A method and system for regulating marine ranch resources

Abstract

The present invention discloses a method and system for regulating marine ranch resources. By uniformly scoring and mapping multi-dimensional time series data such as historical catches, species habits and environmental information to a three-dimensional geometric structure, the present invention effectively eliminates the subjective bias of traditional empirical judgment and realizes the quantification and transparency of decision-making; constructs sphere screening and ecological topology maps with standard scores as anchor points, improves the spatiotemporal resolution of marine ranch habitat connectivity and key ecological patches, and provides precise targeting for directed reproduction and release and habitat restoration; further captures the nonlinear interaction between environmental factors and biological communities through a graph convolutional network, and realizes real-time dynamic adjustment of release density and environmental control measures, thereby improving the response speed to sudden environmental changes and reducing the risk of mass mortality, thereby achieving optimal resource utilization, enhanced ecological resilience and increased sustainable catch.

Description

Technical Field

[0001] The present invention relates to the technical field of resource regulation, and in particular to a method and system for regulating marine ranch resources. Background Art

[0002] Population growth and consumption upgrades are driving a surge in demand for aquatic products. Marine ranching, as a "blue granary," has become a crucial tool for addressing food security and ecological restoration. Ecological integrity is a core pillar of sustainable marine ranching, impacting the entire chain from resource production to environmental quality and economic benefits. Rational resource management can accelerate marine ranching development and enhance its economic potential.

[0003] Most existing resource ranch regulation methods regulate marine ranches through resource assessments, ignoring the ecological integrity of marine ranches. The lack of ecological integrity may lead to "short-sighted economics" and insufficient connectivity in marine ranch ecosystems. For example, although high-density cage aquaculture increases production in the short term, it will cause water quality deterioration in the long term, increase disease risks and treatment costs, and have poor disaster resistance. Summary of the Invention

[0004] The present invention provides a method and system for regulating marine ranch resources, so as to rationally regulate marine ranch resources.

[0005] In order to solve the above technical problems, the present invention provides a method for regulating marine ranch resources, comprising:

[0006] Acquiring multidimensional time series data of the marine ranch, performing score evaluation on the multidimensional time series data to obtain a fishing score, a habit score, and an environmental score at each moment; and constructing an endpoint set based on the fishing score, habit score, and environmental score at each moment;

[0007] Taking a preset standard score as a base point, connecting the base point and all endpoints in the endpoint set to construct a three-dimensional geometric structure; and taking the base point as the sphere center and constructing a spherical structure based on a preset radius, screening the endpoints within the spherical structure as a candidate set;

[0008] Connect any two endpoints in the candidate set to construct an initial ecological topology map; and modify the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map;

[0009] Constructing an ecological assessment model based on a graph convolution model, and evaluating the ecological integrity of the marine ranch based on the target ecological topology map and the ecological assessment model;

[0010] Resource regulation parameters are generated based on the fishing gain score, habit score, environment score and ecological integrity, so as to perform resource regulation based on the resource regulation parameters.

[0011] The present invention effectively eliminates the subjective bias of traditional empirical judgment and realizes the quantification and transparency of decision-making by uniformly scoring and mapping multi-dimensional time series data such as historical catches, species habits and environmental information into a three-dimensional geometric structure; on this basis, sphere screening and ecological topology maps are constructed with standard scores as anchor points, which improves the spatiotemporal resolution of marine ranch habitat connectivity and key ecological patches, and provides precise targeting for directional reproduction and release and habitat restoration; further, the nonlinear interaction between environmental factors and biological communities is captured through the graph convolutional network, making the predictive performance of ecological integrity assessment significantly better than the traditional linear model; the resource control parameters generated based on this assessment score can be issued in a closed loop through the Internet of Things monitoring platform, realizing real-time dynamic adjustment of release density and environmental control measures, thereby improving the response speed to sudden environmental changes and reducing the risk of mass mortality, thereby achieving optimal resource utilization, enhanced ecological resilience and increased sustainable catch.

[0012] Furthermore, the multidimensional time series data includes historical fishing data, species habit data, and environmental information data; the multidimensional time series data of the marine ranch is obtained, and the multidimensional time series data is scored and evaluated to obtain the fishing score, habit score, and environmental score at each moment; and an endpoint set is constructed based on the fishing score, habit score, and environmental score at each moment, including:

[0013] Obtaining historical catch data, species habit data, and environmental information data of the marine ranch; resampling the historical catch data, species habit data, and environmental information data based on a time series model and a preset standard scale to obtain standard catch data, standard habit data, and standard environmental data;

[0014] Scoring and evaluating the standard catch data, standard habit data, and standard environment data based on a fuzzy comprehensive evaluation method to obtain the catch score, habit score, and environment score at each moment;

[0015] A score vector corresponding to each moment is constructed based on the fishing score, habit score and environment at each moment, and an endpoint set is constructed based on the score vector.

[0016] The present invention achieves standardization and comprehensive quantification of time series data by resampling and fuzzy comprehensive evaluation scores of historical marine ranch catch data, species habit data, and environmental information data. On the one hand, the resampling process ensures the consistency of various types of heterogeneous data on a temporal scale, effectively eliminating information loss and bias caused by different sampling frequencies; on the other hand, the fuzzy comprehensive evaluation method can take into account the mutual influence between multiple indicators and uniformly measure the catch score, habit score, and environmental score, thereby obtaining a score vector that can truly reflect the ecological status of the ranch at different times. By constructing an endpoint set, this method lays a solid data foundation for the subsequent construction and dynamic evaluation of the ecological topology network, significantly improving the accuracy and operability of marine ranch resource regulation.

[0017] Furthermore, connecting any two endpoints in the candidate set to construct an initial ecological topology map; and modifying the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map, including:

[0018] Connect any two endpoints in the candidate set to construct an initial ecological topology graph;

[0019] Calculate the endpoint distance between any two endpoints in the initial ecological topology graph;

[0020] The connection line between two endpoints whose endpoint distance is greater than the preset distance threshold in the initial ecological topology map is eliminated to obtain a target ecological topology map.

[0021] This invention uses a graph topology correction mechanism based on a preset distance threshold, which effectively filters out node connections with weak geographical or ecological connections, avoiding the problem of network bloat caused by excessive connections. At the same time, through the adjustability of the distance threshold, it can flexibly adapt to the scale of different sea areas and species distribution characteristics, ensuring that the target ecological topology map retains key ecological connections while reducing noise interference. This approach not only improves the accuracy of the spatiotemporal representation of ecological networks, but also provides clearer structural input for subsequent feature learning of graph convolution models on complex networks, enhancing the model's generalization performance and computational efficiency.

[0022] Furthermore, the ecological assessment model is constructed based on the graph convolution model, and the ecological integrity of the marine ranch is assessed based on the target ecological topology map and the ecological assessment model, including:

[0023] Constructing an ecological assessment model based on a graph convolution model, wherein the ecological assessment model includes a first graph convolution module, a second graph convolution module, and a spatial attention module;

[0024] Perform node attribute encoding and edge attribute encoding based on the target ecological topology graph to obtain a node feature matrix, an adjacency matrix, and an edge tensor;

[0025] Performing feature extraction on the node feature matrix and the adjacency matrix based on the first graph convolution module to obtain local ecological features;

[0026] Aggregating the local ecological features and the adjacency matrix based on the second graph convolution module to obtain regional ecological features;

[0027] fusing the regional ecological features based on the spatial attention module and preset key nodes to obtain enhanced features;

[0028] A heterogeneous graph feature map is generated based on the enhanced features and the edge tensor, and node prediction is performed on the heterogeneous graph feature map to obtain an ecological integrity score.

[0029] The present invention constructs an ecological assessment model that integrates a two-layer graph convolution module and a spatial attention module to form a multi-scale, multi-perspective ecological feature extraction and fusion architecture. The first graph convolution module focuses on local node neighborhood information and carefully depicts micro-ecological interactions; the second graph convolution module performs cross-regional feature aggregation to capture macro-ecological distribution patterns; the spatial attention module performs weighted fusion of regional ecological features based on preset key nodes, thereby enhancing the expression of important ecological information. This composite structure can generate highly expressive feature maps on heterogeneous graphs, and obtain accurate ecological integrity scores through node prediction, effectively improving the accuracy and reliability of marine ranch ecological health assessments.

[0030] Furthermore, generating a heterogeneous graph feature map based on the enhanced features, and performing node prediction on the heterogeneous graph feature map to obtain an ecological integrity score includes:

[0031] Generating a heterogeneous graph feature map based on the enhanced features, performing node prediction on each node in the heterogeneous graph feature map, and obtaining an integrity score of each node;

[0032] The integrity scores of the nodes are weighted and summed to obtain an ecological integrity score.

[0033] This method predicts an independent integrity score for each node, providing a refined reflection of its contribution and potential risks within the ecosystem. It then uses preset or dynamically adjusted weights to weight the scores across nodes, effectively aggregating the overall ecological integrity score. This approach balances local node diversity with the global nature of the network, providing not only detailed node-level health diagnostics but also a clear global integrity indicator, providing a scientific basis for precise resource regulation and management decisions.

[0034] Furthermore, generating resource control parameters based on the fishing score, habit score, environment score and ecological integrity, and performing resource control based on the resource control parameters, includes:

[0035] Determine regional control objectives and control directions based on the fishing yield score, habit score, environmental score and ecological integrity;

[0036] Based on the control target and the control direction, resource control parameters are matched in a preset parameter rule library, and resource control is performed based on the resource control parameters.

[0037] This method determines the control objectives and directions for each region based on the integrity score and matches the optimal control parameters within a pre-set parameter rule library. Translating complex ecological assessment results into actionable control plans not only improves the scientific nature and accuracy of control decisions, but also enhances the automation and intelligence of resource regulation, contributing to the sustainable utilization of marine ranching resources and the maintenance of ecological balance.

[0038] In a second aspect, the present invention provides a marine ranch resource control system, comprising: a data acquisition module, a screening module, a topology generation module, an evaluation module, and a resource control module;

[0039] The data acquisition module is used to acquire multi-dimensional time series data of the marine ranch, perform score evaluation on the multi-dimensional time series data, obtain the fishing score, habit score and environment score at each moment; and construct an endpoint set based on the fishing score, habit score and environment score at each moment;

[0040] The screening module is configured to use a preset standard score as a base point, connect the base point and all endpoints in the endpoint set to construct a three-dimensional geometric structure; and use the base point as the sphere center and a preset radius to construct a spherical structure, and screen the endpoints within the spherical structure as a candidate set;

[0041] The topology generation module is configured to connect any two endpoints in the candidate set to construct an initial ecological topology map; and to modify the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map;

[0042] The evaluation module is used to construct an ecological evaluation model based on a graph convolution model, and evaluate the ecological integrity of the marine ranch based on the target ecological topology map and the ecological evaluation model;

[0043] The resource control module is used to generate resource control parameters based on the fishing score, habit score, environment score and ecological integrity, so as to perform resource control based on the resource control parameters.

[0044] Furthermore, the multidimensional time series data includes historical fishing data, species habit data, and environmental information data; the data acquisition module is used to acquire the multidimensional time series data of the marine ranch, perform score evaluation on the multidimensional time series data, and obtain the fishing score, habit score, and environmental score at each moment; and construct an endpoint set based on the fishing score, habit score, and environmental score at each moment, including:

[0045] Obtaining historical catch data, species habit data, and environmental information data of the marine ranch; resampling the historical catch data, species habit data, and environmental information data based on a time series model and a preset standard scale to obtain standard catch data, standard habit data, and standard environmental data;

[0046] Scoring and evaluating the standard catch data, standard habit data, and standard environment data based on a fuzzy comprehensive evaluation method to obtain the catch score, habit score, and environment score at each moment;

[0047] A score vector corresponding to each moment is constructed based on the fishing score, habit score and environment at each moment, and an endpoint set is constructed based on the score vector.

[0048] Furthermore, the screening module is configured to connect any two endpoints in the candidate set to construct an initial ecological topology map; and to modify the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map, including:

[0049] Connect any two endpoints in the candidate set to construct an initial ecological topology graph;

[0050] Calculate the endpoint distance between any two endpoints in the initial ecological topology graph;

[0051] The connection line between two endpoints whose endpoint distance is greater than the preset distance threshold in the initial ecological topology map is eliminated to obtain a target ecological topology map.

[0052] Furthermore, the evaluation module is used to construct an ecological evaluation model based on a graph convolution model, and evaluate the ecological integrity of the marine ranch based on the target ecological topology map and the ecological evaluation model, including:

[0053] Constructing an ecological assessment model based on a graph convolution model, wherein the ecological assessment model includes a first graph convolution module, a second graph convolution module, and a spatial attention module;

[0054] Based on the target ecological topology graph, obtaining a node feature matrix and an adjacency matrix;

[0055] Performing feature extraction on the node feature matrix and the adjacency matrix based on the first graph convolution module to obtain local ecological features;

[0056] Aggregating the local ecological features and the adjacency matrix based on the second graph convolution module to obtain regional ecological features;

[0057] fusing the regional ecological features based on the spatial attention module and preset key nodes to obtain enhanced features;

[0058] A heterogeneous graph feature map is generated based on the enhanced features, and node prediction is performed on the heterogeneous graph feature map to obtain an ecological integrity score. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 A schematic flow chart of a method for regulating marine ranch resources provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The following embodiments of the present invention are described in further detail with reference to the accompanying drawings and examples. The following examples are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0061] The terms "first," "second," and the like in the specification, claims, and drawings of this application are used to distinguish between different objects, not to describe a particular order. Furthermore, the terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements, but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0062] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0063] Example 1

[0064] See also Figure 1 , Figure 1 A schematic flow chart of a method for regulating marine ranch resources provided by an embodiment of the present invention. The method includes steps 101 to 105, as follows:

[0065] Step 101: Acquire multidimensional time series data of the marine ranch, perform score evaluation on the multidimensional time series data, obtain fishing scores, habit scores, and environmental scores at each moment; and construct an endpoint set based on the fishing scores, habit scores, and environmental scores at each moment.

[0066] In this embodiment, the multidimensional time series data includes historical fishing data, species habit data, and environmental information data; the multidimensional time series data of the marine ranch is obtained, and the multidimensional time series data is scored and evaluated to obtain the fishing score, habit score, and environmental score at each moment; and an endpoint set is constructed based on the fishing score, habit score, and environmental score at each moment, including:

[0067] Obtaining historical catch data, species habit data, and environmental information data of the marine ranch; resampling the historical catch data, species habit data, and environmental information data based on a time series model and a preset standard scale to obtain standard catch data, standard habit data, and standard environmental data;

[0068] Scoring and evaluating the standard catch data, standard habit data, and standard environment data based on a fuzzy comprehensive evaluation method to obtain the catch score, habit score, and environment score at each moment;

[0069] A score vector corresponding to each moment is constructed based on the fishing score, habit score and environment at each moment, and an endpoint set is constructed based on the score vector.

[0070] In this embodiment, the marine ranch's catch time series data (including catch volume, operation time, etc.), species electronic tag migration trajectory and acoustic monitoring records, as well as environmental information data (sea temperature, salinity, dissolved oxygen, etc.) are obtained.

[0071] In this embodiment, the three types of original time series data are resampled on a daily or weekly basis: the time series resampling method provided by platforms such as Dataiku is used to perform linear or Lagrangian interpolation on missing dates to ensure uniform time intervals.

[0072] In this embodiment, the resampled standard catch data, standard habit data, and standard environment data are respectively subjected to Z-score normalization to map the values ​​of each dimension to the same magnitude, thereby eliminating the influence of different indicator dimensions.

[0073] In this embodiment, an evaluation index set is constructed: standard fish catch, standard habit data, and standard environmental data (such as dissolved oxygen, pH, and suspended solids concentration) are incorporated into the fuzzy evaluation system. For each evaluation index, a semi-sine or triangular membership function is selected, and the normalized value is mapped to the membership interval [0, 1] to describe the "poor-medium-excellent" grade division. The entropy weight method (Entropy Weight Method) or the analytic hierarchy process (AHP) is used to determine the weight of each index and construct the weight matrix W.

[0074] In this embodiment, the index membership vector at each moment Add the weight vector W to get the fishing score. , Habit score and environmental score , the ternary score at each moment ( , , ) as endpoints in three-dimensional space , thereby generating an endpoint set; preset standard scoring points =( , , ) as a base point.

[0075] In this example, the standardization and comprehensive quantification of time series data were achieved by resampling and fuzzy comprehensive evaluation of historical marine ranch catch data, species habit data, and environmental information data. On the one hand, the resampling process ensures the consistency of various heterogeneous data across time scales, effectively eliminating information loss and bias caused by different sampling frequencies. On the other hand, the fuzzy comprehensive evaluation method can take into account the mutual influence between multiple indicators and uniformly measure the catch score, habit score, and environmental score, thereby obtaining a score vector that can truly reflect the ecological status of the ranch at different times. By constructing an endpoint set, this method lays a solid data foundation for the subsequent construction and dynamic evaluation of the ecological topology network, significantly improving the accuracy and operability of marine ranch resource regulation.

[0076] Step 102: Using a preset standard score as a base point, connecting the base point and all endpoints in the endpoint set to construct a three-dimensional geometric structure; using the base point as the sphere center and a preset radius to construct a spherical structure, and selecting endpoints within the spherical structure as a candidate set;

[0077] In this embodiment, a three-dimensional vector is defined =( , , ), representing the standard values ​​of fishing gain score, habit score, and environment score, respectively, as reference points for the ideal ecological state.

[0078] In this embodiment, all endpoints With base point Connect the lines to generate a set of radial three-dimensional line segments, which can be regarded as a set of three-dimensional vector clusters with the standard ecological state as the origin, to construct a three-dimensional geometric structure:

[0079] (1)

[0080] In this embodiment, the base point is used as the center of the sphere and the radius r is set, wherein the preset radius is an adjustable parameter used to remove abnormal point sets according to the preset radius.

[0081] In this embodiment, all The point break store is used as the candidate point set .

[0082] In this embodiment, the multidimensional indicator scores are mapped into a three-dimensional spatial structure, effectively integrating the three types of indicators of catch, habits and environment, which is convenient for visualization and subsequent geometric operations; the deviation range is limited by the spatial spherical area to achieve the clustering and extraction of "ecologically similar states".

[0083] In this example, candidate points represent historical state samples that are "close to the standard ecological state." This helps model a stable subset of ecological states and enhances the robustness of subsequent graph modeling and graph convolution. Furthermore, the preset radius r can be adjusted based on the policy, corresponding to different tolerance requirements, supporting the selection of ecological management strategies from loose to strict.

[0084] Step 103: Connect any two endpoints in the candidate set to construct an initial ecological topology map; and modify the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map;

[0085] In this embodiment, connecting any two endpoints in the candidate set to construct an initial ecological topology map; and modifying the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map, including:

[0086] Connect any two endpoints in the candidate set to construct an initial ecological topology graph;

[0087] Calculate the endpoint distance between any two endpoints in the initial ecological topology graph;

[0088] The connection line between two endpoints whose endpoint distance is greater than the preset distance threshold in the initial ecological topology map is eliminated to obtain a target ecological topology map.

[0089] In this embodiment, the candidate set ,in .

[0090] In this embodiment, all unordered pairs (Pi, Pj) are traversed, where 1≤i <j≤N。

[0091] In this embodiment, in the initial diagram Add an undirected edge for each pair of endpoints , thus obtaining the initial ecological topology .

[0092] In this embodiment, for each edge (Pi, Pj) of the initial ecological topology graph, the Euclidean distance of the endpoints is calculated:

[0093] (2)

[0094] in, , , , , , Endpoints and Projected distance on the stereographic coordinate plane.

[0095] In this embodiment, by presetting the distance ,when When , the endpoint is retained and Otherwise, remove the endpoints and The edges between them are used to generate the target ecological topology graph .

[0096] In this embodiment, by eliminating edges whose distance is greater than a preset distance threshold, point pairs with excessively large score differences are removed, thereby avoiding mistakenly regarding samples with very different ecological states as connected.

[0097] In this embodiment, the target topology diagram depicts the real ecological connectivity path and provides an accurate network structure foundation for resource regulation.

[0098] In this embodiment, a graph topology correction mechanism based on a preset distance threshold effectively filters out node connections with weak geographical or ecological connections, avoiding the problem of network bloat caused by overconnection. Furthermore, the adjustable distance threshold allows for flexible adaptation to different sea area sizes and species distribution characteristics, ensuring that the target ecological topology map retains key ecological connections while reducing noise interference. This approach not only improves the accuracy of the spatiotemporal representation of ecological networks but also provides clearer structural input for subsequent graph convolutional models to learn features on complex networks, enhancing the model's generalization performance and computational efficiency.

[0099] Step 104: constructing an ecological assessment model based on the graph convolution model, and evaluating the ecological integrity of the marine ranch based on the target ecological topology map and the ecological assessment model;

[0100] In this embodiment, the ecological assessment model is constructed based on the graph convolution model, and the ecological integrity of the marine ranch is assessed based on the target ecological topology map and the ecological assessment model, including:

[0101] Constructing an ecological assessment model based on a graph convolution model, wherein the ecological assessment model includes a first graph convolution module, a second graph convolution module, and a spatial attention module;

[0102] Perform node attribute encoding and edge attribute encoding based on the target ecological topology graph to obtain a node feature matrix, an adjacency matrix, and an edge tensor;

[0103] Performing feature extraction on the node feature matrix and the adjacency matrix based on the first graph convolution module to obtain local ecological features;

[0104] Aggregating the local ecological features and the adjacency matrix based on the second graph convolution module to obtain regional ecological features;

[0105] fusing the regional ecological features based on the spatial attention module and preset key nodes to obtain enhanced features;

[0106] A heterogeneous graph feature map is generated based on the enhanced features and the edge tensor, and node prediction is performed on the heterogeneous graph feature map to obtain an ecological integrity score.

[0107] In this embodiment, the target ecological topology map ,in is a list of nodes, is the edge list.

[0108] In this embodiment, all node numbers and corresponding space or time identifiers are directly extracted from the original node set to construct the adjacency matrix. ,like .

[0109] In this embodiment, the continuous three-dimensional scores in the node list are directly normalized (eg, Z-score) and then mapped to a 64-dimensional space using a linear layer to obtain continuous features.

[0110] In this embodiment, the species category of the marine ranch can be mapped to a 64-dimensional vector through a learnable embedding layer to obtain classification features, and the water quality level can be mapped to a 64-dimensional vector through segmented quantization or embedding or position encoding to obtain ordinal features. Finally, the continuous features, classification features, and ordinal features are concatenated or added dimension by dimension to obtain the final 64-dimensional features of each node, thereby generating a node feature matrix. .

[0111] In this embodiment, generating a heterogeneous graph feature map based on the enhanced features, and performing node prediction on the heterogeneous graph feature map to obtain an ecological integrity score includes:

[0112] Generating a heterogeneous graph feature map based on the enhanced features, performing node prediction on each node in the heterogeneous graph feature map, and obtaining an integrity score of each node;

[0113] The integrity scores of the nodes are weighted and summed to obtain an ecological integrity score.

[0114] In this embodiment, the ecological assessment model consists of three major modules:

[0115] The first graph convolution module aggregates the features of the first-order neighbors of each node (time) to extract the local ecological situation. The second graph convolution module performs second-order neighbor aggregation based on the output of the first layer to capture regional ecological connections on a larger scale. The spatial attention module dynamically adjusts the neighbor weights of preset key nodes (such as peak spawning times or artificial reef deployment times) to enhance their influence in the assessment.

[0116] In this embodiment, the first graph convolution module constructs an adjacency matrix with self-connection , calculate the degree matrix , thus obtaining local ecological characteristics based on the degree matrix and adjacency matrix:

[0117] (3)

[0118] in, is a trainable parameter.

[0119] In this embodiment, the local ecological features and the adjacency matrix are aggregated by the second graph convolution module to obtain regional ecological features:

[0120] (4)

[0121] in, is a trainable parameter.

[0122] In this embodiment, the attention energy is calculated for each pair of neighbors (i, j) based on the spatial attention module:

[0123] (5)

[0124] in, Represents the attention energy of node i to node j, which is used to measure the initial weight when aggregating information from node i to node j. LeakyReLU(⋅) represents the linear rectifier unit activation function with leakage. is the leakage coefficient, represents the node i feature output by the second graph convolution module, The node j feature output by the second graph convolution module.

[0125] In this embodiment, key nodes include high-value spawning moments or artificial reef deployment moments. When a node in the target ecological topology is a key node, an amplification coefficient is applied to it, thereby amplifying its fusion weight and obtaining an enhanced feature:

[0126] (6)

[0127] In this embodiment, is the edge weight between node i and node j.

[0128] In this embodiment, Input a fully connected layer or a small MLP to predict the completeness score of each node:

[0129] (7)

[0130] in, is a bias term used to adjust the overall level of prediction, is a weight vector whose dimension is the same as the input feature dimension (i.e. dimensions are consistent).

[0131] In this embodiment, the constructed ecological assessment model integrates a two-layer graph convolution module and a spatial attention module to form a multi-scale, multi-perspective ecological feature extraction and fusion architecture. The first graph convolution module focuses on the local node neighborhood information and carefully depicts the micro-ecological interactions; the second graph convolution module performs cross-regional feature aggregation to capture the macro-ecological distribution pattern; the spatial attention module performs weighted fusion of regional ecological characteristics based on preset key nodes, thereby enhancing the expression of important ecological information. This composite structure can generate highly expressive feature maps on heterogeneous graphs, and obtain accurate ecological integrity scores through node prediction, effectively improving the accuracy and reliability of marine ranch ecological health assessments.

[0132] Step 105: Generate resource control parameters based on the fishing score, habit score, environment score and ecological integrity, and perform resource control based on the resource control parameters.

[0133] In this embodiment, generating resource control parameters based on the fishing score, habit score, environment score, and ecological integrity, and performing resource control based on the resource control parameters, includes:

[0134] Determine regional control objectives and control directions based on the fishing yield score, habit score, environmental score and ecological integrity;

[0135] Based on the control target and the control direction, resource control parameters are matched in a preset parameter rule library, and resource control is performed based on the resource control parameters.

[0136] In this embodiment, the integrity score is mapped to four ecological levels: excellent (≥0.8), good (0.6–0.8), medium (0.4–0.6), and poor (<0.4), and the control targets and directions are set respectively - the "excellent" level focuses on maintaining stability and fine-tuning, the "good" level focuses on moderate proliferation and fishing control, the "medium" level focuses on strengthening ecological restoration and environmental restoration, and the "poor" level focuses on fishing suspension and large-scale habitat restoration.

[0137] In this example, a pre-set resource control parameter rule base is used, encompassing resource control parameters such as species type, geographic location, release volume, fishing intensity, and aeration frequency. For example, if a sub-area is rated as a "medium" restoration target, the rule base will match the following: release density is set at 150% of the standard; 8 artificial reefs are deployed per square kilometer; aerators are activated twice daily; and pH is maintained within the 7.8–8.2 range. If a sub-area is rated "poor," measures such as a fishing moratorium, restoration of 10 mu of seagrass beds, and monthly sediment desilting are implemented.

[0138] In this example, an independent integrity score is predicted for each node to precisely reflect its contribution and potential risks within the ecosystem. Then, using preset or dynamically adjusted weights, these scores are weighted and summed, effectively aggregating the global ecological integrity score. This approach balances local node diversity with the overall network integrity, providing not only detailed node-level health diagnostics but also a clear global integrity indicator, providing a scientific basis for precise resource regulation and management decisions.

[0139] The embodiment of the present invention also provides a marine ranch resource control system, including: a data acquisition module, a screening module, a topology generation module, an evaluation module and a resource control module;

[0140] The data acquisition module is used to acquire multi-dimensional time series data of the marine ranch, perform score evaluation on the multi-dimensional time series data, obtain the fishing score, habit score and environment score at each moment; and construct an endpoint set based on the fishing score, habit score and environment score at each moment;

[0141] The screening module is configured to use a preset standard score as a base point, connect the base point and all endpoints in the endpoint set to construct a three-dimensional geometric structure; and use the base point as the sphere center and a preset radius to construct a spherical structure, and screen the endpoints within the spherical structure as a candidate set;

[0142] The topology generation module is configured to connect any two endpoints in the candidate set to construct an initial ecological topology map; and to modify the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map;

[0143] The evaluation module is used to construct an ecological evaluation model based on a graph convolution model, and evaluate the ecological integrity of the marine ranch based on the target ecological topology map and the ecological evaluation model;

[0144] The resource control module is used to generate resource control parameters based on the fishing score, habit score, environment score and ecological integrity, so as to perform resource control based on the resource control parameters.

[0145] In this embodiment, the multidimensional time series data includes historical fishing data, species habit data, and environmental information data; the data acquisition module is used to acquire the multidimensional time series data of the marine ranch, perform score evaluation on the multidimensional time series data, and obtain the fishing score, habit score, and environmental score at each moment; and construct an endpoint set based on the fishing score, habit score, and environmental score at each moment, including:

[0146] Obtaining historical catch data, species habit data, and environmental information data of the marine ranch; resampling the historical catch data, species habit data, and environmental information data based on a time series model and a preset standard scale to obtain standard catch data, standard habit data, and standard environmental data;

[0147] Scoring and evaluating the standard catch data, standard habit data, and standard environment data based on a fuzzy comprehensive evaluation method to obtain the catch score, habit score, and environment score at each moment;

[0148] A score vector corresponding to each moment is constructed based on the fishing score, habit score and environment at each moment, and an endpoint set is constructed based on the score vector.

[0149] In this embodiment, the screening module is configured to connect any two endpoints in the candidate set to construct an initial ecological topology map; and to modify the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map, including:

[0150] Connect any two endpoints in the candidate set to construct an initial ecological topology graph;

[0151] Calculate the endpoint distance between any two endpoints in the initial ecological topology graph;

[0152] The connection line between two endpoints whose endpoint distance is greater than the preset distance threshold in the initial ecological topology map is eliminated to obtain a target ecological topology map.

[0153] In this embodiment, the evaluation module is used to construct an ecological evaluation model based on a graph convolution model, and evaluate the ecological integrity of the marine ranch based on the target ecological topology map and the ecological evaluation model, including:

[0154] Constructing an ecological assessment model based on a graph convolution model, wherein the ecological assessment model includes a first graph convolution module, a second graph convolution module, and a spatial attention module;

[0155] Perform node attribute encoding and edge attribute encoding based on the target ecological topology graph to obtain a node feature matrix, an adjacency matrix, and an edge tensor;

[0156] Performing feature extraction on the node feature matrix and the adjacency matrix based on the first graph convolution module to obtain local ecological features;

[0157] Aggregating the local ecological features and the adjacency matrix based on the second graph convolution module to obtain regional ecological features;

[0158] fusing the regional ecological features based on the spatial attention module and preset key nodes to obtain enhanced features;

[0159] A heterogeneous graph feature map is generated based on the enhanced features and the edge tensor, and node prediction is performed on the heterogeneous graph feature map to obtain an ecological integrity score.

[0160] In this embodiment, the evaluation module is used to generate a heterogeneous graph feature map based on the enhanced features, and perform node prediction on the heterogeneous graph feature map to obtain an ecological integrity score, including:

[0161] Generating a heterogeneous graph feature map based on the enhanced features, performing node prediction on each node in the heterogeneous graph feature map, and obtaining an integrity score of each node;

[0162] The integrity scores of the nodes are weighted and summed to obtain an ecological integrity score.

[0163] In this embodiment, the resource control module is used to generate resource control parameters based on the fishing score, habit score, environment score and ecological integrity, and perform resource control based on the resource control parameters, including:

[0164] Determine regional control objectives and control directions based on the fishing yield score, habit score, environmental score and ecological integrity;

[0165] Based on the control target and the control direction, resource control parameters are matched in a preset parameter rule library, and resource control is performed based on the resource control parameters.

[0166] In an embodiment of the present invention, a terminal device is also provided, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the above-mentioned marine ranch resource regulation method is implemented.

[0167] In an embodiment of the present invention, a computer-readable storage medium is also provided, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned marine ranch resource regulation method.

[0168] For example, a computer program may be divided into one or more modules, one or more of which are stored in a memory and executed by a processor to implement the present invention. One or more modules may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program in a terminal device.

[0169] The terminal device may be a computing device such as a desktop computer, laptop, PDA, or cloud server. The terminal device may include, but is not limited to, a processor, memory, and display. Those skilled in the art will appreciate that the aforementioned components are merely examples of terminal devices and do not constitute a limitation of the terminal device. The terminal device may include more or fewer components, or a combination of certain components, or different components. For example, the terminal device may also include input / output devices, network access devices, buses, and the like.

[0170] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting various parts of the entire terminal device using various interfaces and lines.

[0171] The memory can be used to store computer programs and / or modules. The processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area. The program storage area can store an operating system and at least one application required for a function (such as a sound playback function, a text conversion function, etc.); the data storage area can store data generated based on the use of the mobile phone (such as audio data, text message data, etc.). In addition, the memory can include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0172] If the module for regulating marine ranching resources is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. Computer-readable media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, removable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunications signals, and software distribution media. Persons of ordinary skill in the art can understand and implement this without inventive effort.

[0173] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for regulating marine ranch resources, characterized in that: include: Acquiring multidimensional time series data of the marine ranch, performing score evaluation on the multidimensional time series data, and obtaining a fishing score, a habit score, and an environment score at each moment; Constructing an endpoint set based on the fishing score, habit score and environment score at each moment; Taking a preset standard score as a base point, connecting the base point and all endpoints in the endpoint set to construct a three-dimensional geometric structure; and taking the base point as the sphere center and constructing a spherical structure based on a preset radius, screening the endpoints within the spherical structure as a candidate set; Connect any two endpoints in the candidate set to construct an initial ecological topology graph; and modifying the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map; An ecological assessment model is constructed based on a graph convolution model, and the ecological integrity of the marine ranch is evaluated based on the target ecological topology map and the ecological assessment model, including: constructing an ecological assessment model based on a graph convolution model, the ecological assessment model including a first graph convolution module, a second graph convolution module and a spatial attention module; obtaining a node feature matrix and an adjacency matrix based on the target ecological topology map; performing feature extraction on the node feature matrix and the adjacency matrix based on the first graph convolution module to obtain local ecological features; aggregating the local ecological features and the adjacency matrix based on the second graph convolution module to obtain regional ecological features; fusing the regional ecological features based on the spatial attention module and preset key nodes to obtain enhanced features; generating a heterogeneous graph feature map based on the enhanced features, and performing node prediction on the heterogeneous graph feature map to obtain an ecological integrity score, including: generating a heterogeneous graph feature map based on the enhanced features, performing node prediction on each node in the heterogeneous graph feature map to obtain an integrity score of each node; performing weighted summation on the integrity scores of each node to obtain an ecological integrity score; Generate resource control parameters based on the fishing score, habit score, environmental score and ecological integrity, and perform resource control based on the resource control parameters, including: determining regional control targets and control directions based on the fishing score, habit score, environmental score and ecological integrity; match resource control parameters in a preset parameter rule library based on the control targets and control directions, and perform resource control based on the resource control parameters.

2. A method for regulating marine ranching resources according to claim 1, characterized in that: The multidimensional time series data includes historical fishing data, species habit data, and environmental information data; the multidimensional time series data of the marine ranch is obtained, and the multidimensional time series data is scored and evaluated to obtain the fishing score, habit score, and environmental score at each moment; An endpoint set is constructed based on the fishing score, habit score, and environment score at each moment, including: Obtaining historical catch data, species habit data, and environmental information data of the marine ranch; resampling the historical catch data, species habit data, and environmental information data based on a time series model and a preset standard scale to obtain standard catch data, standard habit data, and standard environmental data; Scoring and evaluating the standard catch data, standard habit data, and standard environment data based on a fuzzy comprehensive evaluation method to obtain the catch score, habit score, and environment score at each moment; A score vector corresponding to each moment is constructed based on the fishing score, habit score and environment score at each moment, and an endpoint set is constructed based on the score vector.

3. A method for regulating marine ranching resources according to claim 2, characterized in that: Connecting any two endpoints in the candidate set to construct an initial ecological topology graph; The initial ecological topology map is modified based on a preset distance threshold to obtain a target ecological topology map, including: Connect any two endpoints in the candidate set to construct an initial ecological topology graph; Calculate the endpoint distance between any two endpoints in the initial ecological topology graph; The connection line between two endpoints whose endpoint distance is greater than the preset distance threshold in the initial ecological topology map is eliminated to obtain a target ecological topology map.

4. A marine ranch resource control system, characterized in that: include: Data acquisition module, screening module, topology generation module, evaluation module and resource control module; The data acquisition module is used to acquire multi-dimensional time series data of the marine ranch, perform score evaluation on the multi-dimensional time series data, and obtain the fishing score, habit score, and environment score at each moment; Constructing an endpoint set based on the fishing score, habit score and environment score at each moment; The screening module is configured to use a preset standard score as a base point, connect the base point and all endpoints in the endpoint set to construct a three-dimensional geometric structure; and use the base point as the sphere center and a preset radius to construct a spherical structure, and screen the endpoints within the spherical structure as a candidate set; The topology generation module is used to connect any two endpoints in the candidate set to construct an initial ecological topology map; and modifying the initial ecological topology map based on a preset distance threshold to obtain a target ecological topology map; The evaluation module is used to construct an ecological evaluation model based on a graph convolution model, and evaluate the ecological integrity of the marine ranch based on the target ecological topology map and the ecological evaluation model, including: constructing an ecological evaluation model based on a graph convolution model, the ecological evaluation model including a first graph convolution module, a second graph convolution module and a spatial attention module; obtaining a node feature matrix and an adjacency matrix based on the target ecological topology map; performing feature extraction on the node feature matrix and the adjacency matrix based on the first graph convolution module to obtain local ecological features; aggregating the local ecological features and the adjacency matrix based on the second graph convolution module to obtain regional ecological features; fusing the regional ecological features based on the spatial attention module and preset key nodes to obtain enhanced features; generating a heterogeneous graph feature map based on the enhanced features, and performing node prediction on the heterogeneous graph feature map to obtain an ecological integrity score, including: generating a heterogeneous graph feature map based on the enhanced features, performing node prediction on each node in the heterogeneous graph feature map to obtain an integrity score of each node; performing weighted summation on the integrity scores of each node to obtain an ecological integrity score; The resource control module is used to generate resource control parameters based on the fishing score, habit score, environmental score and ecological integrity, so as to perform resource control based on the resource control parameters, including: determining regional control targets and control directions based on the fishing score, habit score, environmental score and ecological integrity; matching resource control parameters in a preset parameter rule library based on the control targets and control directions, and performing resource control based on the resource control parameters.

5. The marine ranch resource control system according to claim 4, characterized in that: The multidimensional time series data includes historical fishing data, species habit data, and environmental information data; the data acquisition module is used to acquire the multidimensional time series data of the marine ranch, perform score evaluation on the multidimensional time series data, and obtain the fishing score, habit score, and environmental score at each moment; An endpoint set is constructed based on the fishing score, habit score, and environment score at each moment, including: Obtaining historical catch data, species habit data, and environmental information data of the marine ranch; resampling the historical catch data, species habit data, and environmental information data based on a time series model and a preset standard scale to obtain standard catch data, standard habit data, and standard environmental data; Scoring and evaluating the standard catch data, standard habit data, and standard environment data based on a fuzzy comprehensive evaluation method to obtain the catch score, habit score, and environment score at each moment; A score vector corresponding to each moment is constructed based on the fishing score, habit score and environment at each moment, and an endpoint set is constructed based on the score vector.

6. The marine ranch resource control system according to claim 5, characterized in that: The screening module is used to connect any two endpoints in the candidate set to construct an initial ecological topology map; The initial ecological topology map is modified based on a preset distance threshold to obtain a target ecological topology map, including: Connect any two endpoints in the candidate set to construct an initial ecological topology graph; Calculate the endpoint distance between any two endpoints in the initial ecological topology graph; The connection line between two endpoints whose endpoint distance is greater than the preset distance threshold in the initial ecological topology map is eliminated to obtain a target ecological topology map.

Images