CN112668790A - Thunder and lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network - Google Patents

Thunder and lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network Download PDF

Info

Publication number
CN112668790A
CN112668790A CN202011617345.3A CN202011617345A CN112668790A CN 112668790 A CN112668790 A CN 112668790A CN 202011617345 A CN202011617345 A CN 202011617345A CN 112668790 A CN112668790 A CN 112668790A
Authority
CN
China
Prior art keywords
lightning
neural network
center
lstm neural
time
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
CN202011617345.3A
Other languages
Chinese (zh)
Other versions
CN112668790B (en
Inventor
孟宇航
付章杰
孙星明
孟若涵
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.)
Nanjing University of Information Science and Technology
Original Assignee
Nanjing University of Information Science and Technology
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 Nanjing University of Information Science and Technology filed Critical Nanjing University of Information Science and Technology
Priority to CN202011617345.3A priority Critical patent/CN112668790B/en
Publication of CN112668790A publication Critical patent/CN112668790A/en
Application granted granted Critical
Publication of CN112668790B publication Critical patent/CN112668790B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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 thunder and lightning prediction method based on a space-time sequence clustering algorithm and an LSTM neural network, which belongs to the technical field of computer science. The invention can automatically calculate the clustering radius of the density clustering DBSCAN, the prediction error of the LSTM neural network on the longitude and latitude of the thunder center is small, the precision is high, and the actual thunder prediction requirement can be basically met. The invention firstly tries to solve the lightning prediction problem by using the LSTM neural network, and the prior method generally uses polynomial fitting or other fitting methods, so that the complicated process of lightning center movement is not completely simulated.

Description

Thunder and lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network
Technical Field
The invention belongs to the technical field of computer science, and particularly relates to a thunder and lightning prediction method based on a space-time sequence clustering algorithm and an LSTM neural network.
Background
Lightning disasters are one of the most serious ten natural disasters. Lightning is often accompanied by gusts and rainstorms, which cause damage to the distribution system, communication equipment, and residential household appliances, and even more to human and animal casualties. In the 18 th century, franklin invented lightning rods by means of the philadelphia lightning test. People never stop and go by foot for three hundred years to prevent, control and early warn of thunder and lightning. In the modern technology, lightning is monitored in real time by measuring tools such as radar, satellite, lightning locator and the like.
At present, the methods for lightning prediction include: according to the traditional thunder prediction method, a potential force prediction equation is established through historical observation data by using a related calculation method, and a thunder short-time approach prediction model and method are established. Weather Research and Forecast Model WRF (Weather Research and Forecast Model) based early warning is a Research on Weather Forecast and Weather.
However, research personnel research shows that the WRF model has less application in lightning early warning. And carrying out proximity early warning by extrapolation of various observation data, processing and modeling observation data by using a currently popular neural network algorithm, a clustering algorithm and the like, and carrying out proximity early warning by reasonable extrapolation. The method is based on the thunder and lightning forecast of a numerical model and the recent forecast based on data fusion and effective extrapolation of various high observation frequencies.
In recent years, the research on the lightning prediction problem by utilizing a neural network is common, Geng and the like propose lightNet based on ConvLstm, an encoder-decoder model and a deconvolution model are added, WRF prediction data is introduced in the aspect of data to correct the prediction data, but the method has no obvious effect on short-time lightning prediction. The Chong et al provides clustering-based lightning nowcasting, starts from research of spatial data characteristics and analysis methods, simulates and analyzes distribution characteristics of lightning location data on different planes, determines distribution characteristics of the lightning location data on longitude and latitude planes, uses a classic algorithm DBSCAN to perform clustering analysis on the lightning data, analyzes influences of input parameters on clustering results through a large number of simulations, selects appropriate parameters to obtain corresponding clustering results, and finally establishes a polynomial fitting prediction method for predicting short-term lightning through analyzing cloud cluster centroid distance, boundary point distance and lightning stroke times. The method and the method both use a DBSCAN clustering algorithm, but the method uses a large number of parameters to analyze clustering results, increases the difficulty for finding a clustering center by adopting a large number of parameter comparison modes, also reduces the accuracy for determining the clustering center, has no universality for selecting different lightning data parameters, and is very unfriendly to users without professional backgrounds.
At present, the problem of thunder prediction is solved by using a clustering algorithm and a neural network, but the clustering radius in the selected DBSCAN clustering algorithm is not automatically given and is obtained through empirical values or after a large number of clustering radii are given, so that the required clustering radius is obtained.
Disclosure of Invention
The purpose of the invention is as follows: the invention aims to provide a thunder and lightning prediction method based on a space-time sequence clustering algorithm and an LSTM neural network, which can automatically give a clustering radius value through thunder and lightning data input on a time slice without repeatedly adjusting the size of the clustering radius and can determine a thunder and lightning center more accurately, more quickly and more conveniently by combining the clustering algorithm.
The technical scheme is as follows: in order to achieve the purpose, the invention adopts the following technical scheme:
the thunder and lightning prediction method based on the space-time sequence clustering algorithm and the LSTM neural network comprises the following steps of:
1) improving the density clustering DBSCAN algorithm based on the thunder and lightning prediction problem;
2) and (4) carrying out a prediction process on the short-time lightning by using the LSTM neural network.
Further, in the step 1), time slice division is carried out on the thunder and lightning data to be predicted within a period of time, and the thunder and lightning data to be predicted within the period of time are selected; the data comprises longitude and latitude of lightning occurrence, lightning intensity, lightning gradient and lightning occurrence moment; dividing the data into time slices; it is assumed that the movement of the lightning centre is jump-like from one divided time slice to the next, i.e. the lightning centre is stationary during the divided time period.
Further, in the step 1), the Euclidean distance between all the lightning activity geographic positions on a time slice is calculated to obtain a distance set D(i)＝(d1，d2，...，dh，...)
Figure BDA0002871604010000021
Wherein i represents the number of lightning activities shared over the time slice, dhRepresenting the distance between some two lightning activities at this time slice, h representing the distance in the set D(i)Of the distance between two lightning activities at that time slice.
Further, in the step 1), a distance set D is counted(i)The distance number of the middle intervals V1[0, 0.1), V2[0.1, 0.2), V3[0.2, 0.3), V4[0.3, 0.4), V5[0.4, 0.5) is m1，m2，m3，m4，m5Performing representation and constructing statistical matrix
Figure BDA0002871604010000022
Taking the median values of the intervals V1, V2, V3, V4 and V5 as p1, p2, p3, p4 and p5, and calculating by using a formula
Figure BDA0002871604010000031
Searching clusters by the obtained Eps neighborhood value and the defined Minpts 2 and by examining the Eps neighborhood of each point in the lightning data set; if the Eps neighborhood of the point O contains more than MinPts minimum points, a cluster taking the point O as a core object is created; when no new points are added to any cluster, a cluster of lightning aggregations is determined. If q does not belong to any cluster of lightning gathers, q is called a noise point.
Further, in said step 1), the lightning activity (o) on the determined lightning cluster is determined1，o2，...，on) Geographic location o1(w1，v1)，o2(w2，v2)，......，，on(wn，vn) And by the formula
Figure BDA0002871604010000032
Calculating the lightning center k by an average value method
Figure BDA0002871604010000033
Wherein o isnRepresenting a certain lightning central activity on the cluster, wnAnd vnRepresenting the latitude and longitude of a certain lightning center activity on the cluster,
Figure BDA0002871604010000034
and
Figure BDA0002871604010000035
represents the average of all lightning center latitudes and longitudes over the cluster.
Further, in the step 2), a lightning center change data is processed, where the lightning center change data includes a start-stop time T of a time slice, a longitude L of a lightning center, a latitude W of the lightning center, an intensity ST of the lightning center (an average value of all lightning activity intensities of a cluster), a slope SL of the lightning center (an average value of all lightning activity slopes of a cluster), and a lightning stroke number Num on the time slice is counted.
Further, in the step 2), the data T, L, ST, SL and Num on different time slices of a lightning center are input into the longitudinal LSTM neural network, so as to obtain a longitude value of the lightning activity in the next time sequence; inputting all lightning center data T, W, ST, SL, Num of a lightning activity into a latitude LSTM neural network to obtain latitude values of the lightning activity in the next time sequence.
Has the advantages that: compared with the prior art, the thunder and lightning prediction method based on the space-time sequence clustering algorithm and the LSTM neural network obtains the Eps value and the thunder and lightning center of each time slice by utilizing the improved DBSCAN density clustering algorithm based on the thunder and lightning prediction according to the change of the longitude and latitude of the thunder and lightning center, and predicts the geographical position of the thunder and lightning center of the next time slice through the LSTM neural network. The invention can automatically calculate the clustering radius of the density clustering DBSCAN, the prediction error of the LSTM neural network on the longitude and latitude of the thunder center is small, the precision is high, and the actual thunder prediction requirement can be basically met. The invention firstly tries to solve the lightning prediction problem by using the LSTM neural network, and the prior method generally uses polynomial fitting or other fitting methods, so that the complicated process of lightning center movement is not completely simulated.
Drawings
FIG. 1 is a frame diagram of an improved process of a density clustering DBSCAN algorithm of the present invention;
FIG. 2 is a block diagram of the prediction process of the LSTM neural network of the present invention for lightning strike;
fig. 3 is a diagram of a change in lightning center position.
Detailed Description
The present invention will be further described with reference to the following embodiments.
The thunder and lightning prediction method based on the space-time sequence clustering algorithm and the LSTM neural network is characterized by comprising an improvement process of a density clustering DBSCAN algorithm based on a thunder and lightning prediction problem and a short-time thunder and lightning prediction process by utilizing the LSTM neural network; the specific description is as follows:
a, an improvement process of a density clustering DBSCAN algorithm based on a thunder and lightning prediction problem comprises the following steps:
a1, carrying out time slice division on the thunder and lightning data to be predicted within a period of time, and selecting the thunder and lightning data to be predicted within the period of time; the data comprises longitude and latitude of lightning occurrence, lightning intensity, lightning gradient and lightning occurrence moment; data was divided into time slices every 10 minutes.
Suppose that: the movement of the lightning centre is jump-like from one divided time slice to the next, i.e. the lightning centre is stationary during the divided time period.
Step A2, calculating Euclidean distances among all the lightning activity geographic positions on a time slice to obtain a distance set D(i)＝(d1，d2，...，dh，...)
Figure BDA0002871604010000041
Wherein i represents the number of lightning activities shared over the time slice, dhRepresenting the distance between some two lightning activities at this time slice, h representing the distance in the set D(i)Of the distance between two lightning activities at that time slice.
Step A3, counting a distance set D(i)The distance number of the middle intervals V1[0, 0.1), V2[0.1, 0.2), V3[0.2, 0.3), V4[0.3, 0.4), V5[0.4, 0.5) is m1，m2，m3，m4，m5Performing representation and constructing statistical matrix
Figure BDA0002871604010000042
Description of the drawings: and calculating the lightning centers by dividing the lightning activities with the closer geographic positions to one lightning center, wherein if the two lightning activities are more than 0.5 degree away from the geographic positions (the actual distance is about more than 50 kilometers), namely, the two lightning activities are not in the divided intervals of V1, V2, V3, V4 and V5, statistics is not needed, and the two lightning activities are not necessarily in the division of one lightning center.
Step A4, taking the median values of the intervals V1, V2, V3, V4 and V5 as p1, p2, p3, p4 and p5, and calculating by using a formula
Figure BDA0002871604010000051
Step a5, the clusters are searched by obtaining Eps neighborhood values and defining Minpts 2 and by examining the Eps neighborhood of each point in the lightning data set. If the Eps neighborhood of the point O contains more than MinPts minimum points, a cluster taking the point O as a core object is created;
step A6, when no new points are added to any cluster, a cluster of lightning aggregations is determined. If q does not belong to any cluster of thunder and lightning accumulation, the q is called a noise point;
step A7, determining lightning activity (o) on the cluster of lightning aggregates1，o2，...，on) Geographic location o1(w1，v1)，o2(w2，v2)，......，，on(wn，vn) And by the formula
Figure BDA0002871604010000052
Figure BDA0002871604010000053
Calculating the lightning center k by an average value method
Figure BDA0002871604010000054
Wherein o isnRepresenting a certain lightning central activity on the cluster, wnAnd vnRepresenting the latitude and longitude of a certain lightning center activity on the cluster,
Figure BDA0002871604010000055
and
Figure BDA0002871604010000056
represents the average of all lightning center latitudes and longitudes over the cluster.
B, a short-time lightning prediction process is carried out by using an LSTM neural network, and the steps are as follows:
step B1, processing lightning center change data, wherein the lightning center change data comprises the starting and ending time T of a time slice, the longitude L of a lightning center, the latitude W of the lightning center, the intensity ST of the lightning center (the average value of all lightning activity intensities of a cluster), the gradient SL of the lightning center (the average value of all lightning activity gradients of a cluster), and counting the lightning stroke times Num on the time slice;
step B2, inputting data T, L, ST, SL, Num on different time slices of a lightning center into a longitude LSTM neural network to obtain longitude values of the lightning activity in the next time sequence; inputting all lightning center data T, W, ST, SL, Num of a lightning activity into a latitude LSTM neural network to obtain latitude values of the lightning activity in the next time sequence.
Examples
The thunder and lightning prediction method based on the space-time sequence clustering algorithm and the LSTM neural network specifically comprises the following steps:
before calculating the lightning center of lightning activity, firstly, the time slice is divided into 10 minutes every selected lightning data which is most frequent in lightning activity and needs to be predicted within a period of time, and the data of the invention is from the provincial bureau of Hunan province. In this embodiment, the lightning data of the 6 th and 17 th 2010-th days of Hunan province are selected for analysis. After a large amount of thunder and lightning data are collected, it is known from meteorological experience that the data error obtained by the positioning mode of the three-station time difference direction finding method is small, and the data of the data positioning mode which is not the three-station time difference direction finding method is deleted. The lightning center is calculated by an improved density clustering DBSCAN method, and the method comprises the following steps:
first, lightning activities from 3:55 in the early morning are found to be relatively intensive from lightning data of 6, 17 and 2010 in Hunan province, lightning data of 10 minutes two hours after 3:55 minutes are selected for analyzing the movement trend of a lightning center, and the data are divided into 13 time slices every 10 minutes. The following lightning activity tables are 4: 15-4: 25 time slices:
Figure BDA0002871604010000061
the second step is that: calculating Euclidean distances between all the lightning activity geographic positions on a time slice to obtain a distance set Di＝(d1，d2，...，dh，...)
Figure BDA0002871604010000062
Wherein the Euclidean distance has the unit: degree; set of statistical distances DiThe distances between intervals V1[0, 0.1), V2[0.1, 0.2), V3[0.2, 0.3), V4[0.3, 0.4), V5[0.4, 0.5), and constructing a statistical matrix
Figure BDA0002871604010000063
Taking the median values of the intervals V1, V2, V3, V4 and V5 as p1, p2, p3, p4 and p5, and calculating by using a formula
Figure BDA0002871604010000064
The third step: the clusters are searched by obtaining an Eps neighborhood value and defining a Minpts of 2 and by examining the Eps neighborhood of each point in the lightning data set. If the Eps neighborhood of the point O contains more than MinPts minimum points, a cluster taking the point O as a core object is created; when no new points are added to any cluster, a cluster of lightning aggregations is determined. If q does not belong to any cluster of thunder and lightning accumulation, the q is called a noise point; lightning (o) on a determined cluster of lightning aggregations1，o2，...，on) Geographic location o1(w1，v1)，o2(w2，v2))，......，，on(wn，vn) And by formula
Figure BDA0002871604010000071
Figure BDA0002871604010000072
Calculating the lightning center k by an average value method
Figure BDA0002871604010000073
The following table shows the lightning centers of all time slices of the improved DBSCAN algorithm:
Figure BDA0002871604010000074
Figure BDA0002871604010000081
the following is a lightning center change table:
Figure BDA0002871604010000082
Figure BDA0002871604010000091
FIG. 3 is a diagram showing the change of lightning center position, and the short-term lightning is predicted by using the LSTM neural network according to the data of the change of the geographic position of the lightning center, and the specific operation steps are as follows:
the first step is as follows: processing lightning center change data in lightning centers C1, C2 and C3, wherein the lightning center change data comprise starting and ending time T of a time slice, longitude L of the lightning centers, latitude W of the lightning centers, strength ST of the lightning centers (an average value of all lightning activity strength of a cluster), gradient SL of the lightning centers (an average value of all lightning activity gradients of a cluster), and counting lightning stroke times m on the time slice;
the following is a table of lightning center C1 variations:
T L W ST SL Num
4:05～4:15 109.9085 26.6765 -34.919000 -3.550000 2
4:15～4:25 110.1515 26.75475 -38.259750 -7.325000 4
4:25～4:35 110.228667 26.748333 -50.165500 -8.200000 6
4:35～4:45 110.197000 26.734500 -12.904100 0.680000 10
4:45～4:55 110.011000 26.729714 -35.824286 -9.628571 5
4:55～5:05 110.244857 26.785857 18.799143 3.571429 7
5:05～5:15 110.308667 26.697667 18.403000 4.533333 3
5:35～5:45 110.138000 26.761000 -44.340500 -11.85000 2
5:45～5:55 110.226000 26.733750 -57.509250 -11.22500 4
inputting all lightning center data T, L, ST, SL and Num of a lightning activity into a longitude LSTM neural network to obtain longitude values of the lightning activity in the next time sequence; inputting all lightning center data T, W, ST, SL, Num of a lightning activity into a latitude LSTM neural network to obtain latitude values of the lightning activity in the next time sequence.
The above description is only a preferred embodiment of the present invention, and it will be apparent to those skilled in the art that various modifications and variations can be made without departing from the technical principles of the present invention, and these modifications and variations should also be construed as the scope of the present invention.

Claims (7)

1. The thunder and lightning prediction method based on the space-time sequence clustering algorithm and the LSTM neural network is characterized by comprising the following steps of: the method comprises the following steps:
1) improving the density clustering DBSCAN algorithm based on the thunder and lightning prediction problem;
2) and (4) carrying out a prediction process on the short-time lightning by using the LSTM neural network.
2. The method of claim 1 for predicting lightning based on spatio-temporal sequence clustering algorithm and LSTM neural network, wherein: in the step 1), time slice division is carried out on the thunder and lightning data to be predicted within a period of time, and the thunder and lightning data to be predicted within the period of time are selected; the data comprises longitude and latitude of lightning occurrence, lightning intensity, lightning gradient and lightning occurrence moment; dividing the data into time slices; it is assumed that the movement of the lightning center is jumping from divided time slice to time slice.
3. The method of claim 2 based on spatio-temporal sequence clustering algorithm and LSTM neural network lightning prediction, characterized in that: in the step 1), the Euclidean distance between all the lightning activity geographic positions on a time slice is calculated to obtain a distance set
Figure FDA0002871604000000013
Wherein i represents the number of lightning activities shared over the time slice, dhRepresenting the distance between some two lightning activities at this time slice, h representing the distance in the set D(i)Of the distance between two lightning activities at that time slice.
4. The method of claim 3 for predicting lightning based on spatio-temporal sequence clustering algorithm and LSTM neural network, wherein: in the step 1), a distance set D is counted(i)The distance number of the middle intervals V1[0, 0.1), V2[0.1, 0.2), V3[0.2, 0.3), V4[0.3, 0.4), V5[0.4, 0.5) is m1，m2，m3，m4，m5Performing representation and constructing a statistical matrix:
Figure FDA0002871604000000011
taking the median values of the intervals V1, V2, V3, V4 and V5 as p1, p2, p3, p4 and p5, and calculating by using a formula
Figure FDA0002871604000000012
Searching clusters by the obtained Eps neighborhood value and the defined Minpts 2 and by examining the Eps neighborhood of each point in the lightning data set; if the Eps neighborhood of the point O contains more than MinPts minimum points, a cluster taking the point O as a core object is created; when no new points are added to any cluster, a cluster of lightning aggregations is determined; if q does not belong to any cluster of lightning gathers, q is called a noise point.
5. The method of claim 4 for predicting lightning based on spatio-temporal sequence clustering algorithm and LSTM neural network, wherein: in said step 1), the lightning activity (o) on the determined cluster of lightning aggregates1，o2，...，on) Geographic location o1(w1，v1)，o2(w2，v2)，......，，on(wn，vn) And by the formula
Figure FDA0002871604000000021
Calculating the center of lightning by mean value method
Figure FDA0002871604000000022
Wherein o isnRepresenting a certain lightning central activity on the cluster, wnAnd vnRepresenting the latitude and longitude of a certain lightning center activity on the cluster,
Figure FDA0002871604000000023
and
Figure FDA0002871604000000024
represents the average of all lightning center latitudes and longitudes over the cluster.
6. The method of claim 1 for predicting lightning based on spatio-temporal sequence clustering algorithm and LSTM neural network, wherein: in the step 2), processing lightning center change data, wherein the lightning center change data comprises the starting time T and the ending time T of a time slice, the longitude L of a lightning center, the latitude W of the lightning center, the strength ST of the lightning center, the gradient SL of the lightning center, and the lightning number Num of the time slice is counted.
7. The method of claim 6 based on spatio-temporal sequence clustering algorithm and LSTM neural network lightning prediction, characterized in that: in the step 2), data T, L, ST, SL and Num on different time slices of a lightning center are input into a longitude LSTM neural network to obtain longitude values of the lightning activity in the next time sequence; inputting all lightning center data T, W, ST, SL, Num of a lightning activity into a latitude LSTM neural network to obtain latitude values of the lightning activity in the next time sequence.
CN202011617345.3A 2020-12-30 2020-12-30 Lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network Active CN112668790B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011617345.3A CN112668790B (en) 2020-12-30 2020-12-30 Lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011617345.3A CN112668790B (en) 2020-12-30 2020-12-30 Lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network

Publications (2)

Publication Number Publication Date
CN112668790A true CN112668790A (en) 2021-04-16
CN112668790B CN112668790B (en) 2023-07-25

Family

ID=75411445

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011617345.3A Active CN112668790B (en) 2020-12-30 2020-12-30 Lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network

Country Status (1)

Country Link
CN (1) CN112668790B (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113327022A (en) * 2021-05-18 2021-08-31 重庆莱霆防雷技术有限责任公司 Lightning protection safety risk management system and method

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103389388A (en) * 2012-05-08 2013-11-13 华锐风电科技(集团)股份有限公司 Method and device for predicting wind speed in wind power plant and method and system for predicting power in wind power plant
CN105118047A (en) * 2015-07-15 2015-12-02 陕西师范大学 Auroral oval boundary position prediction method based on interplanetary and geomagnetic parameters
CN106251026A (en) * 2016-08-16 2016-12-21 南京信息工程大学 Thunder and lightning based on PDBSCAN algorithm closes on trend prediction method
CN108427041A (en) * 2018-03-14 2018-08-21 南京中科九章信息技术有限公司 Lightning Warning method, system, electronic equipment and storage medium
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
CN109061774A (en) * 2018-06-26 2018-12-21 合肥佳讯科技有限公司 A kind of thunderstorm core relevance processing method
CN109738970A (en) * 2018-12-07 2019-05-10 国网江苏省电力有限公司电力科学研究院 The method, apparatus and storage medium for realizing Lightning Warning are excavated based on lightning data
CN111290053A (en) * 2020-02-29 2020-06-16 华南理工大学 Thunderstorm path prediction method based on Kalman filtering

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103389388A (en) * 2012-05-08 2013-11-13 华锐风电科技(集团)股份有限公司 Method and device for predicting wind speed in wind power plant and method and system for predicting power in wind power plant
CN105118047A (en) * 2015-07-15 2015-12-02 陕西师范大学 Auroral oval boundary position prediction method based on interplanetary and geomagnetic parameters
CN106251026A (en) * 2016-08-16 2016-12-21 南京信息工程大学 Thunder and lightning based on PDBSCAN algorithm closes on trend prediction method
CN108427041A (en) * 2018-03-14 2018-08-21 南京中科九章信息技术有限公司 Lightning Warning method, system, electronic equipment and storage medium
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
CN109061774A (en) * 2018-06-26 2018-12-21 合肥佳讯科技有限公司 A kind of thunderstorm core relevance processing method
CN109738970A (en) * 2018-12-07 2019-05-10 国网江苏省电力有限公司电力科学研究院 The method, apparatus and storage medium for realizing Lightning Warning are excavated based on lightning data
CN111290053A (en) * 2020-02-29 2020-06-16 华南理工大学 Thunderstorm path prediction method based on Kalman filtering

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
吴敏;王韬;赵淳;王佩;郭钧天;陈路;潘佳鑫;罗飞;: "基于过程的输电线路雷电活动描述新方法", 电网技术 *
黎维;陶蔚;周星宇;潘志松;: "时空序列预测方法综述", 计算机应用研究 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113327022A (en) * 2021-05-18 2021-08-31 重庆莱霆防雷技术有限责任公司 Lightning protection safety risk management system and method

Also Published As

Publication number Publication date
CN112668790B (en) 2023-07-25

Similar Documents

Publication Publication Date Title
WO2022089031A1 (en) Network optimization method based on big data and artificial intelligence
CN110928993A (en) User position prediction method and system based on deep cycle neural network
CN107918830B (en) Power distribution network running state evaluation method based on big data technology
CN107610021A (en) The comprehensive analysis method of environmental variance spatial and temporal distributions
CN112054943B (en) Traffic prediction method for mobile network base station
CN108632832B (en) Network coverage analysis method and system
CN113327022B (en) Lightning protection safety risk management system and method
CN107832876A (en) Subregion peak load Forecasting Methodology based on MapReduce frameworks
CN108898533A (en) Acquisition methods, device and the computer readable storage medium of movement of population data
Riihijärvi et al. Demonstrating radio environment map construction from massive data sets
CN110659672B (en) Method and device for predicting step-by-step uncertainty of output of wind turbine generator
CN110717619A (en) Multi-scale space-time load prediction method and system for bottom-up power distribution network
CN110322067A (en) Location of mobile users prediction technique based on factor graph model
CN114205831B (en) Method, device, storage medium and equipment for determining optimization scheme
CN110727752B (en) Position fingerprint database processing method, device and computer readable storage medium
CN112668790A (en) Thunder and lightning prediction method based on space-time sequence clustering algorithm and LSTM neural network
CN111833224A (en) Urban main and auxiliary center boundary identification method based on population grid data
Hou et al. Damage prediction of 10 kV power towers in distribution network under typhoon disaster based on data-driven model
He et al. Network traffic prediction method based on multi-channel spatial-temporal graph convolutional networks
CN102254090B (en) Method for estimating lightning density distribution and annual average ground flash density (NG) value by kernel density estimation
CN112561121B (en) Rework trend prediction method and system based on mobile phone signaling data
CN115630337A (en) Quantitative evaluation method and system for extreme rainfall attribution based on large-scale climate remote correlation
CN114239983A (en) Target area population flow prediction method and related device
CN114648156A (en) Time series load prediction method based on convolutional neural network
CN109446264B (en) Urban mobile data analysis method based on flow visualization

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