CN111290053A - Thunderstorm path prediction method based on Kalman filtering - Google Patents

Thunderstorm path prediction method based on Kalman filtering Download PDF

Info

Publication number
CN111290053A
CN111290053A CN202010132885.6A CN202010132885A CN111290053A CN 111290053 A CN111290053 A CN 111290053A CN 202010132885 A CN202010132885 A CN 202010132885A CN 111290053 A CN111290053 A CN 111290053A
Authority
CN
China
Prior art keywords
thunderstorm
thunderstorms
lightning
time period
kalman filtering
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202010132885.6A
Other languages
Chinese (zh)
Other versions
CN111290053B (en
Inventor
汤德佑
陈靖宇
伍光胜
胡鹏
奚建清
张平健
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
South China University of Technology SCUT
Original Assignee
South China University of Technology SCUT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by South China University of Technology SCUT filed Critical South China University of Technology SCUT
Priority to CN202010132885.6A priority Critical patent/CN111290053B/en
Publication of CN111290053A publication Critical patent/CN111290053A/en
Application granted granted Critical
Publication of CN111290053B publication Critical patent/CN111290053B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01W—METEOROLOGY
    • G01W1/00—Meteorology
    • G01W1/10—Devices for predicting weather conditions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Abstract

The invention discloses a thunderstorm path prediction method based on Kalman filtering, which updates and refines the moving path of the thunderstorm in a continuous time period sequence in an iteration mode and comprises the following steps: step 01, performing thunderstorm initial identification on lightning data by using a clustering analysis method; step 02, forecasting the thunderstorm state according to Kalman filtering; step 03, tracking the thunderstorm of the adjacent time period based on the state information of the thunderstorm; step 04, updating the thunderstorm state according to Kalman filtering; and step 05, extrapolating to obtain the thunderstorm predicted path. According to the method, a Kalman filtering model is established by observing lightning observation data in time and space sequences, the noise influence of the observation data is reduced by using a two-step program of prediction and update of a Kalman filtering technology, and the thunderstorm path prediction result is improved through continuous iteration, so that the precision is improved.

Description

Thunderstorm path prediction method based on Kalman filtering
Technical Field
The invention relates to the field of meteorological lightning protection, in particular to a thunderstorm path prediction method based on Kalman filtering.
Background
Lightning is one of the most serious natural disasters in the world, not only causes social resource and economic loss, but also threatens the life safety of human beings. In the face of lightning disasters, how to effectively forecast the arrival of thunderstorms and improve the prediction accuracy rate becomes the key point of lightning early warning.
The method for carrying out lightning early warning is more, thunderstorm path prediction is one of the main methods, and comprises identification and tracking of thunderstorms, historical analysis of the thunderstorms, possible future position prediction of the thunderstorms and the like, and possible future positions of the thunderstorms in a plurality of time periods are obtained through a prediction algorithm to form a general thunderstorm moving path.
Lightning data is a stream object with the characteristic of being significantly time-varying in thunderstorm path prediction. The lightning data in China has the characteristics of high accuracy, wide detection range and the like, contains lightning information such as lightning occurrence time, the longitude and latitude positions of lightning strokes and the like, and can effectively help to predict paths of thunderstorms.
Kalman filtering (Kalman Filter) is a high-efficiency state filtering technology for a linear system, is one of the most widely applied filtering methods, and has good application in a plurality of fields such as communication, navigation, weather forecast and the like. Kalman filtering can analyze data containing noise from a time sequence, estimate the state change of the system, and reduce errors by using a two-step program of prediction and updating.
Huang etiquette et al in "LLS-based thunderstorm movement trend approach prediction" indicate that the prior art can predict the thunderstorm cloud center within 10 to 15 minutes in the future, the prediction deviation degree is low, the accuracy rate is high, but the prediction accuracy rate is greatly reduced along with the increase of the time needing prediction. The thunderstorm prediction using Kalman filtering provided by the invention can smoothly process the movement route of the thunderstorm, so that the accuracy rate of long-time prediction is reduced and the amplitude is reduced.
Disclosure of Invention
In order to overcome the defects that the traditional thunderstorm prediction algorithm is not flexible enough and accurate enough, the invention provides a method for processing lightning data to realize thunderstorm path prediction based on Kalman filtering. The method is realized based on a Kalman filtering technology, a Kalman filtering model for the thunderstorm is established by analyzing lightning data and combining algorithms of thunderstorm identification and thunderstorm tracking, and the state matrix of the thunderstorm is subjected to filtering smoothing treatment, so that the influence of noise can be effectively reduced, and the state change of the estimation system is optimized; by means of continuous iteration in the time sequence, the moving route of the thunderstorm can be continuously tracked, and the prediction result of the thunderstorm path is gradually accurate.
The invention is realized by at least one of the following technical schemes.
A thunderstorm path prediction method based on Kalman filtering comprises the following steps:
step 1, identifying the thunderstorm at first;
step 2, predicting the thunderstorm state according to Kalman filtering;
step 3, tracing thunderstorm;
step 4, updating the thunderstorm state according to Kalman filtering;
and 5, extrapolating to obtain the thunderstorm predicted path.
Further, the thunderstorm initial identification in step 1 comprises the following steps:
step 1.1, in the time sequence of lightning data, detecting a time period of first occurrence of lightning, and acquiring tnLightning data for a time period;
step 1.2, meshing lightning coordinates;
step 1.3, scanning grids meeting the lightning frequency threshold and the thunderstorm area threshold and combining the grids;
step 1.4, recording grids meeting the threshold value requirement as a lightning cluster area;
step 1.5, combining adjacent lightning cluster areas and numbering;
step 1.6, scanning all lightning cluster areas;
step 1.7, enveloping the lightning cluster by using an ellipse method, and returning the obtained thunderstorm which is initially identified;
after the step 1 is finished, judging whether a thunderstorm List exists, and if the List exists, namely the thunderstorm information obtained in the previous time period exists, entering a step 2; and if the List does not exist, adding the originally identified thunderstorm into the List, and directly entering next iteration.
Further, the step 2 of predicting the thunderstorm state according to the kalman filter comprises the following steps:
step 2.1, by traversing t in Listn-1All identified thunderstorms in the time period, acquiring state information such as position coordinates, speed vectors and the like of the thunderstorms i, and initializing a state matrix Xn-1Sum covariance matrix Pn-1；
Step 2.2, updating an equation according to Kalman time:
Figure BDA0002396274960000021
Figure BDA0002396274960000022
calculating the time period t of the thunderstorm inThe predicted value comprises a state prediction matrix Xn -Sum covariance prediction matrix Pn -Where F is the state transition matrix and Q is the state transition noise matrix
Figure BDA0002396274960000023
Figure BDA0002396274960000031
Wherein r is the central noise intensity of the measured thunderstorm; t is the interval of time period, the unit is minute, and t is generally more than or equal to 5 and less than or equal to 10.
Further, the thunderstorm tracking in step 3 is forSeveral successive time periods t1，t2，……，tnThe method for continuously tracking the thunderstorm comprises the following steps:
step 3.1, thunderstorm matching;
and 3.2, dividing and combining the thunderstorms.
Further, the thunderstorm matching of step 3.1 comprises the following steps:
step 3.1.1, obtain at t from the List of thunderstormsn-1N of the time period1Individual thunderstorms, t is obtained in the initial thunderstorm identification stepnTime period identified N2Taking the thunderstorms of two adjacent time periods as two sets A and B in the bipartite graph;
step 3.1.2, calculating cost function C corresponding to two thunderstorm setsijTaking the negative number as the weight value on the edge between the thunderstorm bipartite graphs;
and 3.1.3, aiming at the thunderstorm bipartite graph with the weight value, finding a matching scheme capable of enabling a cost function by using the minimum weight matching of the bipartite graph.
Further, the thunderstorm split merge process described in step 3.2 comprises the following steps:
step 3.2.1, time period tn-1And tnDividing the thunderstorms into four sets of matched sets and unmatched sets respectively;
step 3.2.2, combining the predicted value of step 2, according to tn-1Whether the predicted value of the unmatched thunderstorm is tnJudging whether the matched thunderstorms are overlapped, and if the matched thunderstorms are overlapped, judging that the thunderstorms are combined; if the matching thunderstorms are not overlapped, taking the unmatched thunderstorms as residual thunderstorms;
step 3.2.3, combining the predicted value of step 2, according to tn-1Whether the predicted value of the matched thunderstorm is equal to tnJudging whether unmatched thunderstorms are overlapped, and if the unmatched thunderstorms are overlapped, judging that thunderstorm splitting occurs; if the matching thunderstorms are not overlapped, taking the unmatched thunderstorms as residual thunderstorms;
step 3.2.4, analyze and process the rest thunderstorms, and combine tn-1Unmatched thunderstorms are treated as new thunderstorms, and t isnAn unmatched thunderstorm is considered a deceased thunderstorm,and outputting the result of thunderstorm tracking as an observed value to form an observed value list.
Further, the updating the thunderstorm state according to the kalman filter in step 4 comprises the following steps:
step 4.1, traverse the observation value list, each observation value in the list is Yn；
Step 4.2, if the observed value YnInstead of combining thunderstorms, the equations are updated according to kalman states:
Figure BDA0002396274960000041
Figure BDA0002396274960000042
Figure BDA0002396274960000043
in combination with Xn -And Pn -Calculating the Kalman gain KnAnd update XnAnd PnWherein X isnIs an updated a posteriori state matrix, PnIs the updated posterior covariance matrix, H is the observation model, R is the observation noise matrix, I is the identity matrix
Figure BDA0002396274960000044
Figure BDA0002396274960000045
If the observed value Y isnGenerated by combining thunderstorms, combining the velocity vectors of the combined thunderstorms, and updating Xn；
And 4.3, updating the thunderstorm List after all the observation values are updated.
Further, the step 5 of extrapolating to obtain the thunderstorm prediction path is to extrapolate a plurality of thunderstorm states in m time periods after Kalman filtering is carried out, and the states are divided intoIs given by tn+1、tn+2、……、tn+mSo as to draw a thunderstorm prediction path; extrapolation, updating and refining the predicted path is performed at the end of each iteration.
Compared with the prior art, the invention has the beneficial effects that: the accuracy reduction range of the prediction algorithm is only within 20% when the thunderstorm is changed from 10-15 minutes of the prediction short time to about 60 minutes of the prediction long time.
Drawings
FIG. 1 is a flowchart of a thunderstorm path prediction algorithm for performing an iteration according to the present embodiment;
fig. 2 is a flow chart of the initial thunderstorm identification in the present embodiment;
FIG. 3 is a flow chart of the thunderstorm tracking according to the present embodiment;
fig. 4 is a schematic diagram of the thunderstorm path extrapolation of the embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
In this embodiment, a thunderstorm path prediction method based on kalman filtering is set in the nth iteration with a time period of tnStoring the List of thunderstorms t a period of time before Kalman updaten-1And after the Kalman update, the List stores the thunderstorm in the current time period.
Fig. 1 is a flow of a thunderstorm path prediction algorithm for completing one iteration, which includes the following steps:
step 1, identifying the thunderstorm at first; thunderstorm initial identification is carried out through lightning data obtained from a lightning positioning system, and t is carried outnInitial identification of the thunderstorm. If t is presentn-1Go to step 2; otherwise, generating a List by initially identifying the thunderstorm, and entering the next time period;
FIG. 2 is a flow chart of the initial thunderstorm identification of this embodiment to obtain tnThe method comprises the following steps of carrying out a series of processing by taking the initial identification thunderstorm observed value of the time period as a target:
step 1.1, in flashIn the time sequence of the electrical data, detecting the time period of the first occurrence of lightning and acquiring tnLightning data for a time period;
step 1.2, defining a plane rectangular coordinate system, taking longitude and latitude as an X axis and a Y axis of the plane coordinate system, taking the longitude and the latitude from the west to the east as the positive direction of the X axis, taking the longitude and the latitude from the south to the north as the positive direction of the Y axis, and importing lightning data into a gridded coordinate system;
and 1.3, scanning the grid, and setting the lightning quantity scanned in the grid i as i.count.
Step 1.4, defining a lightning frequency threshold value a, and if i.count is larger than or equal to a, recording a grid i as a lightning cluster area; if i.count < a, grid i is not recorded. Judging whether all the grids are scanned or not, if not, turning to the step 1.3 to continuously scan the next grid; if the scanning is completed, entering the step 1.5;
step 1.5, combining adjacent lightning cluster areas and numbering;
and 1.6, scanning all lightning cluster areas, and setting the number of the grids occupied in the area j to be j.
Step 1.7, defining the thunderstorm area threshold b if j<b, not recording the area j; if j.count is greater than or equal to b, an entity observed value having attributes such as area and center point is formed by enveloping the lightning cluster in the area j by an ellipsometry according to Principal Component Analysis (PAC). The lightning cluster is a list of a group of lightning coordinates, and one lightning cluster is provided with n lightning rays which can be regarded as a matrix formed by a set
Figure BDA0002396274960000051
Where x, y are the longitude and latitude coordinates of the lightning, respectively.
The principal component directions of lightning distribution on a plane coordinate system in the same lightning cluster are analyzed by using principal component analysis (PAC), which is specifically described as follows:
firstly, the coordinate in each lightning cluster is averaged to obtain the central point of an ellipse
Figure BDA0002396274960000052
And calculating the cooperator of each lightning clusterA difference matrix. Deriving two eigenvalues (λ) from the covariance matrix1,λ2) And two feature vectors corresponding thereto
Figure BDA0002396274960000053
Due to the maximum eigenvalue λmax＝max(λ1,λ2) The corresponding feature vector is the direction pointing to the maximum data variance, namely the principal component direction, and can be taken as the major axis direction of the ellipse; and the other feature vector perpendicular thereto is taken as the direction of the minor axis. And (4) taking an included angle between the long axis direction and the plane coordinate system to obtain the offset angle of the ellipse. The eigenvalue is equal to the variance of the data in the corresponding dimension on the coordinates after the plane coordinate system is rotated to the direction of the eigenvector, and in order to include all lightning rays in the lightning cluster as much as possible, the standard deviation may be slightly enlarged as the lengths of the major axis and the minor axis of the ellipse in the plane coordinate system. By combining the center point, major axis, minor axis, and angle, an ellipse example can be obtained.
Judging whether all the areas are scanned or not, if not, turning to the step 1.6 to continue scanning the next area; if the scan is complete, the process is terminated and the resulting initially identified thunderstorm is returned.
Step 2, predicting the thunderstorm state according to Kalman filtering, comprising the following steps:
step 2.1, by traversing t in Listn-1All identified thunderstorms in the time period, acquiring state information such as position coordinates, speed vectors and the like of the thunderstorms i, and initializing a state matrix Xn-1Sum covariance matrix Pn-1；
Step 2.2, updating an equation according to Kalman time:
Figure BDA0002396274960000061
Figure BDA0002396274960000062
calculating the time period t of the thunderstorm inThe predicted value comprises a state prediction matrix Xn -Sum covariance prediction matrix Pn -Where F is the state transition matrix and Q is the state transition noise matrix
Figure BDA0002396274960000063
Figure BDA0002396274960000064
Wherein r is the central noise intensity of the measured thunderstorm; t is the interval of time period, the unit is minute, and t is generally more than or equal to 5 and less than or equal to 10.
Step 03, as shown in fig. 3, tracking the mines in the adjacent time periods based on the state information of the thunderstorms includes the following steps:
step 3.1, thunderstorm matching, which specifically comprises the following steps:
step 3.1.1, obtain at t from the List of thunderstormsn-1N of the time period1Individual thunderstorms, t is obtained in the initial thunderstorm identification stepnTime period identified N2And taking the thunderstorms of two adjacent time periods as two sets A and B in the bipartite graph. The state of thunderstorm i in A is
Figure BDA0002396274960000065
The state of thunderstorm j in B is
Figure BDA0002396274960000066
Wherein 0<i≤N1，0<j≤N2，
Figure BDA0002396274960000067
And
Figure BDA0002396274960000068
longitude and latitude coordinates of the thunderstorm center point respectively, and S is the area of the thunderstorm;
step 3.1.2, calculating a cost function C of each thunderstorm i in A corresponding to each thunderstorm j in B according to the state of the thunderstormsijTaking the negative number as the weight on the edge between the bipartite graphs, and let
Cij＝dp+ds
Wherein:
Figure BDA0002396274960000069
Figure BDA00023962749600000610
dpis the difference in location of the center point of the thunderstorm; dsIs the difference of the square root of the thunderstorm area.
Step 3.1.3, aiming at the thunderstorm bipartite graph with the weight value, finding the sum sigma C which can enable the cost function by using the method idea of matching the minimum weight of the bipartite graphijA minimum matching scheme;
the matching method for the minimum weight of the bipartite graph specifically comprises the following steps:
(1) and preprocessing the weight matrix. The weight matrix is arranged in a row set A and a column set B, if the lengths of A and B are not equal, the smaller one is expanded to make the rows and the columns equal, and the expanded row and column weight is set to be minus infinity. Dividing the thunderstorms of two adjacent time periods into two sets, calculating the measurement of the position difference and the area difference between each thunderstorm in the set A and each thunderstorm in the set B, and calculating the negative number-C of a cost functionijAnd substituting the weight value into a weight value matrix. (ii) a
(2) The top mark value of the preprocessing set A is the maximum value of the weight value of the corresponding edge of each element, and the top mark value of the set B is 0. Preprocessing a matching matrix of the set A and the set B, wherein the matching matrix is used for indicating that each element in the set is matched with the element number of the other set;
(3) depth First Search (DFS) augmentation is performed on each thunderstorm in set a, finding matching edges such that d equals 0, where d equals min (a)iTop mark + BjTop mark-Cij) Wherein A isi、BjRepresenting the ith and jth thunderstorms in traversal set A, B, respectively. And if the thunderstorm which matches the conflict is met (namely, the augmentation path cannot be found), modifying the top mark of the corresponding thunderstorm, and then continuing to perform DFS augmentation on the modified thunderstorm. Let the conflicted thunderstorms be A in A respectively1、A2And B of B1The modification is that A is1Top marks of-d, A2Top marks of-d, B1Top mark + d;
(4) and (4) all the thunderstorms in the set A are traversed, and the obtained result is the matching scheme with the minimum edge weight.
If unmatched thunderstorms still exist after the step 3.1, turning to a step 3.2; otherwise, ending the task.
Step 3.2, carrying out thunderstorm splitting and merging treatment, specifically comprising the following steps:
step 3.2.1, after matching, the time period tn-1And tnAre divided into two sets. Time period tnThe matched thunderstorm is Mn-1Unmatched thunderstorm is Un-1(ii) a Time period tnThe matched thunderstorm is MnUnmatched thunderstorm is Un. The objects to be processed are respectively Un-1And Un；
Step 3.2.2 for tn-1Any unmatched thunderstorm Un-1Predicting the thunderstorm at t by using the step 2nStatus values of time periods. If the predicted thunderstorm center point is at any matched tnThunderstorm MnIf it is, then the U is consideredn-1Through combination, M is formedn；
Step 3.2.3 for tnAny unmatched thunderstorm MnPredicting all t using step 2n-1Matched thunderstorm Mn-1The state value of (2). If the predicted center point is at any unmatched tnThunderstorm UnIn, then consider Mn-1Through splitting generate Un；
And 3.2.4, analyzing and processing the residual thunderstorms. If U is presentn-1If the combination is not carried out, the thunderstorm is considered to be died, and the thunderstorm is abandoned; if U is presentnInstead of splitting, the thunderstorm is considered to be a new born thunderstorm.
And finally, ending the task and returning a matching result as an observed value.
Step 04, updating the thunderstorm state according to the Kalman filtering comprises the following steps:
step 4.1, traverse the observation value list, each observation value in the list is Yn；
Step 4.2, if the observed value YnInstead of combining thunderstorms, the equations are updated according to kalman states:
Figure BDA0002396274960000081
Figure BDA0002396274960000082
Figure BDA0002396274960000083
in combination with Xn -And Pn -Calculating the Kalman gain KnAnd update XnAnd PnWherein X isnIs an updated a posteriori state matrix, PnIs the updated posterior covariance matrix, H is the observation model, R is the observation noise matrix, I is the identity matrix
Figure BDA0002396274960000084
Figure BDA0002396274960000085
If the observed value Y isnGenerated by combining thunderstorms, combining the velocity vectors of the combined thunderstorms, and updating Xn；
And 4.3, updating the thunderstorm List after all the observation values are updated.
Step 05, extrapolating to obtain a thunderstorm predicted path, and entering the next time period; if the lightning exists in the next time period, entering the next iteration; otherwise, ending the task.
The thunderstorm path extrapolation diagram shown in FIG. 4 illustrates t for the first detected lightning0Time period to t3An extrapolated path of the time segment. Due to t0The time period is the first time lightning is detected, the trend of a thunderstorm cannot be predicted yet, and therefore no extrapolation path exists. At t1In, extrapolating to obtain t2、t3、……、tm+1Respectively X, are predicted states of thunderstorms2 -、X3 -、……、Xm+1 -And the m predicted thunderstorms can be connected to form a thunderstorm path. At t2After iteration of the algorithm, t1All at t2Is updated to obtain X3 -To Xm+2 -The predicted thunderstorm and the updated extrapolation path; in the same way at t3In update t2The updated extrapolation path is obtained by extrapolation prediction; and then, carrying out iterative extrapolation until the thunderstorm disappears. The dashed lines in the figure represent updates of the predicted thunderstorm state.
The method can be applied to weather forecast prediction equipment, and the result is output by receiving the data of the lightning positioning system, so that the purpose of thunderstorm early warning is achieved.
The above description is not intended to limit the present invention, and the present invention is not limited to the above examples, and any other changes, modifications, substitutions and alterations without departing from the spirit and principles of the present invention are also encompassed in the scope of the present invention.

Claims (8)

1. A thunderstorm path prediction method based on Kalman filtering is characterized by comprising the following steps:
step 1, identifying the thunderstorm at first;
step 2, predicting the thunderstorm state according to Kalman filtering;
step 3, tracing thunderstorm;
step 4, updating the thunderstorm state according to Kalman filtering;
and 5, extrapolating to obtain the thunderstorm predicted path.
2. The method according to claim 1, wherein the initial thunderstorm identification in step 1 comprises the following steps:
step 1.1, in the time sequence of lightning data, detecting a time period of first occurrence of lightning, and acquiring tnLightning data for a time period;
step 1.2, meshing lightning coordinates;
step 1.3, scanning grids meeting the lightning frequency threshold and the thunderstorm area threshold and combining the grids;
step 1.4, recording grids meeting the threshold value requirement as a lightning cluster area;
step 1.5, combining adjacent lightning cluster areas and numbering;
step 1.6, scanning all lightning cluster areas;
step 1.7, enveloping the lightning cluster by using an ellipse method, and returning the obtained thunderstorm which is initially identified;
after the step 1 is finished, judging whether a thunderstorm List exists, and if the List exists, namely the thunderstorm information obtained in the previous time period exists, entering a step 2; and if the List does not exist, adding the originally identified thunderstorm into the List, and directly entering next iteration.
3. The method of claim 1, wherein the step 2 of predicting the thunderstorm state according to the kalman filter comprises the steps of:
step 2.1, by traversing t in Listn-1All identified thunderstorms in the time period, acquiring state information such as position coordinates, speed vectors and the like of the thunderstorms i, and initializing a state matrix Xn-1Sum covariance matrix Pn-1；
Step 2.2, updating an equation according to Kalman time:
Figure FDA0002396274950000011
Figure FDA0002396274950000012
calculating the time period t of the thunderstorm inThe predicted value comprises a state prediction matrix Xn -Sum covariance prediction matrix Pn -Where F is the state transition matrix and Q is the state transition noise matrix
Figure FDA0002396274950000013
Figure FDA0002396274950000021
Wherein r is the central noise intensity of the measured thunderstorm; t is the interval of time period, the unit is minute, and t is generally more than or equal to 5 and less than or equal to 10.
4. The method according to claim 1, wherein the thunderstorm tracking in step 3 is performed for several consecutive time periods t1，t2，……，tnThe method for continuously tracking the thunderstorm comprises the following steps:
step 3.1, thunderstorm matching;
and 3.2, dividing and combining the thunderstorms.
5. The method according to claim 4, wherein the thunderstorm path prediction based on Kalman filtering in step 3.1 comprises the following steps:
step 3.1.1, obtain at t from the List of thunderstormsn-1N of the time period1Individual thunderstorms, t is obtained in the initial thunderstorm identification stepnTime period identified N2Taking the thunderstorms of two adjacent time periods as two sets A and B in the bipartite graph;
step 3.1.2, calculating cost function C corresponding to two thunderstorm setsijTaking the negative number as the weight value on the edge between the thunderstorm bipartite graphs;
and 3.1.3, aiming at the thunderstorm bipartite graph with the weight value, finding a matching scheme capable of enabling a cost function by using the minimum weight matching of the bipartite graph.
6. The method according to claim 4, wherein the thunderstorm path prediction based on Kalman filtering in step 3.2 comprises the following steps:
step 3.2.1, time period tn-1And tnDividing the thunderstorms into four sets of matched sets and unmatched sets respectively;
step 3.2.2, combining the predicted value of step 2, according to tn-1Whether the predicted value of the unmatched thunderstorm is tnJudging whether the matched thunderstorms are overlapped, and if the matched thunderstorms are overlapped, judging that the thunderstorms are combined; if the matching thunderstorms are not overlapped, taking the unmatched thunderstorms as residual thunderstorms;
step 3.2.3, combining the predicted value of step 2, according to tn-1Whether the predicted value of the matched thunderstorm is equal to tnJudging whether unmatched thunderstorms are overlapped, and if the unmatched thunderstorms are overlapped, judging that thunderstorm splitting occurs; if the matching thunderstorms are not overlapped, taking the unmatched thunderstorms as residual thunderstorms;
step 3.2.4, analyze and process the rest thunderstorms, and combine tn-1Unmatched thunderstorms are treated as new thunderstorms, and t isnAnd the unmatched thunderstorms are regarded as extinct thunderstorms, and the result of thunderstorm tracking is output as an observed value to form an observed value list.
7. The method according to claim 1, wherein the updating the thunderstorm state according to kalman filtering in step 4 comprises the following steps:
step 4.1, traverse the observation value list, each observation value in the list is Yn；
Step 4.2, if the observed value YnInstead of combining thunderstorms, the equations are updated according to kalman states:
Figure FDA0002396274950000031
Figure FDA0002396274950000032
Figure FDA0002396274950000033
in combination with Xn -And Pn -Calculating the Kalman gain KnAnd update XnAnd PnWherein X isnIs an updated a posteriori state matrix, PnIs the updated posterior covariance matrix, H is the observation model, R is the observation noise matrix, I is the identity matrix
Figure FDA0002396274950000034
Figure FDA0002396274950000035
If the observed value Y isnGenerated by combining thunderstorms, combining the velocity vectors of the combined thunderstorms, and updating Xn；
And 4.3, updating the thunderstorm List after all the observation values are updated.
8. The method according to claim 1, wherein the thunderstorm path prediction based on kalman filtering is implemented by extrapolating the thunderstorm path prediction in step 5 by extrapolating a plurality of thunderstorm states in m time periods, which are t respectivelyn+1、tn+2、……、tn+mSo as to draw a thunderstorm prediction path; extrapolation, updating and refining the predicted path is performed at the end of each iteration.
CN202010132885.6A 2020-02-29 2020-02-29 Thunderstorm path prediction method based on Kalman filtering Active CN111290053B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010132885.6A CN111290053B (en) 2020-02-29 2020-02-29 Thunderstorm path prediction method based on Kalman filtering

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010132885.6A CN111290053B (en) 2020-02-29 2020-02-29 Thunderstorm path prediction method based on Kalman filtering

Publications (2)

Publication Number Publication Date
CN111290053A true CN111290053A (en) 2020-06-16
CN111290053B CN111290053B (en) 2021-12-17

Family

ID=71024134

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010132885.6A Active CN111290053B (en) 2020-02-29 2020-02-29 Thunderstorm path prediction method based on Kalman filtering

Country Status (1)

Country Link
CN (1) CN111290053B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112668790A (en) * 2020-12-30 2021-04-16 南京信息工程大学 Thunder and lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network
CN113466567A (en) * 2021-06-08 2021-10-01 广东电网有限责任公司广州供电局 Lightning activity prediction method based on Kalman filter
CN115079306A (en) * 2022-05-23 2022-09-20 远景智能国际私人投资有限公司 Lightning forecasting method, device, equipment, storage medium and computer program product

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102662173A (en) * 2012-04-24 2012-09-12 上海交通大学 Thunderstorm forecasting method based on level set
CN102692656A (en) * 2012-06-08 2012-09-26 河源职业技术学院 Thunderstorm data processing method and system
CN103337133A (en) * 2013-06-14 2013-10-02 广东电网公司中山供电局 System and method for power grid thunderstorm disaster early warning based on recognition and forecast
US20180031737A1 (en) * 2016-07-29 2018-02-01 Vaisala, Inc. Short-term thunderstorm forecast and severe weather alert system and method
CN108520023A (en) * 2018-03-22 2018-09-11 合肥佳讯科技有限公司 A kind of identification of thunderstorm core and method for tracing based on Hybrid Clustering Algorithm
CN109754124A (en) * 2019-01-14 2019-05-14 兰州大学 A kind of lightning Application in Potential Prediction method

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102662173A (en) * 2012-04-24 2012-09-12 上海交通大学 Thunderstorm forecasting method based on level set
CN102692656A (en) * 2012-06-08 2012-09-26 河源职业技术学院 Thunderstorm data processing method and system
CN103337133A (en) * 2013-06-14 2013-10-02 广东电网公司中山供电局 System and method for power grid thunderstorm disaster early warning based on recognition and forecast
US20180031737A1 (en) * 2016-07-29 2018-02-01 Vaisala, Inc. Short-term thunderstorm forecast and severe weather alert system and method
CN108520023A (en) * 2018-03-22 2018-09-11 合肥佳讯科技有限公司 A kind of identification of thunderstorm core and method for tracing based on Hybrid Clustering Algorithm
CN109754124A (en) * 2019-01-14 2019-05-14 兰州大学 A kind of lightning Application in Potential Prediction method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
周康辉 郑永光 蓝渝: "新型聚类算法在雷暴单体临近预报中的应用", 《第32届中国气象学会年会》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112668790A (en) * 2020-12-30 2021-04-16 南京信息工程大学 Thunder and lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network
CN112668790B (en) * 2020-12-30 2023-07-25 南京信息工程大学 Lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network
CN113466567A (en) * 2021-06-08 2021-10-01 广东电网有限责任公司广州供电局 Lightning activity prediction method based on Kalman filter
CN115079306A (en) * 2022-05-23 2022-09-20 远景智能国际私人投资有限公司 Lightning forecasting method, device, equipment, storage medium and computer program product
CN115079306B (en) * 2022-05-23 2023-12-05 远景智能国际私人投资有限公司 Lightning prediction method, device, apparatus, storage medium and computer program product

Also Published As

Publication number Publication date
CN111290053B (en) 2021-12-17

Similar Documents

Publication Publication Date Title
CN111290053B (en) Thunderstorm path prediction method based on Kalman filtering
US11488069B2 (en) Method for predicting air quality with aid of machine learning models
CN107103754B (en) Road traffic condition prediction method and system
CN112526513B (en) Millimeter wave radar environment map construction method and device based on clustering algorithm
CN112230243B (en) Indoor map construction method for mobile robot
CN107194408B (en) Target tracking method of mixed block sparse cooperation model
CN110119438B (en) Airborne LiDAR point cloud filtering method based on active learning
CN111209457B (en) Target typical activity pattern deviation warning method
CN111814597A (en) Urban function partitioning method coupling multi-label classification network and YOLO
CN108426582B (en) Indoor three-dimensional map matching method for pedestrians
CN111722214A (en) Radar multi-target tracking PHD implementation method
CN114564545A (en) System and method for extracting ship experience course based on AIS historical data
Han et al. A matching algorithm based on the nonlinear filter and similarity transformation for gravity-aided underwater navigation
CN111259332B (en) Fuzzy data association method and multi-target tracking method in clutter environment
Chen et al. Proactive quality control: Observing system simulation experiments with the Lorenz’96 model
CN114216434A (en) Target confirmation method, system, equipment and storage medium for mobile measurement and control station
CN108415105B (en) Method for inspecting relative humidity value observed by ground meteorological observation station
Li et al. Exploring label probability sequence to robustly learn deep convolutional neural networks for road extraction with noisy datasets
CN111681313B (en) Space vision analysis method based on digital topography and electronic equipment
CN111121778B (en) Navigation system initialization method
CN104684083B (en) A kind of AP systems of selection based on sub-clustering thought
CN109375160B (en) Angle measurement error estimation method in pure-azimuth passive positioning
CN113887590B (en) Target typical track and area analysis method
Heyns et al. Terrain visibility-dependent facility location through fast dynamic step-distance viewshed estimation within a raster environment
CN111896002A (en) Method and system for online planning and adaptability analysis of terrain-assisted navigation matching area

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant