CN116740586A - Hail identification method, hail identification device, electronic equipment and computer readable storage medium - Google Patents
Hail identification method, hail identification device, electronic equipment and computer readable storage medium Download PDFInfo
- Publication number
- CN116740586A CN116740586A CN202310821981.5A CN202310821981A CN116740586A CN 116740586 A CN116740586 A CN 116740586A CN 202310821981 A CN202310821981 A CN 202310821981A CN 116740586 A CN116740586 A CN 116740586A
- Authority
- CN
- China
- Prior art keywords
- hail
- matrix
- attention
- early
- satellite
- 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.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 63
- 239000011159 matrix material Substances 0.000 claims abstract description 168
- 230000007246 mechanism Effects 0.000 claims abstract description 35
- 238000004364 calculation method Methods 0.000 claims abstract description 26
- 238000012549 training Methods 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 18
- 230000003044 adaptive effect Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 11
- 238000010586 diagram Methods 0.000 claims description 8
- 238000010606 normalization Methods 0.000 claims description 8
- 238000001514 detection method Methods 0.000 claims description 7
- 230000009466 transformation Effects 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 5
- 239000000178 monomer Substances 0.000 claims description 4
- 238000007405 data analysis Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 description 6
- 238000011161 development Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 3
- 238000004590 computer program Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 125000004122 cyclic group Chemical group 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 230000001537 neural effect Effects 0.000 description 2
- 230000008447 perception Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
- G06V20/13—Satellite images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Multimedia (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Astronomy & Astrophysics (AREA)
- Remote Sensing (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a hail identification method, a hail identification device, electronic equipment and a computer readable storage medium, comprising the following steps: determining a hail to-be-early-warning area based on current ground station observation data, and analyzing satellite data of the hail to-be-early-warning area to obtain a target data set; performing self-adaptive attention mechanism calculation on an adjacent matrix corresponding to the target data set to obtain a first attention characteristic matrix; through a pre-trained hail recognition model, hail recognition is carried out on a hail to-be-early-warning area based on a first attention feature matrix, and a first hail recognition result corresponding to the hail to-be-early-warning area is obtained. The invention can efficiently and accurately identify hail weather and can also realize early warning of hail.
Description
Technical Field
The invention relates to the technical field of meteorological monitoring and early warning, in particular to a hail identification method, a hail identification device, electronic equipment and a computer readable storage medium.
Background
Hail is a product of a severely convective weather process that can cause significant damage to crops and property. The spatial and temporal scales involved in hail occurrence are small, the impact range of each hail is typically about tens to thousands of meters wide, the duration is mostly 2-10 minutes, the minority is more than 30 minutes, and the longer the duration the greater the losses. Strong convective weather may occur at any time and hail drop zones are difficult to determine, preventing detailed investigation of hail properties while limiting the development of identification and forecasting tools. Due to the lack of instruments and methods, the difficulty in identifying hail is high, the most common method is to obtain atmospheric parameters triggering convection conditions from numerical forecasting, or to realize early warning of hail by utilizing derivative features of radar data, but timeliness of early warning of hail cannot be guaranteed.
Disclosure of Invention
Accordingly, the present invention is directed to a hail identifying method, apparatus, electronic device and computer readable storage medium, which can identify hail weather efficiently and accurately, and can also realize early warning of hail.
In a first aspect, an embodiment of the present invention provides a hail identifying method, including:
determining a hail to-be-early-warning area based on current ground station observation data, and analyzing satellite data of the hail to-be-early-warning area to obtain a target data set;
performing self-adaptive attention mechanism calculation on an adjacent matrix corresponding to the target data set to obtain a first attention characteristic matrix;
and performing hail recognition on the hail to-be-early-warning area based on the first attention feature matrix through a pre-trained hail recognition model to obtain a first hail recognition result corresponding to the hail to-be-early-warning area.
In one embodiment, analyzing satellite data of the hail to-be-early-warning area to obtain a target data set includes:
extracting initial cloud top bright temperature from satellite data of the hail to-be-early-warning area; extracting bright temperature data from satellite data of the hail to-be-early-warning area, and determining an initial bright temperature gradient based on the bright temperature data; extracting designated channel data from satellite data of the hail to-be-early-warned area, and determining an initial hail storm index based on the designated channel data;
Respectively carrying out normalization processing on the initial cloud top bright temperature, the initial bright temperature gradient and the initial hailstorm index to obtain a target cloud top bright temperature, a target bright temperature gradient and a target hailstorm index;
and taking one or more of the target cloud top bright temperature, the target bright temperature gradient and the target hailstorm index as a target data set.
In one embodiment, performing adaptive attention mechanism calculation on an adjacency matrix corresponding to the target data set to obtain a first attention feature matrix, including:
constructing an initial satellite characteristic sequence matrix based on the target data set, and determining an adjacent matrix corresponding to the initial satellite characteristic sequence matrix;
performing linear transformation on the initial satellite feature sequence matrix by using a plurality of preset weights to obtain a plurality of target satellite feature sequence matrices;
and carrying out self-adaptive attention mechanism calculation based on the adjacent matrix, the initial satellite feature sequence matrix and each target satellite feature sequence matrix to obtain a first attention feature matrix.
In one embodiment, performing adaptive attention mechanism calculation based on the adjacency matrix, the initial satellite feature sequence matrix and each target satellite feature sequence matrix to obtain a first attention feature matrix, including:
The attention profile matrix is determined according to the following formula:
Q＝linear(X)＝AXW Q
K＝linear(X)＝AXW K
V＝linear(X)＝AXW V
Attention(Q,K,V)＝softmax(QK T )V+b；
wherein Attention (Q, K, V) is the first Attention feature matrix, g 1 ,g 2 ∈R N×N ，As a learnable parameter, A is E R N×N Is an adjacent matrix, X is an initial satellite characteristic sequence matrix, Q is a first target satellite characteristic sequence matrix, K is a second target satellite characteristic sequence matrix, V is a third target satellite characteristic sequence matrix, QK T As the attention matrix, +..
In one embodiment, the hail identification model employs an adaptive attention space-time graph convolutional loop network;
through a pre-trained hail recognition model, hail recognition is carried out on the hail to-be-early-warning area based on the first attention feature matrix, and a first hail recognition result corresponding to the hail to-be-early-warning area is obtained, and the hail recognition method comprises the following steps:
and inputting the first attention characteristic matrix into the self-adaptive attention time-space diagram convolution circulation network to obtain a first hail identification result corresponding to the hail to-be-early-warned area.
In one embodiment, the training step of the hail identification model comprises:
acquiring a training data set and an actual hail observation result; wherein the actual hail observations are used to characterize the training dataset labels and detection ranges;
Performing self-adaptive attention mechanism calculation on an adjacent matrix corresponding to the training data set to obtain a second attention feature matrix, and performing hail identification on the storm single region based on the second attention feature matrix through a hail identification model to obtain a second hail identification result corresponding to the storm single region;
adjusting network parameters of the hail identification model based on the actual hail observation result and the second hail identification result by using a cross entropy loss function; the cross entropy loss function is minimized by adopting an Adam algorithm, and L2 regularization is added into the cross entropy loss function.
In one embodiment, the method further comprises:
determining a first number of actual hail observations as true and the first hail recognition results as true, and a second number of actual hail observations as false and the first hail recognition results as true, and a third number of actual hail observations as true and the first hail recognition results as false;
determining a critical success index, a hit probability index, and a false alarm index, respectively, based on the first number, the second number, and the third number;
And evaluating the hail identification model by using one or more of the critical success index, the hit probability index and the false alarm index to obtain an evaluation result of the hail identification model.
In a second aspect, an embodiment of the present invention further provides a hail identifying device, including:
the data analysis module is used for determining a hail to-be-early-warning area based on current ground station observation data and analyzing satellite data of the hail to-be-early-warning area to obtain a target data set;
the attention module is used for carrying out self-adaptive attention mechanism calculation on the adjacent matrix corresponding to the target data set to obtain a first attention characteristic matrix;
and the hail recognition module is used for carrying out hail recognition on the hail to-be-early-warning area based on the first attention feature matrix through a pre-trained hail recognition model to obtain a first hail recognition result corresponding to the hail to-be-early-warning area.
In a third aspect, an embodiment of the present invention further provides an electronic device comprising a processor and a memory storing computer-executable instructions executable by the processor to implement the method of any one of the first aspects.
In a fourth aspect, embodiments of the present invention also provide a computer-readable storage medium storing computer-executable instructions which, when invoked and executed by a processor, cause the processor to implement the method of any one of the first aspects.
According to the hail identification method, the hail identification device, the electronic equipment and the computer readable storage medium, firstly, a hail to-be-early-warning area is determined based on current ground station observation data, satellite data of the hail to-be-early-warning area are analyzed to obtain a target data set, then adaptive attention mechanism calculation is carried out on an adjacent matrix corresponding to the target data set, so that a first attention feature matrix is obtained, finally, hail identification is carried out on the hail to-be-early-warning area based on the first attention feature matrix through a pre-trained hail identification model, and a first hail identification result corresponding to the hail to-be-early-warning area is obtained. According to the method, the spatial distribution and the observation time frequency of the hail occurrence scale can be effectively monitored on the basis of satellite data of the hail to-be-early-warning area, the hail cloud is completely monitored from the initial occurrence stage to the development process to the end of dissipation, a new identification method is provided for hail early warning, and accordingly early warning of hail is achieved; in addition, a self-adaptive attention mechanism is calculated by fusing the distance adjacency matrix, the distance adjacency matrix and the attention self-adaptation are combined together, the distance adjacency matrix is used as part of characteristic information to be input into the self-adaptive attention mechanism to obtain a first attention characteristic matrix, and fine-granularity data attributes among nodes are captured, so that the hail recognition accuracy is improved; in addition, a first hail recognition result is obtained based on a first attention feature matrix through a pre-trained hail recognition model, the spatial and time correlation of specific nodes in satellite data is captured through the hail recognition model, data information in the satellite data is fully learned, and the hail recognition accuracy is further improved.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
In order to make the above objects, features and advantages of the present invention more comprehensible, preferred embodiments accompanied with figures are described in detail below.
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 needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a schematic flow chart of a hail identifying method according to an embodiment of the present invention;
FIG. 2 is a flow chart of another hail identification method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a hail identifying device according to an embodiment of the present invention;
Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described in conjunction with the embodiments, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
At present, the timeliness of early warning of hail cannot be guaranteed by the existing hail identification method, and based on the timeliness, the hail identification method, the hail identification device, the electronic equipment and the computer readable storage medium can be used for efficiently and accurately identifying hail weather and also can be used for early warning of hail.
For the convenience of understanding the present embodiment, a detailed description will be first given of a hail identifying method disclosed in the present embodiment, referring to a schematic flow chart of a hail identifying method shown in fig. 1, the method mainly includes the following steps S102 to S106:
And step S102, determining a hail to-be-early-warning area based on current ground station observation data, and analyzing satellite data of the hail to-be-early-warning area to obtain a target data set. The target data set comprises a target cloud top bright temperature, a target bright temperature gradient and a target hailstorm index.
In one embodiment, the longitude and latitude of the ground station (i.e., the ground automatic station) for observing the hail are found based on historical meteorological data, the observed data of the ground station at least comprises the hail occurrence time and the longitude and latitude, so that a hail to-be-early-warning area is determined based on the hail occurrence time and the longitude and latitude by combining with the hail scale area, and a target data set such as a target cloud top bright temperature, a target bright temperature gradient, a target hail storm index and the like can be obtained by analyzing satellite data of the hail to-be-early-warning area.
Step S104, performing self-adaptive attention mechanism calculation on the adjacent matrix corresponding to the target data set to obtain a first attention characteristic matrix.
In one embodiment, an initial satellite feature sequence matrix is first constructed, an adjacent matrix of the initial satellite feature sequence matrix is determined according to the initial satellite feature sequence matrix, a plurality of target satellite feature sequence matrices after linear transformation of the initial satellite feature sequence matrix are determined, and adaptive attention mechanism calculation is performed based on the adjacent matrix, the initial satellite feature sequence matrix and each target satellite feature sequence matrix to obtain a first attention feature matrix.
And S106, performing hail recognition on the hail to-be-early-warning area based on the first attention feature matrix through a pre-trained hail recognition model to obtain a first hail recognition result corresponding to the hail to-be-early-warning area. The input of the hail identification model is a first attention feature matrix, and the output is a first hail identification result. In one embodiment, the hail to-be-early-warning area can be subjected to high-dimensional map convolution operation and perception mechanism processing through the hail identification model, and a first hail identification result corresponding to the hail to-be-early-warning area can be obtained.
According to the hail identification method provided by the embodiment of the invention, the spatial distribution and the observation time frequency of the hail occurrence scale can be effectively monitored on the basis of satellite data of the hail to-be-early-warning area, the hail cloud is completely monitored from the initial occurrence stage to the development process to the end of dissipation, a new identification method is provided for hail early warning, and accordingly early warning of hail is realized; in addition, a self-adaptive attention mechanism is calculated by fusing the distance adjacency matrix, the distance adjacency matrix and the attention self-adaptation are combined together, the distance adjacency matrix is used as part of characteristic information to be input into the self-adaptive attention mechanism to obtain a first attention characteristic matrix, and fine-granularity data attributes among nodes are captured, so that the hail recognition accuracy is improved; in addition, a first hail recognition result is obtained based on a first attention feature matrix through a pre-trained hail recognition model, the spatial and time correlation of specific nodes in satellite data is captured through the hail recognition model, data information in the satellite data is fully learned, and the hail recognition accuracy is further improved.
At present, FY4A satellite adopts multichannel scanning imaging, and experimental analysis is carried out on historical satellite observation data by referring to site observation data and combining the change of cloud cluster generated during hail reduction to analyze the structure and characteristics of imaging cloud cluster. The satellite detection has the characteristics of wide monitoring range, low detection cost, high operation stability and the like, a machine learning algorithm is adopted to identify hail by utilizing satellite data, the probability classification level of hail is realized, the effective early warning duration of hail is improved, the structure and time information of the satellite data are analyzed and learned, and the problem of missing report height in service is solved. For the above reasons, the embodiment of the invention provides a hail identification method. In order to facilitate understanding of the foregoing embodiments, embodiments of the present invention provide a specific implementation of a hail identifying method.
For the foregoing step S102, the hail to-be-early-warning area may be determined according to the longitude and latitude and the hail scale area observed by the ground station, and then the steps of analyzing satellite data of the hail to-be-early-warning area to obtain the target data set are performed according to the following steps a to c:
step a, extracting initial cloud top bright temperature from satellite data of a hail to-be-early-warning area; extracting bright temperature data from satellite data of a hail to-be-early-warned area, and determining an initial bright temperature gradient based on the bright temperature data; and extracting designated channel data from satellite data of the hail to-be-early-warned area, and determining an initial hailstorm index based on the designated channel data.
In one example, the initial cloud top bright temperature may be read directly from satellite data;
in one example, satellite channel bright temperature data associated with the occurrence of hail may be selected. Calculating an initial bright temperature gradient;
in one example, the specified channel data may be 0.55 to 0.75 band data (i.e., data of satellite 2 channels of the cloud 4) and 3.5 to 4.0 band data (i.e., data of satellite 7 channels of the cloud 4), that is, the initial hail storm index is calculated by selecting the data of satellite 2 channels of the cloud 4 and the data of satellite 7 channels, the situation that the non-strong convection point is identified as the strong convection point is effectively reduced by using the initial hail storm index, and the calculation formula is as follows:
H I ＝(ρ Channel2 -ρ Channel7 )/(ρ Channel2 +ρ Channel7 )；
wherein H is I To an initial hailstorm index ρ Channel2 For 2-channel data ρ Channel7 7-channel data.
And b, respectively carrying out normalization treatment on the initial cloud top bright temperature, the initial bright temperature gradient and the initial hailstorm index to obtain the target cloud top bright temperature, the target bright temperature gradient and the target hailstorm index.
In one embodiment, before inputting all data into the model, quality inspection is performed on all data, missing data and null values are interpolated, unreasonable data is removed, and normalization processing is performed on all data.
Taking the initial hailstorm index as an example, because the units of the detected objects of the channel 2 and the channel 7 are not uniform, the initial hailstorm index is subjected to Min-Max standardized normalization treatment according to the following formula:
Wherein, when the initial hailstorm index is normalized, x is shown in the formula min Refers to the smallest initial hailstorm index, x max Refers to the maximum initial hailstorm index, x' refers to the normalized target hailstorm index, and x refers to the current initial hailstorm index.
Similarly, the initial cloud top bright temperature and the initial bright temperature gradient can be normalized according to the normalization processing formula, and parameter meanings in the normalization processing formula are adaptively adjusted, which is not described in detail in the embodiment of the present invention.
And c, integrating one or more of the target cloud top bright temperature, the target bright temperature gradient and the target hailstorm index to obtain a target data set.
For the foregoing step S104, the embodiment of the present invention further provides an implementation manner of performing adaptive attention mechanism calculation on the adjacency matrix corresponding to the target data set to obtain the first attention feature matrix, which is referred to as step 1 to step 3 below:
and step 1, constructing an initial satellite characteristic sequence matrix based on the target data set, and determining an adjacent matrix corresponding to the initial satellite characteristic sequence matrix. In one embodiment, the sequence matrix for a plurality of initial satellite features may be expressed as:
X＝{X 1 ,X 2 ,...,X t ,...} T ∈R N×F×t
where N is denoted as the number of nodes, F is denoted as the number of features, t is denoted as the time step, Indicating that each time step contains N nodes, < > N->The node N is shown to contain F features at time step t, namely the target cloud top bright temperature, target bright temperature gradient and target hailstorm index.
Based on the above embodiments, the embodiments of the present invention provide an adjacent matrix corresponding to an initial satellite feature sequence matrix, which is specifically as follows:
wherein A is i,j Represented as a distance adjacency matrix between nodes i and j,representing the euclidean distance between nodes i and j.
And 2, performing linear transformation on the initial satellite characteristic sequence matrix by using a plurality of preset weights to obtain a plurality of target satellite characteristic sequence matrices. Wherein the preset weights may includeIn one embodiment, a linear mapping of the input satellite data is calculated, three weights are assigned, respectively,/->F in For input dimension F out Is the output dimension.
The three weights are utilized to perform linear transformation on the initial satellite feature sequence matrix to form three matrices (namely, target satellite feature sequence matrices), wherein the three matrices comprise a first target satellite feature sequence matrix Q (short for Query matrix), a second target satellite feature sequence matrix K (short for Key matrix) and a third target satellite feature sequence matrix V (short for Value matrix), and the three matrices are specifically shown as follows:
Q＝linear(X)＝AXW Q ；
K＝linear(X)＝AXW K ；
V＝linear(X)＝AXW V 。
And step 3, performing self-adaptive attention mechanism calculation based on the adjacent matrix, the initial satellite feature sequence matrix and each target satellite feature sequence matrix to obtain a first attention feature matrix. In one embodiment, the attention profile matrix may be determined according to the following formula:
Attention(Q,K,V)＝softmax(QK T )V+b；
wherein Attention (Q, K, V) is the first Attention feature matrix, g 1 ,g 2 ∈R N×N ，As a learnable parameter, A is E R N×N Is an adjacent matrix, X is an initial satellite characteristic sequence matrix, Q is a first target satellite characteristic sequence matrix, K is a second target satellite characteristic sequence matrix, V is a third target satellite characteristic sequence matrix, QK T For the attention matrix, +.: />
For the foregoing step S106, the embodiment of the present invention further provides an implementation manner of obtaining, by using a hail recognition model, a first hail recognition result based on a first attention feature matrix, where the first attention feature matrix may be input to the adaptive attention space-time diagram convolutional loop network to obtain the first hail recognition result corresponding to the hail to-be-early-warned area. In one embodiment, the hail identification model may employ an adaptive attention-space-graph convolutional loop network, the first attention feature matrix is input into the adaptive attention-space-graph convolutional loop network, the spatiotemporal correlation and heterogeneity of hail are extracted, and the hail identification result is output.
In specific implementation, the first order chebyshev polynomial expansion of the high-dimensional graph convolution operation is:
wherein, is G For the graph rolling operation, the feature sequence data X is represented by a parameter matrix A θ Filtered on graph G.Is part of a Laplace matrix, D is a degree matrix, A is an adjacency matrix, Θ and b are respectively a learnable weight and bias, I N Is an identity matrix. In addition:
the sensing mechanism in the GRU (threshold cyclic unit) is replaced by the self-adaptive attention attempt convolution, and the specific calculation flow is as follows:
A s ＝softmax(g 1 ⊙A+g 2 ⊙QK T )V+b；
z t ＝σ(A s [X t ,h t-1 ]W z +b z )；
r t ＝σ(A s [X t ,h t-1 ]W r +b r )；
wherein X is t And h t Respectively, the input and the output of the time step t, sigma is a sigmoid activation function, tanh is an activation function,is a parameter that can be learned in the model.
In one embodiment, in order to enable the hail recognition model to better recognize the hail, training is further required to be performed on the hail recognition model to obtain the hail recognition model with better recognition effect, and specifically, the training process of the hail recognition model is as follows (one) to (three):
firstly, acquiring a training data set and an actual hail observation result; wherein the actual hail observations are used to characterize training dataset labels as well as detection ranges. Wherein the training data set comprises a cloud top bright temperature, a bright temperature gradient, a hailstorm index and the like. In one embodiment, the training data set may be determined based on historical ground station observations. And (3) according to the hail site observation data corresponding to the storm monomers, the specific labels are 1 and 0, and whether hail weather occurs in the storm monomer area is indicated.
And secondly, performing self-adaptive attention mechanism calculation on the adjacent matrix corresponding to the training data set to obtain a second attention feature matrix, and performing hail identification on the storm single region based on the second attention feature matrix through a hail identification model to obtain a second hail identification result corresponding to the storm single region. In particular, reference may be made to the foregoing step S104 and step S106, which are not described in detail in the embodiment of the present invention.
(iii) utilizing a cross entropy loss function to adjust network parameters of the hail identification model based on the actual hail observation and the second hail identification; the cross entropy loss function is minimized by adopting an Adam algorithm, and L2 regularization is added into the cross entropy loss function. In one embodiment, a cross entropy loss function training model is selected, and parameters are optimized to obtain a hail identification model, wherein the cross entropy loss function is as follows:
where P (i) is the probability distribution of the discrete variable i.
Furthermore, adam minimized loss function is adopted, and L2 regularization is added, and the specific steps comprise:
m t ＝β 1 m t-1 +(1-β 1 )g t ；
wherein the gradient g t Is a loss functionFor theta t Deviation determination, m t G is g t Moment t moment momentum form next step distance estimation, v t G is g t Second order distance estimation in moment of t momentum form, < >>And->For the first-order distance and the second-order distance estimation after the deviation correction, the above steps are solvedParameter update θ t Obtaining theta t+1 Epsilon is a random parameter, beta 1 、β 2 Is a learnable parameter.
In one embodiment, after training to obtain the hail identification model, the hail identification model may be evaluated, and specifically, the following (a) to (c) may be referred to:
(a) A first number H that the actual hail observation is true and the first hail identification is true, a second number F that the actual hail observation is false and the first hail identification is true, and a third number M that the actual hail observation is true and the first hail identification is false are determined. Wherein, the actual hail observation result is true, namely, hail actually occurs, the actual hail observation result is false, namely, hail does not actually occur, the first hail recognition result is true, namely, the hail recognition model predicts that hail occurs, and the first hail recognition result is false, namely, the hail recognition model predicts that hail does not occur.
(b) Based on the first number, the second number, and the third number, a critical success index, a hit probability index, and a false alarm index are determined, respectively. In one embodiment, the critical success index, hit probability index, and false alarm index may be determined separately according to the following formulas:
Wherein TS is critical success index, POD is hit probability index, and FAR is false alarm index.
(c) And evaluating the hail identification model by using one or more of a critical success index, a hit probability index and a false alarm index to obtain an evaluation result of the hail identification model.
For example, the critical success index, hit probability index, false alarm index may be sent to a designated associated terminal for display, so that the user may evaluate the hail identification model based on the critical success index, hit probability index, false alarm index.
For example, a critical success index threshold, a hit probability index threshold, and a false alarm index threshold may be preconfigured, and the critical success index, the hit probability index, and the false alarm index are compared with the critical success index threshold, the hit probability index threshold, and the false alarm index threshold, respectively.
In summary, the hail identifying method provided by the embodiment of the invention has at least the following characteristics:
(1) The satellite data is used for identifying hail early warning areas, the spatial distribution and the observation time frequency of hail occurrence scale are effectively monitored, hail clouds are completely monitored from the initial occurrence stage to the development process to the end of dissipation, and a new identification method is provided for hail early warning.
(2) The fusion distance adjacency matrix calculation self-adaptive attention mechanism combines the distance adjacency matrix and the attention self-adaptation, the distance adjacency matrix replaces position embedded information in a transducer and is input into the attention mechanism as part of characteristic information, the fusion distance adjacency matrix calculation self-adaptive attention mechanism models the global dependency relationship of all nodes, and fine-grained data attributes among the nodes are captured.
(3) In order to adapt to a specific hail identification task of satellite data, a graph neural network and a threshold memory network (GRU) are selected and combined with a space-time graph neural cycle network, similarity among nodes is measured in a self-adaptive mode and is input into the neural network, time correlation among satellite sequence data is learned by the GRU, gradient explosion is avoided, space and time correlation of specific nodes in a satellite sequence are captured by the space-time graph neural cycle network, data information in the satellite sequence is fully learned, and hail early warning accuracy is improved.
In order to overcome the defects in the prior art, an adaptive attention-seeking convolution model is established by taking FY-4A satellite inversion products as input data; in addition, because satellite data has complex space-time correlation, reasonable construction and design of a model structure are important, and based on the space-time diagram convolution cyclic network (AASTGCRN) model based on self-adaptive attention is provided by the embodiment of the invention to solve the hail recognition problem. The embodiment of the invention combines the self-adaptive attention, the graph convolution network and the threshold memory network to perform space-time attribute analysis on satellite data, thereby improving hail detection capability of a research area and supporting future practical application thereof compared with the prior tool.
For ease of understanding, referring to the flow chart of another hail identification method shown in fig. 2, the method mainly includes the following steps S202 to S210:
and step S202, acquiring ground automatic station observation data and satellite data of a hail to-be-early-warning area for analysis.
And step S204, acquiring storm characteristic parameters of the hail to-be-early-warned area, and preprocessing data.
Step S206, constructing a training subset and a testing subset, and fusing distance adjacency matrix to calculate an adaptive attention mechanism for extracting satellite feature structure information.
Step S208, the training data after preprocessing is input into a self-adaptive attention time space graph convolution circulation network, the time-space correlation and heterogeneity when hail occurs are extracted, and a hail identification result is output.
And S210, training and optimizing a model, and verifying the generalization performance of the model by using a test set to realize hail weather early warning.
In summary, in the hail identifying method provided by the embodiment of the present invention, first, the hail occurrence area matching satellite product is obtained through the ground automatic station data and the past half hour to one half hour data is extracted. And secondly, integrating the self-adaptive attention into the inter-node initialization adjacency matrix for highlighting the node specific attribute mode. Finally, respectively considering time and space information extraction, integrating self-adaptive attention into a graph convolution network, dynamically capturing the dependency relationship between adjacent nodes, and describing time characteristics by adopting GRU (generic routing unit) for time information extraction. Based on the three parts, the space-time correlation of the fine granularity in the hail feature sequence is automatically captured, and the hail early warning duration can be improved and the hail early warning recognition accuracy can be improved by utilizing satellite data.
For the hail identifying method provided in the foregoing embodiment, the embodiment of the present invention provides a hail identifying device, referring to a schematic structural diagram of a hail identifying device shown in fig. 3, the device mainly includes the following parts:
the data analysis module 302 is configured to determine a hail to-be-early-warning area based on current ground station observation data, and analyze satellite data of the hail to-be-early-warning area to obtain a target data set;
the attention module 304 is configured to perform adaptive attention mechanism calculation on an adjacent matrix corresponding to the target data set, so as to obtain a first attention feature matrix;
the hail recognition module 306 is configured to perform hail recognition on a hail to-be-early-warning area based on a first attention feature matrix through a pre-trained hail recognition model, and obtain a first hail recognition result corresponding to the hail to-be-early-warning area.
According to the hail identification device provided by the embodiment of the invention, the spatial distribution and the observation time frequency of the hail occurrence scale can be effectively monitored on the basis of satellite data of the hail to-be-early-warning area, the hail cloud is completely monitored from the initial occurrence stage to the development process to the end of dissipation, a new identification method is provided for hail early warning, and accordingly early warning of hail is realized; in addition, a self-adaptive attention mechanism is calculated by fusing the distance adjacency matrix, the distance adjacency matrix and the attention self-adaptation are combined together, the distance adjacency matrix is used as part of characteristic information to be input into the self-adaptive attention mechanism to obtain a first attention characteristic matrix, and fine-granularity data attributes among nodes are captured, so that the hail recognition accuracy is improved; in addition, a first hail recognition result is obtained based on a first attention feature matrix through a pre-trained hail recognition model, the spatial and time correlation of specific nodes in satellite data is captured through the hail recognition model, data information in the satellite data is fully learned, and the hail recognition accuracy is further improved.
In one embodiment, the data parsing module 302 is further configured to:
extracting initial cloud top bright temperature from satellite data of a hail to-be-early-warning area; extracting bright temperature data from satellite data of a hail to-be-early-warned area, and determining an initial bright temperature gradient based on the bright temperature data; extracting appointed channel data from satellite data of a hail to-be-early-warned area, and determining an initial hailstorm index based on the appointed channel data;
respectively carrying out normalization treatment on the initial cloud top bright temperature, the initial bright temperature gradient and the initial hailstorm index to obtain a target cloud top bright temperature, a target bright temperature gradient and a target hailstorm index;
and taking one or more of the target cloud top bright temperature, the target bright temperature gradient and the target hailstorm index as a target data set.
In one embodiment, the attention module 304 is further configured to:
constructing an initial satellite characteristic sequence matrix based on the target data set, and determining an adjacent matrix corresponding to the initial satellite characteristic sequence matrix;
performing linear transformation on the initial satellite feature sequence matrix by using a plurality of preset weights to obtain a plurality of target satellite feature sequence matrices;
and performing self-adaptive attention mechanism calculation based on the adjacent matrix, the initial satellite feature sequence matrix and each target satellite feature sequence matrix to obtain a first attention feature matrix.
In one embodiment, the attention module 304 is further configured to:
the attention profile matrix is determined according to the following formula:
Attention(Q,K,V)＝softmax(QK T )V+b；
wherein Attention (Q, K, V) is the first Attention feature matrix, g 1 ,g 2 ∈R N×N ，As a learnable parameter, A is E R N×N Is an adjacent matrix, X is an initial satellite characteristic sequence matrix, Q is a first target satellite characteristic sequence matrix, K is a second target satellite characteristic sequence matrix, V is a third target satellite characteristic sequence matrix, QK T As the attention matrix, +.. />
In one embodiment, hail identification module 306 is further to:
and performing high-dimensional graph convolution operation and perception mechanism processing on the first attention feature matrix through a pre-trained hail identification model to obtain a first hail identification result corresponding to the hail to-be-early-warned area.
In one embodiment, the method further comprises a model training module for:
acquiring a training data set and an actual hail observation result; wherein the actual hail observations are used to characterize the training dataset labels and detection ranges;
performing self-adaptive attention mechanism calculation on an adjacent matrix corresponding to the training data set to obtain a second attention feature matrix, and performing hail identification on the storm single region based on the second attention feature matrix through a hail identification model to obtain a second hail identification result corresponding to the storm single region;
Adjusting network parameters of the hail identification model based on the actual hail observation result and the second hail identification result by using a cross entropy loss function; the cross entropy loss function is minimized by adopting an Adam algorithm, and L2 regularization is added into the cross entropy loss function.
In one embodiment, the method further comprises a model evaluation module for:
determining a first number of actual hail observations as true and a first hail identification as true, and a second number of actual hail observations as false and a first hail identification as true, and a third number of actual hail observations as true and a first hail identification as false;
determining a critical success index, a hit probability index and a false alarm index based on the first number, the second number and the third number, respectively;
and evaluating the hail identification model by using one or more of a critical success index, a hit probability index and a false alarm index to obtain an evaluation result of the hail identification model.
The device provided by the embodiment of the present invention has the same implementation principle and technical effects as those of the foregoing method embodiment, and for the sake of brevity, reference may be made to the corresponding content in the foregoing method embodiment where the device embodiment is not mentioned.
The embodiment of the invention provides electronic equipment, which comprises a processor and a storage device; the storage means has stored thereon a computer program which, when executed by the processor, performs the method of any of the embodiments described above.
Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, where the electronic device 100 includes: a processor 40, a memory 41, a bus 42 and a communication interface 43, the processor 40, the communication interface 43 and the memory 41 being connected by the bus 42; the processor 40 is arranged to execute executable modules, such as computer programs, stored in the memory 41.
The memory 41 may include a high-speed random access memory (RAM, random Access Memory), and may further include a non-volatile memory (non-volatile memory), such as at least one magnetic disk memory. The communication connection between the system network element and the at least one other network element is achieved via at least one communication interface 43 (which may be wired or wireless), which may use the internet, a wide area network, a local network, a metropolitan area network, etc.
Bus 42 may be an ISA bus, a PCI bus, an EISA bus, or the like. The buses may be classified as address buses, data buses, control buses, etc. For ease of illustration, only one bi-directional arrow is shown in FIG. 4, but not only one bus or type of bus.
The memory 41 is configured to store a program, and the processor 40 executes the program after receiving an execution instruction, and the method executed by the apparatus for flow defining disclosed in any of the foregoing embodiments of the present invention may be applied to the processor 40 or implemented by the processor 40.
The processor 40 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuitry in hardware or instructions in software in processor 40. The processor 40 may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU), a network processor (Network Processor, NP), etc.; but may also be a digital signal processor (Digital Signal Processing, DSP for short), application specific integrated circuit (Application Specific Integrated Circuit, ASIC for short), off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA for short), or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. The disclosed methods, steps, and logic blocks in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory 41 and the processor 40 reads the information in the memory 41 and in combination with its hardware performs the steps of the method described above.
The computer program product of the readable storage medium provided by the embodiment of the present invention includes a computer readable storage medium storing a program code, where the program code includes instructions for executing the method described in the foregoing method embodiment, and the specific implementation may refer to the foregoing method embodiment and will not be described herein.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Finally, it should be noted that: the above examples are only specific embodiments of the present invention, and are not intended to limit the scope of the present invention, but it should be understood by those skilled in the art that the present invention is not limited thereto, and that the present invention is described in detail with reference to the foregoing examples: any person skilled in the art may modify or easily conceive of the technical solution described in the foregoing embodiments, or perform equivalent substitution of some of the technical features, while remaining within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. A hail identification method, comprising:
determining a hail to-be-early-warning area based on current ground station observation data, and analyzing satellite data of the hail to-be-early-warning area to obtain a target data set;
performing self-adaptive attention mechanism calculation on an adjacent matrix corresponding to the target data set to obtain a first attention characteristic matrix;
And performing hail recognition on the hail to-be-early-warning area based on the first attention feature matrix through a pre-trained hail recognition model to obtain a first hail recognition result corresponding to the hail to-be-early-warning area.
2. The hail identification method according to claim 1, wherein analyzing satellite data of the hail to-be-early-warned area to obtain a target data set comprises:
extracting initial cloud top bright temperature from satellite data of the hail to-be-early-warning area; extracting bright temperature data from satellite data of the hail to-be-early-warning area, and determining an initial bright temperature gradient based on the bright temperature data; extracting designated channel data from satellite data of the hail to-be-early-warned area, and determining an initial hail storm index based on the designated channel data;
respectively carrying out normalization processing on the initial cloud top bright temperature, the initial bright temperature gradient and the initial hailstorm index to obtain a target cloud top bright temperature, a target bright temperature gradient and a target hailstorm index;
and taking one or more of the target cloud top bright temperature, the target bright temperature gradient and the target hailstorm index as a target data set.
3. The hail identification method according to claim 1, wherein performing adaptive attention mechanism calculation on an adjacency matrix corresponding to the target data set to obtain a first attention feature matrix comprises:
Constructing an initial satellite characteristic sequence matrix based on the target data set, and determining an adjacent matrix corresponding to the initial satellite characteristic sequence matrix;
performing linear transformation on the initial satellite feature sequence matrix by using a plurality of preset weights to obtain a plurality of target satellite feature sequence matrices;
and carrying out self-adaptive attention mechanism calculation based on the adjacent matrix, the initial satellite feature sequence matrix and each target satellite feature sequence matrix to obtain a first attention feature matrix.
4. A hail identification method according to claim 3 wherein performing an adaptive attention mechanism calculation based on the adjacency matrix, the initial satellite signature sequence matrix and each of the target satellite signature sequence matrices to obtain a first attention signature matrix comprises:
the attention profile matrix is determined according to the following formula:
Q＝linear(X)＝AXW Q
K＝linear(X)＝AXW K
V＝linear(X)＝AXW V
Attention(Q,K,V)＝softmax(QK T )V+b；
wherein Attention (Q, K, V) is the first Attention feature matrix, g 1 ,g 2 ∈R N×N ，As a learnable parameter, A is E R N×N Is an adjacent matrix, X is an initial satellite characteristic sequence matrix, Q is a first target satellite characteristic sequence matrix, K is a second target satellite characteristic sequence matrix, V is a third target satellite characteristic sequence matrix, QK T As the attention matrix, +..
5. The hail identification method of claim 1 wherein the hail identification model employs an adaptive attention space-time graph convolutional loop network;
through a pre-trained hail recognition model, hail recognition is carried out on the hail to-be-early-warning area based on the first attention feature matrix, and a first hail recognition result corresponding to the hail to-be-early-warning area is obtained, and the hail recognition method comprises the following steps:
and inputting the first attention characteristic matrix into the self-adaptive attention time-space diagram convolution circulation network to obtain a first hail identification result corresponding to the hail to-be-early-warned area.
6. The hail identification method according to any one of claims 1-5 wherein the training step of the hail identification model comprises:
acquiring a training data set and an actual hail observation result; wherein the actual hail observations are used to characterize the training dataset labels and detection ranges;
performing self-adaptive attention mechanism calculation on an adjacent matrix corresponding to the training data set to obtain a second attention feature matrix, and performing hail identification on a storm monomer region based on the second attention feature matrix through a hail identification model to obtain a second hail identification result corresponding to the storm monomer region;
Adjusting network parameters of the hail identification model based on the actual hail observation result and the second hail identification result by using a cross entropy loss function; the cross entropy loss function is minimized by adopting an Adam algorithm, and L2 regularization is added into the cross entropy loss function.
7. The hail identification method of claim 6, further comprising:
determining a first number of actual hail observations as true and the first hail recognition results as true, and a second number of actual hail observations as false and the first hail recognition results as true, and a third number of actual hail observations as true and the first hail recognition results as false;
determining a critical success index, a hit probability index, and a false alarm index, respectively, based on the first number, the second number, and the third number;
and evaluating the hail identification model by using one or more of the critical success index, the hit probability index and the false alarm index to obtain an evaluation result of the hail identification model.
8. A hail identification device, comprising:
The data analysis module is used for determining a hail to-be-early-warning area based on current ground station observation data and analyzing satellite data of the hail to-be-early-warning area to obtain a target data set;
the attention module is used for carrying out self-adaptive attention mechanism calculation on the adjacent matrix corresponding to the target data set to obtain a first attention characteristic matrix;
and the hail recognition module is used for carrying out hail recognition on the hail to-be-early-warning area based on the first attention feature matrix through a pre-trained hail recognition model to obtain a first hail recognition result corresponding to the hail to-be-early-warning area.
9. An electronic device comprising a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which, when invoked and executed by a processor, cause the processor to implement the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310821981.5A CN116740586A (en) | 2023-07-05 | 2023-07-05 | Hail identification method, hail identification device, electronic equipment and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310821981.5A CN116740586A (en) | 2023-07-05 | 2023-07-05 | Hail identification method, hail identification device, electronic equipment and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116740586A true CN116740586A (en) | 2023-09-12 |
Family
ID=87913352
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310821981.5A Pending CN116740586A (en) | 2023-07-05 | 2023-07-05 | Hail identification method, hail identification device, electronic equipment and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116740586A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117214916A (en) * | 2023-11-08 | 2023-12-12 | 北京英视睿达科技股份有限公司 | Short-time hail prediction method and system based on satellite remote sensing observation data |
-
2023
- 2023-07-05 CN CN202310821981.5A patent/CN116740586A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117214916A (en) * | 2023-11-08 | 2023-12-12 | 北京英视睿达科技股份有限公司 | Short-time hail prediction method and system based on satellite remote sensing observation data |
| CN117214916B (en) * | 2023-11-08 | 2024-04-05 | 北京英视睿达科技股份有限公司 | Short-time hail prediction method and system based on satellite remote sensing observation data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110113226B (en) | Method and device for detecting equipment abnormity | |
| CN112258093B (en) | Data processing method and device for risk level, storage medium and electronic equipment | |
| CN111291264B (en) | Access object prediction method and device based on machine learning and computer equipment | |
| US11520672B2 (en) | Anomaly detection device, anomaly detection method and storage medium | |
| CN112149898B (en) | Training of failure rate prediction model, failure rate prediction method and related device | |
| CN116432091B (en) | Equipment fault diagnosis method based on small sample, construction method and device of model | |
| CN116740586A (en) | Hail identification method, hail identification device, electronic equipment and computer readable storage medium | |
| CN114638633A (en) | Abnormal flow detection method and device, electronic equipment and storage medium | |
| CN113624694A (en) | Inversion method and device for atmospheric methane concentration | |
| CN113536066A (en) | Data anomaly detection algorithm determination method and device and computer equipment | |
| CN117872166B (en) | Method and device for detecting thermal runaway of energy storage battery and electronic equipment | |
| CN113487223B (en) | Risk assessment method and system based on information fusion | |
| CN116821777B (en) | Novel basic mapping data integration method and system | |
| CN110866672B (en) | Data processing method, device, terminal and medium | |
| CN117272145A (en) | Health state evaluation method and device of switch machine and electronic equipment | |
| CN113110961A (en) | Equipment abnormality detection method and device, computer equipment and readable storage medium | |
| CN111091194A (en) | Operation system identification method based on CAVWB _ KL algorithm | |
| CN112668637B (en) | Training method, recognition method and device of network model and electronic equipment | |
| CN110969209B (en) | Stranger identification method and device, electronic equipment and storage medium | |
| CN115757365A (en) | Multi-dimensional time sequence data anomaly detection method, model training method and device | |
| CN113989632A (en) | Bridge detection method and device for remote sensing image, electronic equipment and storage medium | |
| CN112966988A (en) | XGboost model-based data evaluation method, device, equipment and storage medium | |
| CN115514621B (en) | Fault monitoring method, electronic device and storage medium | |
| CN116580025B (en) | Defect filtering method, device, computer equipment and storage medium | |
| KR101367418B1 (en) | Apparatus and method for detecting estrus audio of livestock using a feature vector generation algorithm |
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 |