CN114492569B - Typhoon path classification method based on width learning system - Google Patents
Typhoon path classification method based on width learning system Download PDFInfo
- Publication number
- CN114492569B CN114492569B CN202111565583.9A CN202111565583A CN114492569B CN 114492569 B CN114492569 B CN 114492569B CN 202111565583 A CN202111565583 A CN 202111565583A CN 114492569 B CN114492569 B CN 114492569B
- Authority
- CN
- China
- Prior art keywords
- typhoon
- path
- paths
- points
- matrix
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/231—Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a typhoon path classification method based on a width learning system. The invention is specifically as follows: s1, carrying out cluster analysis on typhoon paths by utilizing an improved DBSCAN algorithm on the existing typhoon path data, and establishing a type label matrix of the typhoon paths; s2, carrying out feature representation on typhoon paths by using an improved hierarchical clustering algorithm, and unifying dimensions of all typhoon paths; s3, constructing a typhoon path classification model based on the width learning system, taking the type label matrix of the typhoon path and the typhoon path characteristic matrix in S2, which are established in S1, as input samples, and solving a weight matrix input to and output by the system, thereby realizing classification of typhoon paths. The method adopts the width learning system to classify the typhoon paths, thereby realizing the automatic classification of the typhoon paths.
Description
Technical Field
The invention belongs to the technical field of typhoon path classification, and particularly relates to a typhoon path classification method based on a width learning system.
Background
Typhoons are one of the most serious natural disasters worldwide, so that the research on the change rule of typhoons and the cause thereof has great scientific significance for typhoons forecast, disaster prevention and disaster reduction.
The width learning system (BLS) is an incremental learning algorithm based on a random vector function linked neural network, and because the training process does not need to iterate sample data repeatedly and calculates a network output layer weight matrix in a pseudo-inverse mode through ridge regression, the method is more suitable for processing work with larger calculation amount, such as data classification and the like, compared with a common neural network.
The width learning system consists of a feature mapping layer, an enhancement node layer and an output layer, wherein the feature mapping layer and the enhancement vector layer are used as the input of the system together. The feature mapping layer randomly generates weights through a feature mapping function to realize feature extraction of the sample. The enhancement node layer carries out enhancement calculation on the feature vector through orthogonal normalized random weights, and introduces nonlinear classification capability of an activation function enhancement model, so that the purpose of fully extracting sample data feature information is achieved. And finally, performing pseudo-inverse operation on the composite matrix of the feature mapping layer and the enhanced node layer to obtain a weight matrix input to and output from the system. When the width learning system is used for solving the classification problem, the data set to be classified and the tag set of the data set to be classified are used as inputs of the width learning system, the width learning system is trained, the width learning system outputs probabilities that all data belong to all categories respectively, and the category with the largest probability is the category in which the data are judged.
The typhoon path is a time sequence formed by positions of typhoons passing through a center point in the movement process, and the typhoons are complicated in movement path due to the influence of regional environment, climate and the like, so that classification analysis of the typhoons is difficult. Currently, classification researches on typhoon paths are mainly focused on cluster analysis on typhoons, and classification is carried out on typhoons by using similar thresholds. However, this method is only suitable for classifying few typhoon paths, and when the typhoon path set is large, the definition of the similarity threshold is difficult and the general type is poor.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a typhoon path classification method which is represented by combining a width learning system with typhoon path characteristics.
The method comprises the following steps:
s1, constructing a typhoon path category set. Clustering analysis is carried out on typhoon paths by utilizing an improved DBSCAN algorithm, the similarity among paths is calculated by utilizing a dynamic time warping algorithm instead of a similarity calculation method in the DBSCAN algorithm, paths with high similarity are automatically gathered into one type, and a type label matrix of the typhoon paths is established;
s2, constructing a feature matrix of the typhoon path. The method comprises the steps of unifying lengths of typhoon paths through characteristic representation by utilizing an improved hierarchical clustering algorithm, wherein, because typhoon path points are data sequences which are strictly arranged according to time sequence, the hierarchical clustering algorithm is improved, hierarchical clustering analysis is only carried out on adjacent points of each path, euclidean distances (Euclidean distances) of the adjacent points are used as measurement standards, the points with smaller Euclidean distances are combined, the lengths of all typhoon paths are unified, the characteristic representation of each typhoon path is realized, and a typhoon path characteristic matrix is obtained;
s3, establishing a typhoon path multi-classification model based on the width learning system, and classifying typhoon paths by using the model. The invention relates to a width learning system (BLS) which is an incremental learning algorithm based on a random vector function linked neural network, wherein a typhoon path type label matrix obtained in S1 and a typhoon path characteristic matrix obtained in S2 are used as the input of a model, a characteristic mapping layer randomly generates weight to perform characteristic extraction on an input sample, an enhancement node layer enhances the characteristic vector by orthogonal normalized random weight, the nonlinear classification capacity of the model is enhanced by an activation function, finally, the pseudo-inverse operation is performed on a composite matrix of the characteristic mapping layer and the enhancement node layer, the weight matrix input to the output of the system is obtained, the output value of each row of the output matrix is the probability that each path obtained by the BLS belongs to each category respectively, and the position index of the maximum value of each row is the category determined by each test path, so that the classification prediction of the typhoon path is realized.
Compared with the prior art, the invention has the following advantages:
first, the method of the invention adopts an improved DBSCAN clustering algorithm to establish the category set of typhoon paths, and can automatically generate the category set according to the similarity of each typhoon path of the history.
Secondly, the method adopts a characteristic representation method for improving hierarchical clustering, unifies the data of different air paths, and provides a basis for classification.
Third, the method adopts the width learning system to classify the typhoon paths, thereby realizing the automatic classification of the typhoon paths.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings may be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a typhoon path classification method based on a width learning system in an embodiment of the present invention.
Detailed Description
As shown in fig. 1, the present invention includes the steps of:
s1, establishing a typhoon path category set through a clustering algorithm.
The typhoon path is a track sequence of typhoon centers arranged according to time sequence in the typhoon movement process, and the DBSCAN algorithm is a clustering algorithm based on density, and can find clusters with any shape, automatically determine the number of the clusters and have robustness to noise, so that the method is very suitable for carrying out cluster analysis on track data such as typhoon paths, and paths with high similarity are automatically clustered into one type. Similarity is generally measured in terms of distance, with smaller distances providing greater similarity. When the original DBSCAN algorithm performs cluster analysis, the Euclidean distance measurement method is adopted, and the method is only suitable for typhoon paths with the same number of points and does not consider the fluctuation condition of the paths, so that the method utilizes the dynamic time warping algorithm as the distance measurement method of the paths. The dynamic time warping algorithm is to lengthen or shorten different paths to be compared until the lengths are consistent, and then calculate the distance between the paths. The typhoon path category set establishment process is therefore:
(1) And (3) setting a scanning neighborhood radius eps, wherein the minimum number of paths is minPts, optionally starting an unaccessed path in a typhoon path sample, calculating the distance between the path and other paths by using a dynamic time warping algorithm, and finding out all nearby paths with the distance within eps (including eps).
(2) If the number of nearby paths is equal to or greater than minPts, the current path forms a cluster with its nearby paths, and the departure path is marked as visited.
(3) Repeating steps (1) - (2), processing all paths within the cluster that are not marked as visited, and if the number of nearby paths is less than minPts, the paths are marked as noisy data.
(4) If all paths in the cluster are marked as accessed, repeating the steps (2) - (3) until all objects are classified as a certain cluster or marked as noise data, and outputting the category label matrix of each path. The feature matrix of each path is denoted as the label matrix of path features and is denoted as Y.
S2, the data length of each typhoon path may be different, so that the original data cannot be directly used for constructing the feature matrix for classification. In consideration of the fact that typhoon paths cannot be accurately fitted by using a functional relation, the typhoon paths are represented by extracting the same number points from all typhoon paths by using an improved hierarchical clustering algorithm. The hierarchical clustering algorithm is based on the principle that the distances between samples are calculated first, and the closest points are merged into the same class each time. Then, the distance between classes is calculated again, the class closest to the class is combined into a large class, and the combination is continued until the specified condition is reached. The invention utilizes an improved hierarchical clustering method to perform characteristic representation on typhoon paths, only uses Euclidean distance (Euclidean distance) of adjacent points as a measurement standard, combines points with small Euclidean distance, and uses the combined points as characteristic points of the paths. If l i For point T i To the pointT i+1 Is used for the distance of (a),wherein x is i And y i Respectively are points T i Longitude and latitude, x i+1 And y i+1 Respectively are points T i+1 The specific calculation steps are as follows:
a1, calculating the distance l between adjacent points on the original typhoon path i 。
A2, all distances l i Stored in set S, s= { l 1 ,l 2 ,…,l n-1 -wherein, l 1 For point T 1 Sum point T 2 Distance between points, l 2 For point T 2 Sum point T 3 Distance between l n-1 For point T n-1 Sum point T n The distance between the two points, i represents the number of the points and n represents the total number of the points.
And A3, merging two points with the smallest distance in the S set into one point, and representing the point by using the mean value of longitude and latitude of the two points to form a new path.
And A4, recalculating the distance between each two adjacent points on the new path according to the step A2, and updating the set S.
A5, turning to the step A3 until the number of points of all new paths reaches the length of the shortest path set in the sample set, and outputting a characteristic path set F.
And S3, establishing a typhoon path classification model based on the width learning system, and carrying out classification prediction based on the model. The width learning system (BLS) is an incremental learning algorithm based on a random vector function linked neural network, a feature path set F and a label matrix Y are taken as inputs, a feature mapping layer randomly generates weight to conduct feature extraction on input samples, an enhancement node layer conducts enhancement calculation on feature vectors through orthogonal normalized random weight, nonlinear classification capacity of an activation function enhancement model is utilized, finally, a pseudo-inverse operation is conducted on a synthesis matrix of the feature mapping layer and the enhancement node layer, a weight matrix input to the output of the system is obtained, an output value of each row of the output matrix is the probability that each path obtained by the BLS belongs to each category respectively, and a position index of the maximum value of each row is the category determined by each test path, so that classification prediction of typhoon paths is achieved.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.
Claims (3)
1. The typhoon path classification method based on the width learning system is characterized by comprising the following steps of:
s1, constructing a typhoon path type set;
the improved DBSCAN algorithm is utilized to carry out cluster analysis on typhoon paths, and the method specifically comprises the following steps: calculating the similarity between paths by using a dynamic time warping algorithm instead of a similarity calculation method in a DBSCAN algorithm, and automatically gathering paths with high similarity into one type; establishing a category label matrix of the typhoon path;
s2, constructing a feature matrix of the typhoon path;
the length of typhoon paths is unified through characteristic representation by utilizing an improved hierarchical clustering algorithm, and the method specifically comprises the following steps: improving a hierarchical clustering algorithm, performing hierarchical clustering analysis on adjacent points of each path, combining points with smaller Euclidean distances by taking Euclidean distances of the adjacent points as measurement standards, unifying lengths of all typhoon paths, and realizing characteristic representation of each typhoon path; obtaining a typhoon path feature matrix;
s3, classifying typhoon paths by using a typhoon path multi-classification model based on a width learning system;
taking the typhoon path type label matrix obtained in the step S1 and the typhoon path characteristic matrix obtained in the step S2 as the input of a model;
the feature mapping layer in the typhoon path multi-classification model randomly generates weights to conduct feature extraction on input samples, the enhancement node layer conducts enhancement calculation on feature vectors through orthogonal normalized random weights, and the nonlinear classification capacity of the model is enhanced by utilizing an activation function;
and finally, calculating a weight matrix input to and output by the system by carrying out pseudo-inverse operation on the composite matrix of the feature mapping layer and the enhanced node layer, wherein the output value of each row of the output matrix is the probability that each path calculated by the width learning system belongs to each category, and the position index of the maximum value of each row is the category judged by each test path, so that the classification prediction of typhoon paths is realized.
2. The typhoon path classification method based on the width learning system according to claim 1, wherein the step of constructing the typhoon path type tag matrix in S1 is as follows:
s11, setting a scanning neighborhood radius eps, wherein the minimum number of paths is minPts, optionally starting an unviewed path in a typhoon path sample, calculating the distance between the path and other paths by using a dynamic time warping algorithm, and finding out all nearby paths with the distance within eps;
s12, if the number of the nearby paths is greater than or equal to minPts, forming a cluster by the current path and the nearby paths, and marking the departure path as accessed;
s13, repeating S11-S12, processing all paths which are not marked as accessed in the cluster, and marking the paths as noise data if the number of the nearby paths is less than minPts;
s14, if all paths in the cluster are marked as accessed, repeating the steps S12-S13 until all objects are classified into a certain cluster or marked as noise data, and outputting a category label matrix of each path.
3. The typhoon path classification method based on the width learning system according to claim 1, wherein the typhoon path feature matrix construction step in S2 is as follows:
s21, calculating original each stationDistance l between adjacent points on the wind path i ；
S22, setting all distances l i Saving in the set S; s= { l 1 ,l 2 ,…,l n-1 -wherein, l 1 For point T 1 Sum point T 2 Distance between points, l 2 For point T 2 Sum point T 3 Distance between l n-1 For point T n-1 Sum point T n The distance between the two points, i represents the serial number of the points, and n represents the total number of the points;
s23, merging two points with the smallest distance in the set S into a point, and representing the point by using the mean value of longitude and latitude of the two points to form a new path;
s24, recalculating the distance between the adjacent points of the newly generated points of the new path according to the step S22, and updating the set S;
s25, turning to the step S23 until the point number of all the new paths reaches the length of the shortest path set in the sample set, and outputting the characteristic path set.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111565583.9A CN114492569B (en) | 2021-12-20 | 2021-12-20 | Typhoon path classification method based on width learning system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111565583.9A CN114492569B (en) | 2021-12-20 | 2021-12-20 | Typhoon path classification method based on width learning system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114492569A CN114492569A (en) | 2022-05-13 |
| CN114492569B true CN114492569B (en) | 2023-08-29 |
Family
ID=81495059
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111565583.9A Active CN114492569B (en) | 2021-12-20 | 2021-12-20 | Typhoon path classification method based on width learning system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114492569B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115907159B (en) * | 2022-11-22 | 2023-08-29 | 应急管理部国家减灾中心 | Method, device, equipment and medium for determining typhoons in similar paths |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102289570A (en) * | 2011-07-23 | 2011-12-21 | 浙江大学 | Flood forecast method based on rainfall-runoff-flood routing calculation |
| CN103500278A (en) * | 2013-09-30 | 2014-01-08 | 福建四创软件有限公司 | Path similarity typhoon analysis method based on shortest path algorithm |
| CN110209825A (en) * | 2019-06-17 | 2019-09-06 | 大连海事大学 | A kind of fast network representative learning algorithm based on width learning system |
| CN110717390A (en) * | 2019-09-05 | 2020-01-21 | 杭州电子科技大学 | Electroencephalogram signal classification method based on graph semi-supervised width learning |
| CN111401443A (en) * | 2020-03-16 | 2020-07-10 | 重庆大学 | Width learning system based on multi-feature extraction |
| CN112581315A (en) * | 2020-12-25 | 2021-03-30 | 东北电力大学 | Wind power field clustering method based on extreme gradient dynamic density clustering |
| US11121788B1 (en) * | 2020-06-08 | 2021-09-14 | Wuhan University | Channel prediction method and system for MIMO wireless communication system |
| AU2021105840A4 (en) * | 2021-08-18 | 2021-10-28 | Xinjiang Institute Of Ecology And Geography, Chinese Academy Of Sciences | Method for predicting typhoon track |
-
2021
- 2021-12-20 CN CN202111565583.9A patent/CN114492569B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102289570A (en) * | 2011-07-23 | 2011-12-21 | 浙江大学 | Flood forecast method based on rainfall-runoff-flood routing calculation |
| CN103500278A (en) * | 2013-09-30 | 2014-01-08 | 福建四创软件有限公司 | Path similarity typhoon analysis method based on shortest path algorithm |
| CN110209825A (en) * | 2019-06-17 | 2019-09-06 | 大连海事大学 | A kind of fast network representative learning algorithm based on width learning system |
| CN110717390A (en) * | 2019-09-05 | 2020-01-21 | 杭州电子科技大学 | Electroencephalogram signal classification method based on graph semi-supervised width learning |
| CN111401443A (en) * | 2020-03-16 | 2020-07-10 | 重庆大学 | Width learning system based on multi-feature extraction |
| US11121788B1 (en) * | 2020-06-08 | 2021-09-14 | Wuhan University | Channel prediction method and system for MIMO wireless communication system |
| CN112581315A (en) * | 2020-12-25 | 2021-03-30 | 东北电力大学 | Wind power field clustering method based on extreme gradient dynamic density clustering |
| AU2021105840A4 (en) * | 2021-08-18 | 2021-10-28 | Xinjiang Institute Of Ecology And Geography, Chinese Academy Of Sciences | Method for predicting typhoon track |
Non-Patent Citations (1)
| Title |
|---|
| 基于宽度学习系统的内河航道船舶轨迹分类算法;王颢程等;《上海海事大学学报》;第42卷(第3期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114492569A (en) | 2022-05-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109471938B (en) | Text classification method and terminal | |
| CN111368920B (en) | Quantum twin neural network-based classification method and face recognition method thereof | |
| CN108229550B (en) | Cloud picture classification method based on multi-granularity cascade forest network | |
| CN111833172A (en) | Consumption credit fraud detection method and system based on isolated forest | |
| CN108875816A (en) | Merge the Active Learning samples selection strategy of Reliability Code and diversity criterion | |
| CN111008337B (en) | Deep attention rumor identification method and device based on ternary characteristics | |
| CN112069310A (en) | Text classification method and system based on active learning strategy | |
| CN109581339B (en) | Sonar identification method based on automatic adjustment self-coding network of brainstorming storm | |
| Casalino et al. | Incremental adaptive semi-supervised fuzzy clustering for data stream classification | |
| CN112270355A (en) | Active safety prediction method based on big data technology and SAE-GRU | |
| CN113269239B (en) | Relation network node classification method based on multichannel convolutional neural network | |
| Yu et al. | Tulsn: siamese network for trajectory-user linking | |
| CN114492569B (en) | Typhoon path classification method based on width learning system | |
| CN114093445B (en) | Patient screening marking method based on partial multi-marking learning | |
| CN114897085A (en) | Clustering method based on closed subgraph link prediction and computer equipment | |
| CN114549897A (en) | Training method and device for classification model and storage medium | |
| CN112990371B (en) | Unsupervised night image classification method based on feature amplification | |
| CN112529025A (en) | Data processing method and device | |
| CN116304941A (en) | Ocean data quality control method and device based on multi-model combination | |
| Steyn et al. | A nearest neighbor open-set classifier based on excesses of distance ratios | |
| Zhou et al. | Difficult Novel Class Detection in Semisupervised Streaming Data | |
| CN114254738A (en) | Double-layer evolvable dynamic graph convolution neural network model construction method and application | |
| Shaji et al. | Weather Prediction Using Machine Learning Algorithms | |
| CN110738243B (en) | Self-adaptive unsupervised feature selection method | |
| Ali Shatat et al. | Big Data Driven Map Reduce Framework for Automated Flood Disaster Detection Based on Heuristic-Based Ensemble Learning |
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 |