Marine vortex identification and prediction method and system

Abstract

The invention discloses an ocean vortex identification and prediction method and system. The method comprises the following steps: firstly, acquiring an ocean vortex high-frequency ground wave radar signal; constructing a training set and a test set according to the ocean vortex high-frequency ground wave radar signal; training the training set through a black-wing plinual optimization algorithm and a kernel extreme learning machine to obtain an optimal kernel function parameter and a regularization factor, and constructing to obtain an initial model; performing feature extraction on the test set and testing the initial model based on the extracted features to obtain a target model meeting the test requirements; and according to the target model, performing identification prediction on the newly obtained ocean vortex high-frequency ground wave radar signal to obtain an ocean vortex identification result. The embodiment of the invention can improve the accuracy and timeliness of the prediction result, and can be widely applied to the technical field of computers.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to a method and system for identifying and predicting ocean vortices. Background Art

[0002] Ocean vortices are rotating water structures formed in the ocean. They carry various nutrients and nutrient elements, including organisms, and have an important impact on global ocean circulation, marine ecosystems, and weather and climate. Therefore, accurate identification and prediction of ocean vortices are crucial for marine resource development and maritime traffic safety. However, traditional observation methods are limited by time and spatial resolution, making it difficult to comprehensively and accurately monitor the formation and evolution of vortices. At the same time, numerical simulation methods have certain limitations when simulating ocean vortices: insufficiently refined parameterization schemes, limited computing resources, etc. These problems affect the accuracy and timeliness of predictions.

[0003] A mainstream ocean eddy identification technology is a direct identification technology based on satellite altimeter data. This technology has been used by many teams around the world for research and processing in the past two decades, and provides global or regional sea surface height time series products. The main method is to use the satellite radar altimeter echo signal to calculate the distance from the satellite to a certain point under the satellite, and then combine the orbital altitude of WeChat to calculate the height of the point relative to the surface reference surface, and finally obtain the characteristic image of the ocean surface in the area, and then extract the characteristic lines of abnormal height of the ocean surface on the image to achieve the purpose of eddy identification. At present, the technology for identifying anomalies in satellite altitude data is relatively mature, and can be processed automatically, which is suitable for real-time monitoring and large-scale data analysis.

[0004] Traditional hydrological data are mostly time series data, so the standard recurrent neural network represented by LSTM (Long Short-Term Memory Network) is still the core model structure of deep learning hydrological forecasting at this stage. This model is a "data hungry" model, and research relies on a large amount of fine data. The resolution of satellite altimeter data images limits the minimum vortex scale that can be identified, so the accuracy of identifying small-scale or irregularly shaped ocean vortices is limited. Secondly, the phenomenon of ocean vortices is constantly changing dynamically with human and non-human environmental changes, and existing vortex identification and prediction technologies lack the ability to respond to this dynamic change in a timely manner, especially in basins that are greatly affected by human activities. Hydrological modeling is still a problem. Summary of the invention

[0005] The main purpose of the embodiments of the present invention is to provide a method and system for identifying and predicting ocean vortices, which can improve the accuracy and timeliness of prediction results.

[0006] To achieve the above object, an embodiment of the present invention provides a method for identifying and predicting ocean vortices, comprising the following steps:

[0007] Obtain high-frequency ground wave radar signals of ocean eddies;

[0008] Constructing a training set and a test set according to the ocean eddy high-frequency ground wave radar signal;

[0009] The training set is trained by using a black kite optimization algorithm and a kernel extreme learning machine to obtain optimal kernel function parameters and regularization factors, and an initial model is constructed;

[0010] By extracting features from the test set and testing the initial model based on the extracted features, a target model that meets the test requirements is obtained;

[0011] According to the target model, the newly acquired ocean vortex high-frequency ground wave radar signal is identified and predicted to obtain an ocean vortex identification result.

[0012] In some embodiments, the method further comprises:

[0013] After completing the identification and prediction of ocean eddies, the knowledge graph is used to verify the eddy identification data.

[0014] In some embodiments, the expression of the black kite optimization algorithm is: i =BK lh +rand(BK ub -BK lb ), where X i represents the position of the i-th individual in the iterative process of the optimization algorithm; BK lb represents the lower bound of the j-th dimension black-winged kite; rand() represents a randomly selected value between [0, 1]; BK ub Represents the upper bound of the j-th dimension black-winged kite;

[0015] During the initialization process, the black kite optimization algorithm selects the individual with the best fitness value as the leader X in the initial group. L ; The leader X L The expression is:

[0016] f best =min(f(X i ))

[0017] X L =X(find(f best = = f(X i )))

[0018] Among them, f bestrepresents the minimum fitness in the group; X L represents the leader in the initial group, which is the individual with the best fitness value; f(X i ) represents the fitness value of the ith individual; X() represents the position of the individual that meets the condition of the function;

[0019] The attack behavior of the black kite optimization algorithm is expressed as:

[0020]

[0021] in, represents the position of the i-th black-winged kite in the j-th dimension and (t+1)-th iteration step; r represents a random number; p is a calculated constant; T is the total number of iterations; t is the number of iterations completed so far; n is a control parameter related to the number of iterations, and its value is determined by the following expression:

[0022]

[0023] The expression of the migration behavior of the black kite optimization algorithm is:

[0024]

[0025]

[0026] in, represents the leading scorer of the j-th dimension black-wing kite at the t-th iteration so far; F i represents the j-th dimension current position obtained by any black-winged kite in the t-th iteration; F ri represents the fitness value of any black-winged kite at a random position in the jth dimension in the tth iteration; m is a control parameter whose value is determined by the following expression:

[0027]

[0028] In some embodiments, after the identification and prediction of the ocean vortex is completed, the identification data of the vortex is verified using the knowledge graph, including the following steps:

[0029] Collect and organize the basic attributes of the vortex knowledge graph; the entities of the vortex knowledge graph are the monitoring points on the sea surface, the entity relationship is the spatial distribution relationship of the monitoring points, and the entity attributes are object attributes and data attributes. The object attributes include the monitoring physical quantity of the monitoring point, the warning level of a single measuring point and the location to which it belongs, and the data attributes include three-dimensional coordinates, current monitoring data and historical monitoring data;

[0030] Select a vortex anomaly point as the reference point, retrieve all monitoring points that are spatially adjacent to the reference point based on the multi-level relationship query capability of the knowledge graph, and determine whether these adjacent monitoring points are abnormal; the characteristics of the vortex anomaly point include abnormal sea surface height, abnormal sea surface temperature, and abnormal flow velocity change;

[0031] Continue to search and determine whether there is an abnormal monitoring point outside the abnormal measuring point group that has a spatially adjacent relationship with any monitoring point in the abnormal measuring point group. If so, the new abnormal monitoring point is incorporated into the abnormal measuring point group for processing until there is no abnormal monitoring point outside the abnormal measuring point group that has a spatially adjacent relationship with it;

[0032] All abnormal monitoring points in the sea surface vortex monitoring knowledge graph are classified into different abnormal monitoring point groups;

[0033] Select any two abnormal measurement point groups in the knowledge graph, and use the relationship query capability of the graph database to determine whether there is a data sequence similarity relationship between the corresponding monitoring points in the two abnormal measurement point groups. If the number of monitoring points with a data sequence similarity relationship is greater than 50% of the total number of monitoring points in the two abnormal measurement point groups, then the two abnormal measurement point groups are determined to have a data sequence similarity relationship;

[0034] Perform pairwise analysis on all abnormal measurement point groups in the knowledge graph to obtain abnormal measurement point groups with similar data sequence relationships;

[0035] The detected vortex monitoring points are compared with the results obtained by pattern recognition to verify the vortex identification results; after the identification is completed, a regression prediction model composed of an optimized sparrow algorithm, a convolutional neural network, and a bidirectional gated recursive unit is used to further predict the vortex.

[0036] In some embodiments, selecting a vortex abnormal point as a reference point, retrieving all monitoring points that have a spatially adjacent relationship with the reference point according to the multi-level relationship query capability of the knowledge graph, and determining whether these adjacent monitoring points are abnormal, includes the following steps:

[0037] By building an abnormal mesoscale eddy sample library, and then using the abnormal sea surface height and sea surface temperature data, feature extraction and learning are performed through a densely connected network to obtain fused feature data that integrates sea surface height and sea surface temperature information;

[0038] The improved U-Net image segmentation model is used to build recognition and judge the abnormal situation of the vortex at the monitoring point. The construction process of the U-Net model is as follows: extract features from the monitoring data, including the vortex center point, vortex radius, and ocean parameters; associate the extracted features with the entities in the knowledge graph and output them as the segmentation mask of the vortex generation area; U-Net adopts an encoder-decoder structure, in which the encoder is used to extract features and the decoder is used to map features to segmentation results; add an attention mechanism to U-Net to focus on key areas;

[0039] Use the labeled segmentation data to train the U-Net model, supervise the learning of vortex generation monitoring points, and use the cross entropy loss function and Adam optimizer to optimize the model;

[0040] Among them, the cross entropy loss function The expression is:

[0041]

[0042] Where N represents the number of samples; y i represents the true label; Represents the probability predicted by the model.

[0043] In some embodiments, the detected vortex monitoring points are compared with the results obtained by pattern recognition to verify the vortex recognition results; after the recognition is completed, the vortex is further predicted using a regression prediction model, including the following steps:

[0044] 1D CNN is used to extract the features of the original data, explore the implicit relationship between multiple features and filter out noise and unstable components. The information processed by CNN is then input into the bidirectional gated recurrent unit Bi-GRU for the next step of prediction.

[0045] In the prediction process of the bidirectional gated recurrent unit Bi-GRU, the output information at a certain time t is the sum of the outputs of the forward hidden layer and the backward hidden layer;

[0046] The CNN–BiGRU model is optimized using the optimized sparrow algorithm;

[0047] The normEve algorithm is introduced to optimize the obtained prediction model and complete further prediction of the vortex.

[0048] In some embodiments, the optimization of the CNN-BiGRU model using the optimized sparrow algorithm comprises the following steps:

[0049] Initializing population parameters, wherein the population parameters include a maximum number of iterations and a population size;

[0050] Initialize CNN-BiGRU parameters and set the parameter optimization interval;

[0051] Calculate the current fitness of each sparrow, find the best and worst fitness values, and the corresponding positions;

[0052] After each iteration, the fitness of each sparrow is recalculated, and a dynamic learning factor is introduced to update the position of the discoverer; the Levy strategy is introduced to update the position of the follower; and the position of the vigilant is updated;

[0053] According to the current status of the sparrows, update the optimal and worst fitness values ​​and their positions experienced by the entire sparrow population;

[0054] Determine whether the maximum iteration is met. If so, assign the optimization result to CNN-BiGRU. Otherwise, continue iterating until the maximum iteration is met.

[0055] Input the optimized parameters to build the CNN-BiGRU prediction model.

[0056] Another aspect of the embodiments of the present invention further provides an ocean vortex identification and prediction system, comprising:

[0057] The first module is used to obtain high-frequency ground wave radar signals of ocean eddies;

[0058] The second module is used to construct a training set and a test set according to the ocean eddy high-frequency ground wave radar signal;

[0059] The third module is used to train the training set by using a black kite optimization algorithm and a kernel extreme learning machine to obtain optimal kernel function parameters and regularization factors, and to construct an initial model;

[0060] The fourth module is used to extract features from the test set and test the initial model based on the extracted features to obtain a target model that meets the test requirements;

[0061] The fifth module is used to identify and predict the newly acquired ocean vortex high-frequency ground wave radar signal according to the target model to obtain the ocean vortex identification result.

[0062] To achieve the above objective, another aspect of an embodiment of the present invention provides an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the above-mentioned method when executing the computer program.

[0063] To achieve the above objective, another aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned method is implemented.

[0064] The embodiment of the present invention also discloses a computer program product or a computer program, which includes a computer instruction stored in a computer-readable storage medium. A processor of a computer device can read the computer instruction from the computer-readable storage medium, and the processor executes the computer instruction, so that the computer device executes the above method.

[0065] The embodiments of the present invention include at least the following beneficial effects: the present invention provides a method and system for identifying and predicting ocean vortices, the scheme first obtains ocean vortex high-frequency ground wave radar signals; constructs a training set and a test set according to the ocean vortex high-frequency ground wave radar signals; trains the training set by a black kite optimization algorithm and a kernel extreme learning machine to obtain optimal kernel function parameters and regularization factors, and constructs an initial model; extracts features from the test set and tests the initial model based on the extracted features to obtain a target model that meets the test requirements; and identifies and predicts the newly acquired ocean vortex high-frequency ground wave radar signals according to the target model to obtain an ocean vortex identification result. The embodiments of the present invention can improve the accuracy and timeliness of the prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic diagram of an implementation environment provided by an embodiment of the present invention;

[0067] Figure 2 is a flow chart of overall steps provided by an embodiment of the present invention;

[0068] Figure 3 It is a workflow diagram of the TSMFDE-BKA-KELM model for vortex identification and prediction provided by an embodiment of the present invention;

[0069] Figure 4 is a structural diagram of a bidirectional gated recurrent unit provided by an embodiment of the present invention;

[0070] Figure 5 is an optimization process diagram of the CNN-BiGRU model provided by an embodiment of the present invention;

[0071] Figure 6 It is a schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0072] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of the present invention. They are only examples of devices and methods consistent with some aspects of the embodiments of the present invention as detailed in the attached claims.

[0073] It is understood that the terms "first", "second", etc. used in the present invention may be used herein to describe various concepts, but unless otherwise specified, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of the present invention, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0074] The terms "at least one", "multiple", "each", "any", etc. used in the present invention, at least one includes one, two or more, multiple includes two or more, each refers to each of the corresponding multiple, and any refers to any one of the multiple.

[0075] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein are only for the purpose of describing the embodiments of the present invention and are not intended to limit the present invention.

[0076] The method and system for identifying and predicting ocean vortices provided by the embodiments of the present invention relate to the field of computer technology. The method for identifying and predicting ocean vortices provided by the embodiments of the present invention can be applied to a terminal, can be applied to a server, or can be software running in a terminal or a server. In some embodiments, the terminal can be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, and a car terminal, etc., but is not limited to this; the server side can be configured as an independent physical server, or a server cluster or distributed system composed of multiple physical servers, and can also be configured as a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms. The server can also be a node server in a blockchain network; the software can be an application that implements the method for identifying and predicting ocean vortices, etc., but is not limited to the above forms.

[0077] The present invention can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present invention can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present invention can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0078] like Figure 1 FIG. 1 is a schematic diagram of an implementation environment provided by an embodiment of the present invention. Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected to a network wirelessly or wired to complete data transmission and exchange.

[0079] Server 101 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers. It can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN (Content Delivery Network), as well as big data and artificial intelligence platforms.

[0080] In addition, the server 101 can also be a node server in the blockchain network. Blockchain is a new application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, and encryption algorithm.

[0081] The terminal 102 may be a smart phone, a tablet computer, a laptop computer, a desktop computer, a smart speaker, a smart watch, etc. The terminal 102 may also be a vehicle-mounted terminal of various device types described above, but is not limited thereto. The terminal 102 and the server 101 may be directly or indirectly connected via wired or wireless communication, which is not limited in this embodiment of the present invention.

[0082] Based on the example Figure 1 In the implementation environment shown, an embodiment of the present invention provides a method for identifying and predicting ocean vortices. The following is explained using the example of the method for identifying and predicting ocean vortices being applied to a server 101. It can be understood that the method can also be applied to a terminal 102.

[0083] Reference Figure 2 , Figure 2 The flowchart of the method for identifying and predicting ocean vortices applied to a server provided in an embodiment of the present invention, the execution subject of the method can be any of the aforementioned computer devices (including a server or a terminal). Figure 2 , the method may include the following steps:

[0084] Obtain high-frequency ground wave radar signals of ocean eddies;

[0085] Constructing a training set and a test set according to the ocean eddy high-frequency ground wave radar signal;

[0086] The training set is trained by using a black kite optimization algorithm and a kernel extreme learning machine to obtain optimal kernel function parameters and regularization factors, and an initial model is constructed;

[0087] By extracting features from the test set and testing the initial model based on the extracted features, a target model that meets the test requirements is obtained;

[0088] According to the target model, the newly acquired ocean vortex high-frequency ground wave radar signal is identified and predicted to obtain an ocean vortex identification result.

[0089] In some embodiments, the method further comprises:

[0090] After completing the identification and prediction of ocean eddies, the knowledge graph is used to verify the eddy identification data.

[0091] In some embodiments, the expression of the black kite optimization algorithm is: i =BK lh +rand(BK ub -BK lb ), where X i represents the position of the i-th individual in the iterative process of the optimization algorithm; BK lb represents the lower bound of the j-th dimension black-winged kite; rand() represents a randomly selected value between [0, 1]; BK ub Represents the upper bound of the j-th dimension black-winged kite;

[0092] During the initialization process, the black kite optimization algorithm selects the individual with the best fitness value as the leader X in the initial group. L ; The leader X L The expression is:

[0093] f best =min(f(X i ))

[0094] X L =X(find(f best = = f(X i )))

[0095] Among them, f best represents the minimum fitness in the group; X L represents the leader in the initial group, which is the individual with the best fitness value; f(X i ) represents the fitness value of the ith individual; X() represents the position of the individual that meets the condition of the function;

[0096] The attack behavior of the black kite optimization algorithm is expressed as:

[0097]

[0098] in, represents the position of the i-th black-winged kite in the j-th dimension and (t+1)-th iteration step; r represents a random number; p is a calculated constant; T is the total number of iterations; t is the number of iterations completed so far; n is a control parameter related to the number of iterations, and its value is determined by the following expression:

[0099]

[0100] The expression of the migration behavior of the black kite optimization algorithm is:

[0101]

[0102] in, represents the leading scorer of the j-th dimension black-wing kite at the t-th iteration so far; F i represents the j-th dimension current position obtained by any black-winged kite in the t-th iteration; F ri represents the fitness value of any black-winged kite at a random position in the jth dimension in the tth iteration; m is a control parameter whose value is determined by the following expression:

[0103]

[0104] In some embodiments, after the identification and prediction of the ocean vortex is completed, the identification data of the vortex is verified using the knowledge graph, including the following steps:

[0105] Collect and organize the basic attributes of the vortex knowledge graph; the entities of the vortex knowledge graph are the monitoring points on the sea surface, the entity relationship is the spatial distribution relationship of the monitoring points, and the entity attributes are object attributes and data attributes. The object attributes include the monitoring physical quantity of the monitoring point, the warning level of a single measuring point and the location to which it belongs, and the data attributes include three-dimensional coordinates, current monitoring data and historical monitoring data;

[0106] Select a vortex anomaly point as the reference point, retrieve all monitoring points that are spatially adjacent to the reference point based on the multi-level relationship query capability of the knowledge graph, and determine whether these adjacent monitoring points are abnormal; the characteristics of the vortex anomaly point include abnormal sea surface height, abnormal sea surface temperature, and abnormal flow velocity change;

[0107] Continue to search and determine whether there is an abnormal monitoring point outside the abnormal measuring point group that has a spatially adjacent relationship with any monitoring point in the abnormal measuring point group. If so, the new abnormal monitoring point is incorporated into the abnormal measuring point group for processing until there is no abnormal monitoring point outside the abnormal measuring point group that has a spatially adjacent relationship with it;

[0108] All abnormal monitoring points in the sea surface vortex monitoring knowledge graph are classified into different abnormal monitoring point groups;

[0109] Select any two abnormal measurement point groups in the knowledge graph, and use the relationship query capability of the graph database to determine whether there is a data sequence similarity relationship between the corresponding monitoring points in the two abnormal measurement point groups. If the number of monitoring points with a data sequence similarity relationship is greater than 50% of the total number of monitoring points in the two abnormal measurement point groups, then the two abnormal measurement point groups are determined to have a data sequence similarity relationship;

[0110] Perform pairwise analysis on all abnormal measurement point groups in the knowledge graph to obtain abnormal measurement point groups with similar data sequence relationships;

[0111] The detected vortex monitoring points are compared with the results obtained by pattern recognition to verify the vortex identification results; after the identification is completed, a regression prediction model composed of an optimized sparrow algorithm, a convolutional neural network, and a bidirectional gated recursive unit is used to further predict the vortex.

[0112] In some embodiments, selecting a vortex abnormal point as a reference point, retrieving all monitoring points that have a spatially adjacent relationship with the reference point according to the multi-level relationship query capability of the knowledge graph, and determining whether these adjacent monitoring points are abnormal, includes the following steps:

[0113] By building an abnormal mesoscale eddy sample library, and then using the abnormal sea surface height and sea surface temperature data, feature extraction and learning are performed through a densely connected network to obtain fused feature data that integrates sea surface height and sea surface temperature information;

[0114] The improved U-Net image segmentation model is used to build recognition and judge the abnormal situation of the vortex at the monitoring point. The construction process of the U-Net model is as follows: extract features from the monitoring data, including the vortex center point, vortex radius, and ocean parameters; associate the extracted features with the entities in the knowledge graph and output them as the segmentation mask of the vortex generation area; U-Net adopts an encoder-decoder structure, in which the encoder is used to extract features and the decoder is used to map features to segmentation results; add an attention mechanism to U-Net to focus on key areas;

[0115] Use the labeled segmentation data to train the U-Net model, supervise the learning of vortex generation monitoring points, and use the cross entropy loss function and Adam optimizer to optimize the model;

[0116] Among them, the cross entropy loss function The expression is:

[0117]

[0118] Where N represents the number of samples; y i represents the true label; Represents the probability predicted by the model.

[0119] In some embodiments, the detected vortex monitoring points are compared with the results obtained by pattern recognition to verify the vortex recognition results; after the recognition is completed, the vortex is further predicted using a regression prediction model, including the following steps:

[0120] 1D CNN is used to extract the features of the original data, explore the implicit relationship between multiple features and filter out noise and unstable components. The information processed by CNN is then input into the bidirectional gated recurrent unit Bi-GRU for the next step of prediction.

[0121] In the prediction process of the bidirectional gated recurrent unit Bi-GRU, the output information at a certain time t is the sum of the outputs of the forward hidden layer and the backward hidden layer;

[0122] The CNN–BiGRU model is optimized using the optimized sparrow algorithm;

[0123] The normEve algorithm is introduced to optimize the obtained prediction model and complete further prediction of the vortex.

[0124] In some embodiments, the optimization of the CNN-BiGRU model using the optimized sparrow algorithm comprises the following steps:

[0125] Initializing population parameters, wherein the population parameters include a maximum number of iterations and a population size;

[0126] Initialize CNN-BiGRU parameters and set the parameter optimization interval;

[0127] Calculate the current fitness of each sparrow, find the best and worst fitness values, and the corresponding positions;

[0128] After each iteration, the fitness of each sparrow is recalculated, and a dynamic learning factor is introduced to update the position of the discoverer; the Levy strategy is introduced to update the position of the follower; and the position of the vigilant is updated;

[0129] According to the current status of the sparrows, update the optimal and worst fitness values ​​and their positions experienced by the entire sparrow population;

[0130] Determine whether the maximum iteration is met. If so, assign the optimization result to CNN-BiGRU. Otherwise, continue iterating until the maximum iteration is met.

[0131] Input the optimized parameters to build the CNN-BiGRU prediction model.

[0132] The following describes the specific implementation process of the present invention in detail by taking a specific application scenario as an example:

[0133] The embodiment of the present invention uses a high-frequency ground wave radar to collect images. Compared with the images obtained by the traditional satellite altimeter, this image acquisition method has higher resolution and accuracy. In addition, the high-frequency ground wave radar is relatively less affected by factors such as sea surface temperature and light visibility, which is beneficial to obtain more accurate data. The present invention uses data assimilation technology to combine observation data with numerical simulation results, and further corrects the ocean vortex data on the basis of accurate identification and prediction to ensure that the results are timely.

[0134] like Figure 3 As shown, the embodiment of the present invention adopts the time-shift theory to replace the traditional coarse-graining process in the multi-scale fluctuation spread entropy (MFDE), and proposes the time-shift multi-scale fluctuation spread entropy. Through simulation experiments, it is proved that the proposed method has good time series length robustness, noise resistance and feature extraction capabilities, and solves the problem of insufficient coarse-graining of traditional multi-scale entropy. Then, the black-winged kite optimization algorithm (BKA) with the characteristics of strong portability, strong optimization ability and fast convergence speed is selected to jointly optimize the regularization parameters and kernel function parameters of the kernel extreme learning machine (KELM), and establish a BKA-KELM classifier, which solves the problem that the KELM hyperparameters (kernel function parameters, regularization parameters) are difficult to adjust. Finally, the features extracted by TSMFDE are input into the classifier to complete the pattern recognition work.

[0135] BKA algorithm explanation and related formulas:

[0136] X i =BK lb +rand(BK ub -BK lb )

[0137] Among them, i is an integer between 1 and pop, X i represents the position of the i-th individual during the iteration of the optimization algorithm; BKlb and BKub are the lower and upper bounds of the j-th dimension black-winged kite, respectively, and rand is a randomly selected value between [0, 1].

[0138] During initialization, BKA selects the individual with the best fitness value as the leaderXL in the initial group, which is considered the optimal position for the black-winged kite. Below is the mathematical representation of the initial leader XL, taking the minimum value as an example.

[0139] f best =min(f(X i ))

[0140] X L =X(find(f best = = f(X i )))

[0141] Among them, f best represents the minimum fitness in the group; X L represents the leader in the initial group, that is, the individual with the best fitness value; f(X i ) represents the fitness value of the ith individual; X() represents the position of the individual that meets the condition of the function;

[0142] Aggressive behavior:

[0143]

[0144] and They represent the position of the i-th black-winged kite in the j-th dimension and the (t+1)-th iteration step respectively; r is a random number between 0 and 1, p is a calculated constant; T is the total number of iterations, and t is the number of iterations completed so far.

[0145] Migratory behavior:

[0146]

[0147] in, denotes the leading scorer of the j-th dimension Black Wing Kite at the t-th iteration so far. and represents the position of the i-th black-winged kite in the j-th dimension and the (t+1) iteration step respectively; F i represents the j-th dimension current position obtained by any black-winged kite in the t-th iteration; F ri represents the fitness value of any black-winged kite at a random position in the jth dimension in the tth iteration;

[0148] After the identification is completed, the embodiment of the present invention intends to use the knowledge graph to verify the identification data of the vortex. The knowledge graph is a new technology that uses a graph model to describe knowledge and model the relationship between all things, and has the capabilities of multi-level relationship query, knowledge analysis and reasoning. The knowledge graph is a graph database composed of a large number of triples {(head entity, relationship, tail entity) or (entity, attribute, attribute value)}, and its key elements include entities, entity relationships and entity attributes.

[0149] The step of verifying the vortex identification data using the knowledge graph specifically includes the following steps:

[0150] 1. Collect and organize the basic attributes of the vortex knowledge graph: its entities are the monitoring points on the sea surface, and the entity relationships are 1) the spatial distribution relationship of the monitoring points, that is, whether the observation points are spatially connected, and 2) the data sequence relationship of the monitoring points. This study uses the dynamic time warping (DTW) algorithm, which is good at processing complex data sequence relationships, to determine the data sequence similarity relationship of the sea surface monitoring points. Its entity attributes are object attributes (monitoring physical quantities of monitoring points, warning levels of single measuring points, and their locations) and data attributes (three-dimensional coordinates, current monitoring data, and historical monitoring data). This study uses the Neo4j database to construct the sea surface vortex monitoring knowledge graph, batch processes the fixed instructions of the graph database construction in Python programming, and batch imports a large number of triples in the text into the Neo4j database to complete the construction of the sea surface vortex monitoring knowledge graph.

[0151] 2. Use the established sea surface vortex knowledge graph to identify vortex anomalies:

[0152] 1) Query the attributes of all monitoring points in the sea surface vortex monitoring knowledge graph;

[0153] 2) Select any one of the vortex anomaly points as the reference point, and with the help of the multi-level relationship query capability of the knowledge graph, retrieve all the monitoring points that have a spatial adjacent relationship with the reference point, and judge whether these adjacent monitoring points are abnormal. The characteristics of the vortex anomaly point mainly include abnormal sea surface height, abnormal sea surface temperature and abnormal flow velocity change. This study mainly constructs an abnormal mesoscale vortex sample library, and then uses SSHA (sea surface height) and SSTA (sea surface temperature anomaly) data to extract and learn features through a densely connected network to obtain fusion feature data that integrates SSH (sea surface height) and SST (sea surface temperature) information. After that, the recognition is constructed through the improved U-Net image segmentation model to judge the abnormal situation of the vortex at the monitoring point. The construction of the U-Net model is to extract features from the monitoring data, such as vortex features (center point, radius), ocean parameters, etc., and associate the extracted features with the entities in the knowledge graph to enrich the model input (including remote sensing images and extracted features), and the output is the segmentation mask of the vortex generation area.

[0154] U-Net adopts an encoder-decoder structure, in which the encoder is used to extract features and the decoder is used to map features to segmentation results. With this structure, multi-scale information can be captured. Through jump connections, the feature map of the encoder layer can be directly transferred to the corresponding decoder layer, which can retain high-resolution information and help improve segmentation accuracy. Since this model uses multiple data such as SSH and SST for feature learning at the same time. Therefore, through the forward propagation of the network, the model can automatically extract features related to vortex anomalies. At the same time, in the improved U-Net process, an attention mechanism is added to help the model focus on more important feature areas, thereby improving the recognition accuracy of vortex anomalies. At the final output, the result of the model output is a segmentation map, which reflects the vortex generation area and the normal area.

[0155] Afterwards, the U-Net model will be trained using the labeled segmented data, the identification of vortex generation monitoring points will be supervised, and the cross entropy loss function and Adam optimizer will be used for model optimization.

[0156] The cross entropy loss function is used to calculate the difference between the true label distribution and the model predicted label distribution. For the binary classification problem, the cross entropy loss function is defined as follows:

[0157]

[0158] Where: N is the number of samples. i is the true label (0 or 1). is the probability predicted by the model (between 0 and 1).

[0159] For multi-classification problems, the cross entropy loss function is defined as:

[0160]

[0161] Where C is the number of categories; y i is the one-hot encoding of the true label; is the model's predicted probability for each category.

[0162] 3) According to the method of step 2), continue to search and determine whether there is an abnormal monitoring point outside the abnormal measuring point group that has a spatial adjacent relationship with any monitoring point in the abnormal measuring point group. If there is, then the new abnormal monitoring point is incorporated into the abnormal measuring point group, and the cycle is repeated until there is no abnormal monitoring point outside the abnormal measuring point group that has a spatial adjacent relationship with it;

[0163] 4) Repeat steps 2) and 3) until all abnormal monitoring points in the sea surface vortex monitoring knowledge graph are classified into different abnormal measurement point groups;

[0164] 5) Select any two abnormal measurement point groups in the knowledge graph, and use the relationship query capability of the graph database to determine whether there is a data sequence similarity relationship between the corresponding monitoring points in the two abnormal measurement point groups. If the number of monitoring points with a data sequence similarity relationship is greater than 50% of the total number of monitoring points in the two abnormal measurement point groups, then the two abnormal measurement point groups are considered to have a data sequence similarity relationship;

[0165] 6) Repeat step 5) to perform pairwise analysis on all abnormal measurement point groups in the knowledge graph, and query all abnormal measurement point groups with similar data sequence relationships.

[0166] 7) Compare the detected vortex monitoring points with the results obtained by pattern recognition to verify the vortex recognition results.

[0167] After the identification is completed, the present invention intends to use the ISSA-CNN-BiGRU regression prediction model to further predict the vortex.

[0168] The specific process is as follows:

[0169] 1. Convolutional neural network.

[0170] 1DCNN is used to extract the features of the original data, explore the implicit relationship between multiple features and filter out noise and unstable components. The information processed by CNN is then input into the Bi-GRU network for the next prediction.

[0171] 2. Bidirectional Gated Recurrent Unit (BiGRU), the bidirectional gated recurrent unit of the embodiment of the present invention is as follows Figure 4 shown.

[0172] 3. The output information at a certain time t is the sum of the outputs of the forward hidden layer and the backward hidden layer. The calculation formula is as follows:

[0173]

[0174] 4. ISSA optimized CNN-BiGRU model, reference Figure 5 , the steps specifically include the following 1)-7):

[0175] 1) Initialize population parameters, such as maximum number of iterations, population size, etc.

[0176] 2) Initialize CNN-BiGRU parameters and set the parameter optimization interval.

[0177] 3) Calculate the current fitness of each sparrow, find the optimal and worst fitness values, and the corresponding positions.

[0178] 4) After each iteration, the fitness of each sparrow is recalculated, and a dynamic learning factor is introduced to update the position of the discoverer; the Levy strategy is introduced to update the position of the follower; and the position of the sentinel is updated.

[0179] 5) According to the current status of the sparrows, update the optimal and worst fitness values ​​and their positions experienced by the entire sparrow population.

[0180] 6) Determine whether the maximum iteration is met. If so, assign the optimization result to CNN-BiGRU, otherwise return to 3) and continue iterating.

[0181] 7) Input the optimized parameters to build the CNN-BiGRU prediction model.

[0182] After completing the prediction, we will continue to follow up with the next vortex data to verify the accuracy of the prediction results.

[0183] 5. Introduce the normEve algorithm to optimize the obtained prediction model.

[0184] This algorithm combines the advantages of the Eve optimization algorithm and the AdaNorm optimization algorithm, and proposes a new optimization algorithm normEve. The convergence speed of the new algorithm is better than the traditional Adam optimization algorithm, and it also improves the problem of slow convergence speed of the Eve optimization algorithm when the sub-sample size is small. It has the characteristics of fast convergence speed and better convergence performance, and has better model optimization effect.

[0185] In summary, the present invention has the following characteristics:

[0186] (1) Generally speaking, the TSMFDE feature extraction method is based on the premise that there may be a lot of noise in the high-frequency ground wave radar signal of the ocean eddy. TSMFDE is more stable at any scale and can effectively distinguish different noises. It verifies that TSMFDE has good time series length robustness and recognition ability. TSMFDE has the advantages of good noise resistance.

[0187] (2) In order to solve the problem that KELM hyperparameters are difficult to adjust, the BKA algorithm with strong portability and not easy to fall into the optimal solution is introduced to optimize the kernel function parameters and regularization factor of KELM. Through experimental comparison, it can be seen that the BKA algorithm effectively improves the classification accuracy of the model. This is an optimization of the recognition prediction method KELM.

[0188] (3) The TSMFDE–BKA–KELM model of this method is a model for identifying and predicting ocean eddies with good identification and prediction accuracy, that is, the TSMFDE–BKA–KELM method has the advantage of high precision.

[0189] (4) The BKA algorithm is used for optimization, which solves the problem that the regularization parameters and kernel function parameters of the kernel extreme learning machine (KELM) are difficult to adjust.

[0190] (5) This method uses the knowledge graph method to verify the recognition results. Knowledge graph is a new technology that uses a graph model to describe knowledge and model the relationship between all things. It has the capabilities of multi-level relationship query, knowledge analysis and reasoning.

[0191] (6) The vortex points are identified using the improved U-Net image model. In the improved U-Net process, an attention mechanism is added to help the model focus on more important feature areas, thereby improving the recognition accuracy of vortex anomalies.

[0192] (7) In the judgment of vortex anomalies at sea surface monitoring points based on knowledge graphs, densely connected networks (DenseNet) are used as a tool for feature extraction and learning. By connecting each layer to its subsequent layers, features can be reused, reducing information loss. This feature reuse helps the network learn richer feature representations, especially when processing ocean data, complex vortex structures can be better captured. In addition, since each layer is connected to all previous layers, the densely connected network improves the propagation of gradients, making the training process more stable. For complex ocean phenomena, this stability helps the network effectively learn the characteristics of vortices under big data.

[0193] (8) Densely linked networks can extract multi-scale features at different levels. This is particularly important for ocean monitoring, because the scale of vortices can vary, and the learning of multi-scale features enhances the model's adaptability to different types of vortices. In ocean monitoring, it is usually necessary to integrate multiple data (such as sea surface height anomalies and sea surface temperature anomalies). This connection method can naturally integrate these different data features together, helping the model learn more effective vortex feature judgments.

[0194] (9) A prediction model is established based on a neural network to perform data preprocessing, network structure design, and change prediction. ISSA is used to optimize the CNN-BiGRU model to make the prediction value more in line with the real-time situation.

[0195] (10) The cross-entropy loss function is used to optimize the U-Net model. The identification of vortices involves the classification of each pixel. The cross-entropy loss function can effectively measure the difference between the distribution of true label pixels and the model prediction results, and is very suitable for pixel-level classification problems. In addition, the cross-entropy loss function imposes a large penalty on misclassification, especially for low-probability predictions. This makes it easier to adjust the parameters of the model during training to reduce incorrect predictions of rare classes, thereby improving the overall performance of the model. The most important point is that the cross-entropy loss function is relatively simple and intuitive, easy to implement, and widely supported in many deep learning frameworks, which facilitates further model training and tuning.

[0196] Compared with the prior art, the present invention has the following advantages:

[0197] (1) High-frequency ground wave radar is used to collect images, which effectively improves the resolution and accuracy of images. High-frequency ground wave radar is relatively less affected by factors such as sea surface temperature and light visibility, which is beneficial for obtaining more accurate data.

[0198] (2) Data assimilation technology is used to combine the observational data with the numerical simulation results to further correct the ocean eddy data.

[0199] (3) Establish a prediction model based on neural network to perform data preprocessing, network structure design and change prediction.

[0200] (4) The results are finally output in the form of visual images, which is intuitive and clear.

[0201] (5) Check the results multiple times to ensure their accuracy.

[0202] (6) The normEve algorithm is used to optimize the prediction model to obtain more accurate data in a short time.

[0203] Another aspect of the embodiments of the present invention further provides an ocean vortex identification and prediction system, comprising:

[0204] The first module is used to obtain high-frequency ground wave radar signals of ocean eddies;

[0205] The second module is used to construct a training set and a test set according to the ocean eddy high-frequency ground wave radar signal;

[0206] The third module is used to train the training set by using a black kite optimization algorithm and a kernel extreme learning machine to obtain optimal kernel function parameters and regularization factors, and to construct an initial model;

[0207] The fourth module is used to extract features from the test set and test the initial model based on the extracted features to obtain a target model that meets the test requirements;

[0208] The fifth module is used to identify and predict the newly acquired ocean vortex high-frequency ground wave radar signal according to the target model to obtain the ocean vortex identification result.

[0209] It can be understood that the contents of the above method embodiments are all applicable to the present system embodiments, the functions specifically implemented by the present system embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0210] The embodiment of the present invention further provides an electronic device, the electronic device comprising a memory and a processor, the memory storing a computer program, and the processor implementing the above-mentioned ocean vortex identification and prediction method when executing the computer program. The electronic device can be any intelligent terminal including a tablet computer, a car computer, etc.

[0211] It can be understood that the contents of the above method embodiments are all applicable to the present device embodiments, the functions specifically implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0212] See also Figure 6 , Figure 6 The hardware structure of an electronic device of another embodiment is illustrated, and the electronic device includes:

[0213] The processor 601 may be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention;

[0214] The memory 602 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 602 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 602, and the processor 601 calls and executes the method for identifying and predicting ocean vortices in the embodiment of the present invention;

[0215] Input / output interface 603, used to implement information input and output;

[0216] Communication interface 604, used to realize communication interaction between the device and other devices, which can be realized through wired mode (such as USB, network cable, etc.) or wireless mode (such as mobile network, WI FI, Bluetooth, etc.);

[0217] Bus 605 , which transmits information between various components of the device (e.g., processor 601 , memory 602 , input / output interface 603 , and communication interface 604 );

[0218] The processor 601 , the memory 602 , the input / output interface 603 and the communication interface 604 are connected to each other in communication within the device via a bus 605 .

[0219] An embodiment of the present invention further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the above-mentioned method for identifying and predicting ocean vortices is implemented.

[0220] It can be understood that the contents of the above method embodiments are all applicable to the present storage medium embodiments, the functions specifically implemented by the present storage medium embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0221] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely disposed relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0222] It should be noted that in various specific embodiments of the present invention, when it comes to the need to perform relevant processing based on data related to user identity or characteristics such as user information, user behavior data, user historical data, and user location information, the user's permission or consent will be obtained first, and the collection, use, and processing of these data will comply with relevant laws, regulations, and standards. In addition, when the embodiment of the present invention needs to obtain the user's sensitive personal information, it will obtain the user's separate permission or consent through a pop-up window or jump to a confirmation page, and after clearly obtaining the user's separate permission or consent, it will obtain the necessary user-related data for the normal operation of the embodiment of the present invention.

[0223] The embodiments described in the embodiments of the present invention are intended to more clearly illustrate the technical solutions of the embodiments of the present invention, and do not constitute a limitation on the technical solutions provided by the embodiments of the present invention. Those skilled in the art can appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of the present invention are also applicable to similar technical problems.

[0224] Those skilled in the art will appreciate that the technical solutions shown in the figures do not limit the embodiments of the present invention and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0225] The device embodiments described above are merely illustrative, and the units described as separate components may or may not be physically separated, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0226] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0227] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0228] It should be understood that in the present invention, "at least one (item)" refers to one or more, and "plurality" refers to two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can represent: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0229] In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0230] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0231] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0232] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store programs.

[0233] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but the scope of the rights of the embodiments of the present invention is not limited thereby. Any modification, equivalent substitution and improvement made by those skilled in the art without departing from the scope and essence of the embodiments of the present invention shall be within the scope of the rights of the embodiments of the present invention.

Claims

1. A method for identifying and predicting ocean vortices, characterized in that: The following steps are involved: Obtain high-frequency ground wave radar signals of ocean eddies; Constructing a training set and a test set according to the ocean eddy high-frequency ground wave radar signal; The training set is trained by using a black kite optimization algorithm and a kernel extreme learning machine to obtain optimal kernel function parameters and regularization factors, and an initial model is constructed; By extracting features from the test set and testing the initial model based on the extracted features, a target model that meets the test requirements is obtained; According to the target model, the newly acquired ocean vortex high-frequency ground wave radar signal is identified and predicted to obtain an ocean vortex identification result.

2. The method for identifying and predicting ocean vortices according to claim 1, characterized in that: The method further comprises: After completing the identification and prediction of ocean eddies, the knowledge graph is used to verify the eddy identification data.

3. The method for identifying and predicting ocean vortices according to claim 1, characterized in that: The expression of the black kite optimization algorithm is: i =BK lb +rand(BK ub -BK lb ), where X i represents the position of the i-th individual in the iterative process of the optimization algorithm; BK lb represents the lower bound of the j-th dimension black-winged kite; rand() represents a randomly selected value between [0,1]; BK ub Represents the upper bound of the j-th dimension black-winged kite; During the initialization process, the black kite optimization algorithm selects the individual with the best fitness value as the leader X in the initial group. L ; The leader X L The expression is: f best =min(f(X i )) X L =X(find(f best ==f(X i ))) Among them, f best represents the minimum fitness in the group; X L represents the leader in the initial group, which is the individual with the best fitness value; f(X i ) represents the fitness value of the ith individual; X() represents the position of the individual that meets the condition of the function; The attack behavior of the black kite optimization algorithm is expressed as: in, represents the position of the i-th black-winged kite in the j-th dimension and (t+1)-th iteration step; r represents a random number; p is a calculated constant; T is the total number of iterations; t is the number of iterations completed so far; n is a control parameter related to the number of iterations, and its value is determined by the following expression: The expression of the migration behavior of the black kite optimization algorithm is: in, represents the leading scorer of the j-th dimension black-wing kite at the t-th iteration so far; F i represents the j-th dimension current position obtained by any black-winged kite in the t-th iteration; F ri represents the fitness value of any black-winged kite at a random position in the jth dimension in the tth iteration; m is a control parameter whose value is determined by the following expression:

4. The method for identifying and predicting ocean vortices according to claim 2, characterized in that: After the identification and prediction of the ocean vortex is completed, the knowledge graph is used to verify the identification data of the vortex, including the following steps: Collect and organize the basic attributes of the vortex knowledge graph; the entities of the vortex knowledge graph are the monitoring points on the sea surface, the entity relationship is the spatial distribution relationship of the monitoring points, and the entity attributes are object attributes and data attributes. The object attributes include the monitoring physical quantity of the monitoring point, the warning level of a single measuring point and the location to which it belongs, and the data attributes include three-dimensional coordinates, current monitoring data and historical monitoring data; Select a vortex anomaly point as the reference point, retrieve all monitoring points that are spatially adjacent to the reference point based on the multi-level relationship query capability of the knowledge graph, and determine whether these adjacent monitoring points are abnormal; the characteristics of the vortex anomaly point include abnormal sea surface height, abnormal sea surface temperature, and abnormal flow velocity change; Continue to search and determine whether there is an abnormal monitoring point outside the abnormal measuring point group that has a spatially adjacent relationship with any monitoring point in the abnormal measuring point group. If so, the new abnormal monitoring point is incorporated into the abnormal measuring point group for processing until there is no abnormal monitoring point outside the abnormal measuring point group that has a spatially adjacent relationship with it; All abnormal monitoring points in the sea surface vortex monitoring knowledge graph are classified into different abnormal monitoring point groups; Select any two abnormal measurement point groups in the knowledge graph, and use the relationship query capability of the graph database to determine whether there is a data sequence similarity relationship between the corresponding monitoring points in the two abnormal measurement point groups. If the number of monitoring points with a data sequence similarity relationship is greater than 50% of the total number of monitoring points in the two abnormal measurement point groups, then the two abnormal measurement point groups are determined to have a data sequence similarity relationship; Perform pairwise analysis on all abnormal measurement point groups in the knowledge graph to obtain abnormal measurement point groups with similar data sequence relationships; The detected vortex monitoring points are compared with the results obtained by pattern recognition to verify the vortex identification results; after the identification is completed, a regression prediction model composed of an optimized sparrow algorithm, a convolutional neural network, and a bidirectional gated recursive unit is used to further predict the vortex.

5. The method for identifying and predicting ocean vortices according to claim 4, characterized in that: The method of selecting a vortex abnormal point as a reference point, retrieving all monitoring points that have a spatial adjacent relationship with the reference point according to the multi-level relationship query capability of the knowledge graph, and determining whether these adjacent monitoring points are abnormal, includes the following steps: By building an abnormal mesoscale eddy sample library, and then using the abnormal sea surface height and sea surface temperature data, feature extraction and learning are performed through a densely connected network to obtain fused feature data that integrates sea surface height and sea surface temperature information; The improved U-Net image segmentation model is used to build recognition and judge the abnormal situation of the vortex at the monitoring point. The construction process of the U-Net model is as follows: extract features from the monitoring data, including the vortex center point, vortex radius, and ocean parameters; associate the extracted features with the entities in the knowledge graph and output them as the segmentation mask of the vortex generation area; U-Net adopts an encoder-decoder structure, in which the encoder is used to extract features and the decoder is used to map features to segmentation results; add an attention mechanism to U-Net to focus on key areas; Use the labeled segmentation data to train the U-Net model, supervise the learning of vortex generation monitoring points, and use the cross entropy loss function and Adam optimizer to optimize the model; Among them, the cross entropy loss function The expression is: Where N represents the number of samples; y i represents the true label; Represents the probability predicted by the model.

6. The method for identifying and predicting ocean vortices according to claim 4, characterized in that: The detected vortex monitoring points are compared with the results obtained by pattern recognition to verify the vortex recognition results; after the recognition is completed, the vortex is further predicted using a regression prediction model, including the following steps: 1D CNN is used to extract the features of the original data, explore the implicit relationship between multiple features and filter out noise and unstable components. The information processed by CNN is then input into the bidirectional gated recurrent unit Bi-GRU for the next step of prediction. In the prediction process of the bidirectional gated recurrent unit Bi-GRU, the output information at a certain time t is the sum of the outputs of the forward hidden layer and the backward hidden layer; The CNN–BiGRU model is optimized using the optimized sparrow algorithm; The normEve algorithm is introduced to optimize the obtained prediction model and complete further prediction of the vortex.

7. The method for identifying and predicting ocean vortices according to claim 6, characterized in that: The optimization of the CNN-BiGRU model by using the optimized sparrow algorithm includes the following steps: Initializing population parameters, wherein the population parameters include a maximum number of iterations and a population size; Initialize CNN-BiGRU parameters and set the parameter optimization interval; Calculate the current fitness of each sparrow, find the best and worst fitness values, and the corresponding positions; After each iteration, the fitness of each sparrow is recalculated, and a dynamic learning factor is introduced to update the position of the discoverer; the Levy strategy is introduced to update the position of the follower; and the position of the vigilant is updated; According to the current status of the sparrows, update the optimal and worst fitness values ​​and their positions experienced by the entire sparrow population; Determine whether the maximum iteration is met. If so, assign the optimization result to CNN-BiGRU. Otherwise, continue iterating until the maximum iteration is met. Input the optimized parameters to build the CNN-BiGRU prediction model.

8. A system for identifying and predicting ocean vortices, characterized in that: include: The first module is used to obtain high-frequency ground wave radar signals of ocean eddies; The second module is used to construct a training set and a test set according to the ocean eddy high-frequency ground wave radar signal; The third module is used to train the training set by using a black kite optimization algorithm and a kernel extreme learning machine to obtain optimal kernel function parameters and regularization factors, and to construct an initial model; The fourth module is used to extract features from the test set and test the initial model based on the extracted features to obtain a target model that meets the test requirements; The fifth module is used to identify and predict the newly acquired ocean vortex high-frequency ground wave radar signal according to the target model to obtain the ocean vortex identification result.

9. An electronic device, characterized in that: including a processor and a memory; The memory is used to store programs; The processor executes the program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the method according to any one of claims 1 to 7.

Images