is understandable that in the process of predicting the centroid of lightning ground flashes, since lightning ground flash events are affected by multiple factors, their observation data show a high degree of nonlinearity and uncertainty. Therefore, the present application can utilize the characteristics of the neural network model that can process complex time series data, and use it to capture the long-term dependence and nonlinear relationship between each centroid point, thereby further improving the prediction accuracy of the centroid point.
[0071] S140: Find the midpoint of the first predicted centroid point and the second predicted centroid point to obtain a target predicted centroid point, and determine the target predicted trajectory of the lightning to ground flash based on the target predicted centroid point and the centroid points at each historical moment.
[0072] In this embodiment, after the first predicted centroid point and the second predicted centroid point are predicted respectively through step S120 and step S130, the computer device can obtain the midpoint of the first predicted centroid point and the second predicted centroid point to obtain the target predicted centroid point, and determine the target predicted trajectory of the lightning ground flash based on the target predicted centroid point and the centroid points of each historical moment, thereby improving the accuracy of the target predicted trajectory by improving the accuracy of the target predicted centroid point.
[0073] It is understandable that the first predicted centroid point is predicted by an interpolation fitting algorithm, which is sensitive to noise and can reflect the local characteristics of the data to a certain extent, which has certain advantages for processing observation data containing noise; the second predicted centroid point is predicted based on a neural network model, which can capture long-term dependencies in time series data and predict global trends. Therefore, the target predicted centroid point obtained by the midpoint calculation of the first predicted centroid point and the second predicted centroid point in this application can combine the advantages of the two prediction methods, learn from each other's strengths, improve robustness and accuracy, and thus improve the accuracy of the target prediction trajectory.
[0074] In the above embodiment, when predicting the moving trajectory of lightning to ground flash in the observation area, the observation data of lightning to ground flash in the observation area can be obtained first, and the centroid points of lightning to ground flash at multiple consecutive historical moments can be determined based on the observation data to serve as the basic data for lightning to ground flash trajectory prediction. Then, based on each centroid point, different prediction methods can be used to predict the centroid point at the next moment, such as using an interpolation fitting algorithm to predict the first predicted centroid point and using a neural network model to predict the second predicted centroid point; wherein, in the centroid point prediction process, if the prediction result deviates from the observation area, the prediction result can be corrected to improve the accuracy of the first predicted centroid point and the second predicted centroid point. Finally, the present application can obtain the midpoint of the first predicted centroid point and the second predicted centroid point to obtain the target predicted centroid point, and determine the target predicted trajectory of lightning to ground flash based on the target predicted centroid point and the centroid points at each historical moment. In this way, the present application can combine the advantages of different prediction methods, learn from each other's strengths and overcome their weaknesses, and avoid the low robustness caused by adopting a single prediction method, which affects the accuracy of the prediction result.
[0075] In one embodiment, determining the centroid of lightning to ground lightning at multiple consecutive historical moments according to the observation data in step S110 may include:
[0076] S111: Use spatiotemporal clustering algorithm to perform cluster analysis on the observation data and obtain lightning to ground flash clusters at multiple consecutive historical moments.
[0077] S112: Determine the longitude and latitude coordinates of each data point in each lightning-to-ground lightning cluster, and calculate the centroid of each lightning-to-ground lightning cluster according to the longitude and latitude coordinates of each data point.
[0078] In this embodiment, after the computer device obtains the observation data of lightning to ground flashes, it can use a spatiotemporal clustering algorithm to perform cluster analysis on the observation data to obtain lightning to ground flash clusters at multiple consecutive historical moments. Then, the longitude and latitude coordinates of each data point in each lightning to ground flash cluster can be determined, and the center of mass of each lightning to ground flash cluster can be calculated based on the longitude and latitude coordinates of each data point.
[0079] Among them, the spatiotemporal clustering algorithm is an algorithm for processing spatiotemporal data, which can group data points with similar spatiotemporal attributes into clusters or clusters. Therefore, the present application can determine the similarity of each data point in each observation data according to the spatial distance and time distance between them through the spatiotemporal clustering algorithm, and then obtain lightning ground lightning clusters at multiple consecutive historical moments. Furthermore, the spatiotemporal clustering algorithm used in the present application may include the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) algorithm and the ST-DBSCAN (Spatio-Temporal DBSCAN) algorithm, which are not limited here.
[0080] Specifically, the computer device can perform spatial and temporal clustering on the observation data through the spatiotemporal clustering algorithm. To elaborate, the computer device can first identify the clustering pattern of the observation data in the geographical location based on the spatial dimension of the observation data, and then further analyze the distribution of the observation data in the temporal dimension on the basis of spatial clustering to identify the clustering pattern of the observation data in time. Finally, the computer device can combine the results of spatial clustering and temporal clustering, comprehensively analyze the spatiotemporal distribution of the observation data, identify the spatiotemporal clustering pattern, and obtain lightning ground lightning clusters at multiple consecutive historical moments.
[0081] It should be noted that since the centroid point represents the center position of all points in a data set or cluster, that is, the center position of the lightning-to-ground lightning cluster, the computer equipment can first determine the longitude and latitude coordinates of each data point in the lightning-to-ground lightning cluster, and then obtain the average value of the longitude and latitude coordinates of each data point, and use the average value to characterize the position of the centroid point of the corresponding lightning-to-ground lightning cluster.
[0082] In one embodiment, in step S120, using an interpolation fitting algorithm to perform interpolation fitting prediction on each centroid point to obtain a first prediction result may include:
[0083] S121: Generate a lightning ground lightning centroid point fitting line by connecting the centroid points according to the latitude and longitude coordinates of each centroid point in chronological order.
[0084] S122: Using an interpolation fitting algorithm to perform interpolation fitting on the centroid point fitting polyline to generate a centroid point fitting curve.
[0085] S123: Obtain the moving speed of each centroid point, and determine a first prediction result of a lightning flash to ground at the next moment from a centroid fitting curve according to each moving speed.
[0086] In this embodiment, when the computer device uses the interpolation fitting algorithm to predict the centroid point, it can first connect the latitude and longitude coordinates of each centroid point in chronological order to generate a centroid point fitting line of the lightning to ground flash. Then the computer device can use the interpolation fitting algorithm to interpolate and fit the centroid point fitting line to generate a centroid fitting curve. Finally, the computer device can obtain the moving speed of each centroid point, and determine the first prediction result of the lightning to ground flash at the next moment from the centroid fitting curve according to each moving speed.
[0087] Specifically, the present application can use cubic spline interpolation to perform interpolation fitting prediction on each centroid point. In detail, the computer device can construct a coordinate axis based on the observation area, which is in the horizontal axis interval Contains ( ) centroid points with known ordinates, that is, the interval is divided into For each segment , are all represented by cubic functions. These segmented cubic functions take the same value at the centroid and have the same first-order and second-order derivatives at the centroid. The purpose of cubic spline interpolation is to find the coefficients of each segment of cubic function that meets the above conditions. Since the ordinates of all the centroids are known, we can list equations; except for the first and last points of the interval, the function values and first-order and second-order derivative values of the two adjacent segments at the centroid are equal, and a total of equations. Total number of equations and the number of unknown numbers The difference of 2 can be given by boundary conditions, which generally use natural boundary conditions, that is, the second-order derivative value at the endpoint of the interval is set to 0. Finally, by solving the equation, the expression of the cubic function of each segment can be obtained.
[0088] Indicatively, Figure 2 As shown, Figure 2A schematic diagram of the result of an interpolation fitting prediction provided in an embodiment of the present application; Figure 2 In the above example, the computer device can select 1-3 known centroid points, and then obtain two segments of cubic functions. When fitting the curve, the definition domains of the two segments of functions can be extended by the fitting formula, so as to find the fourth point on the formed centroid fitting curve, i.e., the first prediction result. The specific fitting formula can be expressed as follows:
[0089]
[0090] In the formula, the default , that is, the fitted ground-to-ground lightning trajectory is from west to east; if the trajectory trend is opposite to the normal trend, that is, from east to west, then the interval expression of the above formula can be changed accordingly, and no restriction is made here.
[0091] Furthermore, after obtaining the centroid fitting curve, the computer device can find the latitude and longitude coordinates of the fourth point on the centroid fitting curve. According to the continuity of lightning strikes to ground lightning events, that is, the physical scenario where there will be no sudden change in speed in a short period of time, the speed from the first point to the second point is known. and the speed from point 2 to point 3 , we can take a speed that is the average of these two speeds, that is ,like Figure 2 As shown, with the third point as the center, The intersection of the circle with the centroid fitting curve is the prediction result of the 4th point. If there are two intersections in the prediction result, the previous intersection can be discarded.
[0092] It should be noted that, since the cubic spline interpolation fits the functional relationship between the ordinate and the abscissa, if the movement trend of the centroid fitting curve composed of points 1 to 3 is from south to north or from north to south, the abscissa and the ordinate can be interchanged for processing.
[0093] In one embodiment, determining the first predicted centroid of the lightning to ground flash at the next moment according to the first prediction result in step S120 may include:
[0094] S124: Determine whether the longitude and latitude coordinates in the first prediction result are within the observation area.
[0095] S125: If yes, the first prediction result is used as the first predicted centroid point of the lightning to ground flash at the next moment.
[0096] S126: If not, construct the speed coverage range of the center of mass point at the current moment according to the center of mass point at the current moment and the first prediction result, and correct the longitude and latitude coordinates of the first prediction result based on the speed coverage range, the center of mass point fitting curve and the boundary line of the observation area to obtain the first predicted center of mass point of the lightning to ground flash at the next moment.
[0097] In this embodiment, after the computer device uses interpolation fitting prediction to obtain the first prediction result, it can first determine whether the longitude and latitude coordinates in the first prediction result are within the observation area. If it is within the observation area, it means that the lightning strike point of the lightning to ground flash at the next moment is within the observation range. At this time, the computer device can use the first prediction result as the first predicted centroid point of the lightning to ground flash at the next moment. If it is not within the observation area, it means that the lightning strike point of the lightning to ground flash at the next moment is not within the observation range, and it does not belong to the object to be observed. Therefore, the computer device needs to take corresponding measures to correct the position of the first prediction result, and obtain the first predicted centroid point of the lightning to ground flash at the next moment, thereby improving the accuracy of the first predicted centroid point.
[0098] Specifically, when the longitude and latitude coordinates in the first prediction result are not within the observation area, the computer device can obtain the centroid point at the current moment, that is, the centroid point of the known longitude and latitude coordinates at the last historical moment in the centroid point fitting curve, and construct the speed coverage range of the centroid point at the current moment with the first prediction result, that is, the circular area constructed with the centroid point at the current moment as the center of the circle and the straight-line distance between the centroid point at the current moment and the first prediction result as the radius. Finally, the computer device can correct the longitude and latitude coordinates of the first prediction result based on the speed coverage range, the centroid point fitting curve and the boundary line of the observation area to obtain the first predicted centroid point of the lightning ground flash at the next moment.
[0099] In one embodiment, in step S126, the longitude and latitude coordinates of the first prediction result are corrected based on the speed coverage range, the centroid fitting curve, and the boundary line of the observation area to obtain the first predicted centroid of the lightning to ground flash at the next moment, which may include:
[0100] S1261: Determine the velocity coverage range, the intersection points of the centroid fitting curve and the boundary line of the observation area, and the velocity vector of each intersection point respectively.
[0101] S1262: Determine a reference velocity vector of the lightning to ground flash based on the centroid point at the first moment of the lightning to ground flash and the centroid point at the current moment, and use the intersection point with the smallest angle between the velocity vector and the reference velocity vector among each intersection point as the first predicted centroid point of the lightning to ground flash at the next moment.
[0102] In this embodiment, when the computer device corrects the longitude and latitude coordinates of the first prediction result, it can respectively determine the speed coverage range, the intersection of the centroid fitting curve and the boundary line of the observation area, and the speed vector of each intersection, and then determine the reference speed vector of the lightning to ground flash based on the centroid of the first moment of the lightning to ground flash and the centroid of the current moment, and use the intersection with the smallest angle between the speed vector and the reference speed vector among each intersection as the first predicted centroid of the lightning to ground flash at the next moment.
[0103] Indicatively, Figure 3 As shown, Figure 3 A schematic diagram of a result of a first predicted centroid correction provided in an embodiment of the present application; Figure 3 In the figure, 1, 2, and 3 are the centroids of three consecutive historical moments. The first prediction result of the prediction 4 before correction is outside the boundary line of the observation area. The circular area with 3 as the center is the speed coverage range, which represents the magnitude of the speed, and the fitting curve represents the direction of the speed. The correction method used by the computer at this time is: take the intersection of the speed coverage range, the centroid fitting curve and the boundary line, a total of 3, recorded as points Generally speaking, The magnitudes of the lightning strikes are not much different, but the directions are quite different. Based on the continuity of lightning strikes to ground, The point corresponding to the minimum of is the revised prediction 4. It can be seen that the revised prediction 4 is no longer outside the boundary, and the error between it and the actual 4 is also reduced.
[0104] In one embodiment, in step S130, the position of each centroid point is predicted based on the neural network model to obtain a second prediction result, which may include:
[0105] S131: Determine a centroid prediction model, where the centroid prediction model is obtained by training with a preset neural network model.
[0106] S132: Determine the longitude and latitude coordinates of each centroid point, and input the longitude and latitude coordinates of each centroid point into a centroid prediction model to obtain a second prediction result output by the centroid prediction model.
[0107] In this embodiment, when the computer device uses a neural network model to predict the centroid, it can first determine the centroid model and the longitude and latitude coordinates of each centroid, and then input the longitude and latitude coordinates of each centroid into the centroid prediction model to obtain a second prediction result output by the centroid prediction model. The second prediction result obtained in this way can capture the longitude and latitude coordinates of each centroid, predict the global trend, and thus improve the accuracy of the prediction result.
[0108] It can be understood that the centroid prediction model of the present application refers to a model that performs trend prediction on the longitude and latitude coordinates of the centroid of each continuous historical moment of the input and obtains the longitude and latitude coordinates of the centroid of the next moment. The centroid prediction model can be obtained by training a neural network model. When performing model training, the sample observation data of different lightning strikes to ground lightning events can be used as training samples, and the longitude and latitude coordinates of the centroid of the last moment in the training sample are marked as sample labels. After all training samples are marked, the training samples with sample labels can be input into the preset neural network model for forward propagation to train the model, and the preset target loss function is used to tune the parameters of the model during the back propagation of the model. When the model meets certain training conditions or parameter convergence conditions, such as the number of iterations reaches the set value, the training is considered to be completed, and the trained model can be used as the final centroid prediction model. Therefore, the present application inputs the longitude and latitude coordinates of each centroid into the centroid prediction model to obtain the second prediction result output by the centroid prediction model.
[0109] Specifically, during the training of the neural network model, the computer device can use the given known Get the predicted and conduct and The error calculation is performed each time, and the network parameters are updated each time, so that the error of the model can be reduced next time, until the error is reduced to almost zero, the network converges, and the training is considered complete. At this time, the network training process is performed again, and the output That is the The prediction result at the moment. More specifically, the computer device can train the neural network model 500 times, and use RMSE and loss function to represent the error in the training process. Among them, the loss function is used to estimate the inconsistency between the predicted value of the model and the true value. The smaller the value, the better the robustness of the model. This loss function can effectively eliminate the gradient vanishing and gradient exploding phenomena in the training of deep learning models, making the model converge faster; and RMSE represents the root mean square error of all neurons. For a single neuron, its definition can be as follows:
[0110]
[0111] Indicatively, Figure 4 As shown, Figure 4 A schematic diagram of a neural network prediction result provided in an embodiment of the present application; Figure 4In the process of model prediction, after determining the centroid prediction model, the computer can input the latitude and longitude coordinates of the centroid of the previous three consecutive historical moments into the centroid prediction model, so as to obtain the predicted value of the fourth point predicted by the centroid prediction model. In the process of model prediction, the computer device executes two neural networks respectively, the first time predicting the longitude of the fourth point, and the second time predicting the latitude of the fourth point, thereby forming the predicted longitude and latitude coordinates of the fourth point.
[0112] In one embodiment, determining the second predicted centroid of the lightning to ground flash at the next moment according to the second prediction result in step S130 may include:
[0113] S133: Determine whether the longitude and latitude coordinates in the second prediction result are within the observation area.
[0114] S134: If yes, the second prediction result is used as the second predicted centroid point of the lightning to ground flash at the next moment.
[0115] S135: If not, construct a target auxiliary line of the second prediction result in the observation area, determine at least one intersection point between the target auxiliary line and the boundary line of the observation area, and correct the longitude and latitude coordinates of the second prediction result according to the number of intersection points to obtain the second predicted centroid point of the lightning to ground flash at the next moment.
[0116] In this embodiment, after the computer device obtains the second prediction result based on the neural network model prediction, it can first determine whether the latitude and longitude coordinates in the second prediction result are within the observation area. If it is within the observation area, it means that the lightning strike point of the lightning to ground flash at the next moment is within the observation range. At this time, the computer device can use the second prediction result as the second predicted centroid point of the lightning to ground flash at the next moment. If it is not within the observation area, it means that the lightning strike point of the lightning to ground flash at the next moment is not within the observation range, and it does not belong to the object to be observed. Therefore, the computer device needs to take corresponding measures to correct the position of the second prediction result, and obtain the second predicted centroid point of the lightning to ground flash at the next moment, thereby improving the accuracy of the second predicted centroid point.
[0117] Specifically, when the longitude and latitude coordinates in the second prediction result are not in the observation area, the computer device can construct a target auxiliary line of the second prediction result in the observation area, where the auxiliary line may include a horizontal line and a vertical line, so that it intersects with the boundary line of the observation area to generate at least one intersection point. At this time, the computer device can adopt a corresponding strategy method according to the number of intersection points to correct the longitude and latitude coordinates of the second prediction result to obtain the second predicted centroid point of the lightning ground flash at the next moment.
[0118] In one embodiment, in step S135, the second prediction result is corrected in terms of longitude and latitude coordinates according to the number of intersection points to obtain a second predicted centroid point of the lightning to ground flash at the next moment, which may include:
[0119] S1351: When the number of intersection points is one, the intersection point is used as the second predicted centroid point of the lightning to ground flash at the next moment.
[0120] S1352: When there are two intersection points, determine the velocity vectors of the two intersection points, and connect the intersection line segment between the two intersection points, and determine the second predicted centroid point of the lightning to ground flash at the next moment on the intersection line segment according to the velocity vectors of the two intersection points.
[0121] In this embodiment, when there is only one intersection point between the target auxiliary line and the boundary line of the observation area, the computer device can directly use the intersection point as the second predicted centroid point of the lightning to ground flash at the next moment. When there are two intersection points between the target auxiliary line and the boundary line of the observation area, the computer device can first determine the velocity vectors of the two intersection points, and connect to obtain the intersection line segment between the two intersection points, and then determine the second predicted centroid point of the lightning to ground flash at the next moment on the intersection line segment according to the velocity vectors of the two intersection points.
[0122] Indicatively, Figure 5 As shown, Figure 5 A schematic diagram of a result of a second predicted centroid correction provided in an embodiment of the present application; Figure 5 In the example, the large arc is the boundary line of the observation area, 1, 2, and 3 are the three known centroid points, and the pre-correction prediction 4 is the second prediction result, which is outside the boundary line. The computer device can construct a target auxiliary line of a preset length on the pre-correction prediction 4. If there are two intersection points with the boundary line, denoted as A and B, then the correction target is to find a point 4 on the line segment AB as the corrected prediction 4. Generally, and The angle between the two will not be too large, so the speed is the main consideration. The computer device can select a point 4 on the line segment AB, that is, The corresponding point is taken as the revised prediction 4.
[0123] The lightning-to-ground lightning flash trajectory prediction device provided in an embodiment of the present application is described below. The lightning-to-ground lightning flash trajectory prediction device described below and the lightning-to-ground lightning flash trajectory prediction method described above can be referenced to each other.
[0124] In one embodiment, Figure 6 As shown, Figure 6A schematic diagram of a lightning-to-ground lightning trajectory prediction device provided in an embodiment of the present application; the present application also provides a lightning-to-ground lightning trajectory prediction device, including a centroid determination module 210, a first prediction module 220, a second prediction module 230 and a trajectory determination module 240, specifically including the following:
[0125] The centroid determination module 210 is used to obtain observation data of lightning to ground lightning in the observation area, and determine the centroid of the lightning to ground lightning at multiple consecutive historical moments according to the observation data.
[0126] The first prediction module 220 is used to perform interpolation fitting prediction on each centroid point by using an interpolation fitting algorithm to obtain a first prediction result, and determine a first predicted centroid point of the lightning to ground flash at the next moment according to the first prediction result.
[0127] The second prediction module 230 is used to predict the position of each centroid point based on the neural network model to obtain a second prediction result, and determine the second predicted centroid point of the lightning to ground flash at the next moment according to the second prediction result.
[0128] The trajectory determination module 240 is used to obtain the midpoint of the first predicted centroid point and the second predicted centroid point to obtain the target predicted centroid point, and determine the target predicted trajectory of the lightning ground flash based on the target predicted centroid point and the centroid points at each historical moment.
[0129] In the above embodiment, when predicting the moving trajectory of lightning to ground flash in the observation area, the observation data of lightning to ground flash in the observation area can be obtained first, and the centroid points of lightning to ground flash at multiple consecutive historical moments can be determined based on the observation data to serve as the basic data for lightning to ground flash trajectory prediction. Then, based on each centroid point, different prediction methods can be used to predict the centroid point at the next moment, such as using an interpolation fitting algorithm to predict the first predicted centroid point and using a neural network model to predict the second predicted centroid point; wherein, in the centroid point prediction process, if the prediction result deviates from the observation area, the prediction result can be corrected to improve the accuracy of the first predicted centroid point and the second predicted centroid point. Finally, the present application can obtain the midpoint of the first predicted centroid point and the second predicted centroid point to obtain the target predicted centroid point, and determine the target predicted trajectory of lightning to ground flash based on the target predicted centroid point and the centroid points at each historical moment. In this way, the present application can combine the advantages of different prediction methods, learn from each other's strengths and overcome their weaknesses, and avoid the low robustness caused by adopting a single prediction method, which affects the accuracy of the prediction result.
[0130] In one embodiment, the centroid determination module 210 may include:
[0131] The cluster analysis submodule is used to perform cluster analysis on the observation data using a spatiotemporal clustering algorithm to obtain lightning to ground flash clusters at multiple consecutive historical moments.
[0132] The centroid calculation submodule is used to determine the longitude and latitude coordinates of each data point in each lightning-to-ground lightning cluster, and calculate the centroid of each lightning-to-ground lightning cluster based on the longitude and latitude coordinates of each data point.
[0133] In one embodiment, the first prediction module 220 may include:
[0134] The polyline generation submodule is used to connect the centroid points of lightning ground lightning in time sequence according to the latitude and longitude coordinates of each centroid point.
[0135] The curve fitting submodule is used to perform interpolation fitting on the centroid point fitting polyline using an interpolation fitting algorithm to generate a centroid point fitting curve.
[0136] The fitting prediction submodule is used to obtain the moving speed of each centroid point and determine the first prediction result of the lightning to ground flash at the next moment from the centroid point fitting curve according to each moving speed.
[0137] In one embodiment, the first prediction module 220 may further include:
[0138] The first judgment submodule is used to judge whether the longitude and latitude coordinates in the first prediction result are within the observation area.
[0139] The first correction submodule is used to use the first prediction result as the first predicted centroid point of the lightning to ground flash at the next moment if the latitude and longitude coordinates in the first prediction result are within the observation area.
[0140] The second correction submodule is used to construct the speed coverage range of the center of mass point at the current moment according to the center of mass point at the current moment and the first prediction result if the longitude and latitude coordinates in the first prediction result are not within the observation area, and to correct the longitude and latitude coordinates of the first prediction result based on the speed coverage range, the center of mass point fitting curve and the boundary line of the observation area to obtain the first predicted center of mass point of the lightning to ground flash at the next moment.
[0141] In one embodiment, the second correction submodule may include:
[0142] The intersection determination unit is used to respectively determine the velocity coverage range, the intersection of the centroid fitting curve and the boundary line of the observation area, and the velocity vector of each intersection.
[0143] The centroid selection unit is used to determine the reference velocity vector of the lightning-to-ground flash based on the centroid of the first moment of the lightning-to-ground flash and the centroid of the current moment, and to use the intersection point with the smallest angle between the velocity vector and the reference velocity vector among each intersection point as the first predicted centroid of the lightning-to-ground flash at the next moment.
[0144] In one embodiment, the second prediction module 230 may include:
[0145] The model determination submodule is used to determine the centroid prediction model, which is obtained by training with a preset neural network model.
[0146] The model prediction submodule is used to determine the longitude and latitude coordinates of each centroid point, and input the longitude and latitude coordinates of each centroid point into the centroid point prediction model to obtain a second prediction result output by the centroid point prediction model.
[0147] In one embodiment, determining the second predicted centroid of the lightning to ground flash at the next moment according to the second prediction result in step S130 may include:
[0148] The second judgment submodule is used to judge whether the longitude and latitude coordinates in the second prediction result are within the observation area.
[0149] The third correction submodule is used to use the second prediction result as the second predicted centroid point of the lightning to ground flash at the next moment if the latitude and longitude coordinates in the second prediction result are within the observation area.
[0150] The fourth correction submodule is used to construct a target auxiliary line of the second prediction result in the observation area if the longitude and latitude coordinates in the second prediction result are not within the observation area, and to determine at least one intersection point between the target auxiliary line and the boundary line of the observation area, and to correct the longitude and latitude coordinates of the second prediction result according to the number of intersection points to obtain the second predicted centroid point of the lightning ground flash at the next moment.
[0151] In one embodiment, the fourth correction submodule may include:
[0152] The first determination unit is used to, when the number of intersection points is one, use the intersection point as a second predicted centroid point of the lightning to ground flash at the next moment.
[0153] The second determination unit is used to determine the velocity vectors of the two intersections when there are two intersections, and to connect the intersection line segment between the two intersections, and to determine the second predicted centroid point of the lightning to ground flash at the next moment on the intersection line segment according to the velocity vectors of the two intersections.
[0154] In one embodiment, the present application also provides a storage medium, which stores computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the lightning ground flash trajectory prediction method as described in any of the above embodiments.
[0155] In one embodiment, the present application also provides a computer device, in which computer-readable instructions are stored. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the lightning ground flash trajectory prediction method as described in any of the above embodiments.
[0156] Indicatively, Figure 7 As shown, Figure 7 This is a schematic diagram of the internal structure of a computer device provided in an embodiment of the present application. The computer device 300 may be provided as a server. Figure 7 The computer device 300 includes a processing component 302, which further includes one or more processors, and a memory resource represented by a memory 301, for storing instructions executable by the processing component 302, such as an application. The application stored in the memory 301 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 302 is configured to execute instructions to perform the lightning ground lightning trajectory prediction method of any of the above embodiments.
[0157] The computer device 300 may further include a power supply component 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input / output (I / O) interface 305. The computer device 300 may operate based on an operating system stored in the memory 301, such as Windows Server TM, Mac OS X TM, Unix TM, Linux TM, Free BSD TM, or the like.
[0158] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.
[0159] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0160] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can refer to each other.
[0161] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for predicting lightning ground lightning trajectory, characterized in that: The method comprises: Acquire observation data of lightning to ground flashes in an observation area, and determine the centroid points of the lightning to ground flashes at multiple consecutive historical moments according to the observation data; An interpolation fitting algorithm is used to perform interpolation fitting prediction on each centroid point to obtain a first prediction result, and a first predicted centroid point of the lightning to ground flash at the next moment is determined according to the first prediction result; Predicting the position of each centroid point based on the neural network model to obtain a second prediction result, and determining a second predicted centroid point of the lightning to ground flash at a next moment according to the second prediction result; The midpoint of the first predicted centroid point and the second predicted centroid point is obtained to obtain a target predicted centroid point, and the target predicted trajectory of the lightning ground flash is determined based on the target predicted centroid point and the centroid points of each historical moment.
2. The lightning ground lightning trajectory prediction method according to claim 1 is characterized in that: Determining the centroid of the lightning to ground flash at multiple consecutive historical moments according to the observation data includes: A spatiotemporal clustering algorithm is used to perform cluster analysis on the observation data to obtain lightning to ground flash clusters at multiple consecutive historical moments; The longitude and latitude coordinates of each data point in each lightning-to-ground lightning cluster are determined, and the centroid point of each lightning-to-ground lightning cluster is calculated based on the longitude and latitude coordinates of each data point.
3. The lightning ground lightning trajectory prediction method according to claim 1 is characterized in that: The interpolation fitting algorithm is used to perform interpolation fitting prediction on each centroid point to obtain a first prediction result, including: According to the latitude and longitude coordinates of each centroid point, a centroid point fitting line of the lightning ground lightning is generated by connecting them in chronological order; Using an interpolation fitting algorithm to perform interpolation fitting on the centroid point fitting polyline to generate a centroid point fitting curve; The moving speed of each centroid point is obtained, and a first prediction result of the lightning to ground flash at the next moment is determined from the centroid point fitting curve according to each moving speed.
4. The lightning ground lightning trajectory prediction method according to claim 3 is characterized in that: The determining, according to the first prediction result, a first predicted centroid point of the lightning to ground flash at a next moment comprises: Determine whether the longitude and latitude coordinates in the first prediction result are within the observation area; If yes, taking the first prediction result as the first predicted centroid point of the lightning to ground flash at the next moment; If not, construct the speed coverage range of the center of mass point at the current moment according to the center of mass point at the current moment and the first prediction result, and correct the longitude and latitude coordinates of the first prediction result based on the speed coverage range, the center of mass point fitting curve and the boundary line of the observation area to obtain the first predicted center of mass point of the lightning ground flash at the next moment.
5. The lightning ground lightning trajectory prediction method according to claim 4 is characterized in that: The step of correcting the longitude and latitude coordinates of the first prediction result based on the speed coverage range, the centroid fitting curve, and the boundary line of the observation area to obtain a first predicted centroid of the lightning to ground flash at the next moment includes: Respectively determining the velocity coverage range, the intersection points of the centroid fitting curve and the boundary line of the observation area, and the velocity vector of each intersection point; A reference velocity vector of the lightning to ground flash is determined based on the centroid point at the first moment of the lightning to ground flash and the centroid point at the current moment, and the intersection point at which the angle between the velocity vector and the reference velocity vector is the smallest among each intersection point is used as the first predicted centroid point of the lightning to ground flash at the next moment.
6. The lightning ground lightning trajectory prediction method according to claim 1, characterized in that: The method of performing position prediction on each centroid point based on the neural network model to obtain a second prediction result includes: Determine a centroid prediction model, wherein the centroid prediction model is obtained by training with a preset neural network model; Determine the longitude and latitude coordinates of each centroid point, and input each data into the centroid point prediction model to obtain a second prediction result output by the centroid point prediction model.
7. The lightning ground lightning trajectory prediction method according to claim 1, characterized in that: The determining, according to the second prediction result, a second predicted centroid point of the lightning to ground flash at the next moment comprises: Determine whether the latitude and longitude coordinates in the second prediction result are within the observation area; If yes, the second prediction result is used as the second predicted centroid point of the lightning to ground flash at the next moment; If not, construct a target auxiliary line of the second prediction result in the observation area, determine at least one intersection of the target auxiliary line and the boundary line of the observation area, and correct the longitude and latitude coordinates of the second prediction result according to the number of intersections to obtain the second predicted centroid point of the lightning flash at the next moment.
8. The lightning ground lightning trajectory prediction method according to claim 7, characterized in that: The step of correcting the latitude and longitude coordinates of the second prediction result according to the number of intersection points to obtain a second predicted centroid point of the lightning to ground flash at the next moment includes: When the number of intersection points is one, the intersection point is used as the second predicted centroid point of the lightning to ground flash at the next moment; When there are two intersection points, the velocity vectors of the two intersection points are determined, and the intersection line segment between the two intersection points is connected, and the second predicted centroid point of the lightning to ground flash at the next moment is determined on the intersection line segment according to the velocity vectors of the two intersection points.
9. A lightning ground lightning trajectory prediction device, characterized in that: include: A centroid determination module is used to obtain observation data of lightning to ground flashes in an observation area, and determine the centroid of the lightning to ground flashes at multiple consecutive historical moments according to the observation data; A first prediction module is used to perform interpolation fitting prediction on each centroid point by using an interpolation fitting algorithm to obtain a first prediction result, and determine a first predicted centroid point of the lightning to ground flash at the next moment according to the first prediction result; A second prediction module is used to predict the position of each centroid point based on the neural network model to obtain a second prediction result, and determine a second predicted centroid point of the lightning to ground flash at the next moment according to the second prediction result; The trajectory determination module is used to obtain the midpoint of the first predicted centroid point and the second predicted centroid point to obtain a target predicted centroid point, and determine the target predicted trajectory of the lightning ground flash based on the target predicted centroid point and the centroid points of each historical moment.
10. A storage medium, characterized in that: The storage medium stores computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the lightning ground lightning trajectory prediction method as described in any one of claims 1 to 8.
11. A computer device, characterized in that: include: one or more processors, and memory; The memory stores computer-readable instructions, and when the computer-readable instructions are executed by the one or more processors, the steps of the lightning ground lightning trajectory prediction method according to any one of claims 1 to 8 are performed.
Citation Information
Patent Citations
Thunder and lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network
CN112668790A
Lightning prediction method and system based on lightning monitoring data information
CN116776170A
Lightning strike prediction apparatus, notification system for lightning strike prediction result, and program
JP2005274321A
