A method for evaluating and predicting deep-sea crust resources based on machine learning

Abstract

The present invention relates to the technical field of marine resource development, and specifically to a method for evaluating and predicting deep-sea crust resources based on machine learning, comprising the following steps: S1: collecting multi-band acoustic data, magnetic field data, and seawater chemical parameter data; S2: identifying ecologically sensitive areas; S3: extracting depth and density features related to crust resource distribution; S4: using a generative adversarial network to simulate changes in ecological impacts under different resource exploitation scenarios; S5: constructing a collaborative evaluation model; S6: dynamically updating the collaborative evaluation model; S7: generating a resource and ecological impact distribution map for the target area. The present invention combines the precise evaluation of deep-sea crust resources with the prediction of ecological environmental impacts by constructing a collaborative evaluation model, thereby achieving a scientific division of resource development priorities and a precise setting of ecological protection needs, and providing effective support for the sustainable development of deep-sea resources.

Description

Technical Field

[0001] The present invention relates to the technical field of marine resource development, and in particular to a method for evaluating and predicting deep-sea crust resources based on machine learning. Background Art

[0002] Deep-sea crust resources are rich in rare metal elements and have important economic and strategic value. They have gradually become a hot spot for deep-sea resource development around the world. Traditional deep-sea resource exploration methods mainly rely on acoustic detection, magnetic field measurement and geological sampling. These methods usually take a lot of time and cost, and in the deep-sea environment distributed over a large area, data collection and processing face extremely high difficulties. In addition, the deep-sea environment is complex and changeable, and the density and composition of crust resources at different depths and regions vary significantly. The traditional single data source collection method is difficult to meet the needs of accurate prediction of resource distribution, affecting the accuracy of resource assessment.

[0003] Most of the existing assessment methods for deep-sea crust resources lack comprehensive consideration of the impact on the ecological environment. In particular, in resource development planning, it is difficult to balance the needs of resource development and ecological protection, which may lead to potential ecological risks, especially in the identification of ecologically sensitive areas and the prediction of the impact of resource exploitation on water quality and biological communities. The existing technical methods lack accurate evaluation and dynamic feedback mechanisms, resulting in uncertainty in the setting of resource development priorities and environmental protection measures. Therefore, there is an urgent need for a technical solution that can accurately assess the distribution of crust resources while taking the ecological environment impact into consideration, so as to provide a scientific basis for the sustainable development of deep-sea resources. Summary of the invention

[0004] Based on the above objectives, the present invention provides a deep-sea crust resource evaluation and prediction method based on machine learning.

[0005] A method for evaluating and predicting deep-sea crust resources based on machine learning comprises the following steps:

[0006] S1: Deploy multimodal sensors in the target deep sea area to collect multi-band acoustic data, magnetic field data, and seawater chemical parameter data, and record marine biological activity data and water flow rate data to generate an initial environmental data set;

[0007] S2: Based on the initial environmental data set, the K-means clustering algorithm is used to classify the environmental characteristics of the target area to identify ecologically sensitive areas, and the identified ecologically sensitive areas are marked as priority protection areas in resource development to form an environmental baseline database;

[0008] S3: Use convolutional neural networks to process the acoustic and magnetic field data in the initial environmental data set, extract the depth and density features related to the distribution of crust resources, and generate a crust resource feature data set;

[0009] S4: Based on the crust resource characteristic dataset and environmental baseline database, a generative adversarial network is used to simulate the ecological impact changes under different resource mining scenarios, generate environmental impact data in ecologically sensitive areas, and form an ecological impact feedback dataset;

[0010] S5: Based on the crust resource characteristic data set and the ecological impact feedback data set, a collaborative evaluation model is constructed. By setting weights to calculate the impact of different data characteristics, the crust resource distribution prediction value and the ecological impact prediction value are output to form a resource and ecological collaborative scoring result;

[0011] S6: Dynamically update the collaborative evaluation model, match and adjust the new sample data collected each time with the parameters in the collaborative evaluation model to improve the prediction accuracy and adaptability of the collaborative evaluation model under different time periods and environmental conditions;

[0012] S7: Based on the output of the collaborative assessment model, a resource and ecological impact distribution map of the target area is generated, and resource development priorities and environmental risk distribution are displayed through a graphical user interface.

[0013] Optionally, the S1 specifically includes:

[0014] S11: In the target deep sea area, deploy multimodal sensor arrays at different depths of 50 m, 100 m, 200 m and 300 m using remotely operated submersibles or autonomous underwater vehicles, with a horizontal spacing of 100 m between deployment points at each depth interval to cover the changes in environmental characteristics of the target area;

[0015] S12: adjusting the acoustic acquisition module of the multimodal sensor to a multi-band acquisition mode, setting three acquisition channels of low frequency, medium frequency and high frequency, respectively acquiring acoustic signals of each frequency band, and transmitting the acquired acoustic data to a host computer through a real-time data transmission interface to form a multi-band acoustic data set;

[0016] S13: using the three-axis magnetometer module in the multimodal sensor, based on the magnetic field distribution characteristics of the sensor deployment location, continuously collecting magnetic field data, recording the magnetic field strength, direction changes and interference factors of each deployment point, and generating a magnetic field data set of the target area;

[0017] S14: using a water quality detection module disposed on the multimodal sensor, chemical parameter data of oxygen content, pH value and metal ion concentration in seawater are collected in sequence, and timestamps are marked on the collected data, and the processed data are used to generate a seawater chemical parameter data set;

[0018] S15: Set up biological monitoring modules at the multimodal sensor deployment points to identify and record the characteristics of marine biological activities based on acoustic and optical detection technologies, record the species, frequency and intensity of active organisms, and generate a marine biological activity data set;

[0019] S16: Continuously collect water flow velocity data in the target area through the flow velocity measurement module, record the water flow direction, flow velocity change and water pressure information, and integrate the flow velocity data of each deployment point into a water flow velocity data set;

[0020] S17: Integrate multi-band acoustic datasets, magnetic field datasets, seawater chemical parameter datasets, marine biological activity datasets, and water flow velocity datasets, perform time and space synchronization on each dataset, and generate an initial environmental dataset.

[0021] Optionally, the S2 specifically includes:

[0022] S21: Extract multi-band acoustic data, magnetic field data, seawater chemical parameter data, marine biological activity data and water flow velocity data of the target area from the initial environmental data set as input feature variables of the K-means clustering algorithm, and normalize each feature variable to ensure the consistency of the dimension of the data features;

[0023] S22: setting the number of cluster centers of the K-means clustering algorithm, determining the K value according to the ecological complexity of the target area, preferably 5 to 10 categories, to represent areas with different ecological environmental characteristics;

[0024] S23: using the normalized feature variable data, calculating through the K-means clustering algorithm, dividing each data point in the initial environmental data set into the closest cluster center, forming environmental feature areas of different categories, and generating classification labels;

[0025] S24: Analyze the generated environmental characteristic areas, identify ecologically sensitive areas based on seawater chemical parameters, marine biological activity density, and water flow rate change characteristics, mark them as priority protection areas in resource development, and record the spatial coordinates, characteristic data, and classification labels of the priority protection areas;

[0026] S25: Import the marked data of the priority protection areas and each environmental characteristic area into the environmental baseline database, and save the characteristic classification results and environmental parameters of each area as baseline data.

[0027] Optionally, the S24 specifically includes:

[0028] S241: Based on the seawater chemical parameters in the environmental baseline database, calculate the mean and standard deviation of oxygen content, pH value and metal ion concentration to identify chemical anomaly areas; set the chemical sensitivity score of each monitoring point as S c ; For chemical sensitivity score S c Monitoring points above the set threshold are identified as sensitive areas with significant chemical characteristics;

[0029] S242: Based on the marine biological activity density data, normalize the biological activity frequency F at each monitoring point and calculate the biological activity density score S b ; When S b When the biological activity density is higher than the set threshold, the monitoring point is marked as an ecologically sensitive point with active organisms;

[0030] S243: Based on the water flow velocity variation characteristics, calculate the velocity variation rate R of each monitoring point v , to identify sensitive areas where water flow fluctuates violently; if R v When the rate of change is higher than the set threshold, the monitoring point is identified as a sensitive area with significant water flow fluctuations;

[0031] S244: Combined with the chemical sensitivity score S c , biological activity density score S b and the flow rate change rate R v , the monitoring points that meet at least two of the above three conditions are marked as ecologically sensitive areas; at the same time, their spatial coordinates, comprehensive feature scores and classification labels are recorded, and the ecologically sensitive areas are marked as priority protection areas in resource development.

[0032] Optionally, the S3 specifically includes:

[0033] S31: Preprocess the acoustic data and magnetic field data in the initial environmental data set, remove abnormal data and background noise through data normalization and noise filtering methods, and obtain high-quality input data;

[0034] S32: construct a convolutional neural network model, including an input layer, a convolution layer, a pooling layer and a fully connected layer, for extracting depth and density features from acoustic data and magnetic field data; the convolution layer extracts features from the input data through multiple 3×3 convolution kernels to identify density changes and depth distribution characteristics of crust resources;

[0035] S33: a pooling layer is set after the convolution layer to reduce the dimension of the extracted features through a maximum pooling operation, thereby reducing the data dimension and retaining the main information of the features;

[0036] S34: Input the feature vector output by the pooling layer to the fully connected layer, and generate a comprehensive feature vector of depth and density features by weighted summation;

[0037] S35: Associating the extracted comprehensive feature vector with the corresponding geographic coordinates and timestamp to generate the depth and density characteristics of the crust resources at each monitoring point, and integrating the feature data of all monitoring points to form a crust resource feature data set.

[0038] Optionally, the S4 specifically includes:

[0039] S41: Extract characteristic data of ecologically sensitive areas from the crust resource characteristic dataset and the environmental baseline database, including water quality chemical parameters, biological activity data, and flow rate information, as the initial input dataset of the generative adversarial network to simulate the baseline data of ecological environment changes;

[0040] S42: Construct a generative adversarial network model, which includes a generator and a discriminator. The generator is used to simulate the environmental impact of different resource extraction scenarios, including changes in water quality and changes in the structure of biological communities. The discriminator is used to compare the generated simulated data with the actual historical data to optimize the output effect of the generator.

[0041] S43: Input different mining scenario parameters into the generator, including mining depth, mining intensity and mining time, use the noise vector z combined with the mining scenario parameters as the input of the generator, and generate environmental impact data under these mining conditions through the generator; the output expression of the generator is: G(z, θ G )=X sim , where G represents the generator model, z is the noise vector, and θ G is the parameter of the generator, X sim To generate simulated environmental impact data;

[0042] S44: Input the generated simulated data X into the discriminator sim and the actual baseline data X real , the discriminator calculates the similarity between the two and feeds back the error to the generator; the output formula of the discriminator is: D(X) = σ(W D ·X+b D ), where D(X) is the output of the discriminator, indicating the probability that the input data X belongs to the actual data; W D is the weight matrix of the discriminator, X is the input data, b D is the bias term, σ is the activation function; the discriminator updates the parameter θ by optimizing D To maximize the probability of the discriminator recognizing the actual data, the generator continuously adjusts the parameter θ according to the feedback G To optimize the accuracy of the generated data;

[0043] S45: Through multiple rounds of iterative training of the generator and the discriminator, the generator is gradually able to simulate the changes in water quality and biological community structure under different mining scenarios; the final generated simulation data includes the changes in oxygen content, pH value, metal ion concentration and biological activity frequency environmental parameters;

[0044] S46: The environmental impact data output by the optimized generator is associated and integrated with the geographic coordinates of the ecologically sensitive areas and the mining scenario parameters to generate an ecological impact feedback data set.

[0045] Optionally, the S5 specifically includes:

[0046] S51: Extract depth features and density features from the crust resource feature dataset, and extract environmental impact data, including water quality parameters and biological activity density data, from the ecological impact feedback dataset as input variables of the collaborative evaluation model; and normalize each input variable to eliminate the impact of dimensional differences on the model;

[0047] S52: Construct a collaborative evaluation model, and synthesize the crust resource characteristic data and ecological impact data by weighted summation to form a resource-ecological score; the collaborative evaluation model adopts a linear weighting method, where the resource characteristic score is recorded as R and the ecological impact score is recorded as E; then the comprehensive score function S of the collaborative evaluation model is total Expressed as: S total =αR+βE, where S total is the comprehensive score of the collaborative evaluation model; α and β are the weight coefficients of the resource characteristic score and the ecological impact score, respectively, which are set according to the specific resource development priority and ecological protection needs; the resource characteristic score R and the ecological impact score E are defined as: R = w d D+w m M and E = w o O+w p P+w i I+w b B; where D is the depth characteristic value of the crust resource, M is the density characteristic value, and w d and w m are the weight coefficients of depth and density respectively; O is oxygen content, P is pH value, I is metal ion concentration, B is biological activity density, and w o ,w p ,w i and w b are the weight coefficients of each environmental impact parameter;

[0048] S53: Comprehensive score S calculated based on the collaborative evaluation model total, classify the resource development priorities and ecological protection needs of the target areas; specifically, by setting different scoring intervals, divide the scoring results into high-priority development areas, medium-priority development areas and low-priority development areas, and set corresponding ecological protection needs.

[0049] Optionally, the S53 specifically includes:

[0050] S531: Comprehensive score S calculated based on the collaborative evaluation model total ; Set high priority threshold T high and low priority threshold T low :Specifically when S total ≥T high When T low ≤S total <T high When S total <T low When , the area is marked as a low priority development area;

[0051] S532: For high-priority development zones, the ecological protection requirement is set at the lowest level, indicating that development activities should be given priority; for medium-priority development zones, a medium-level ecological protection requirement is set, indicating that development activities must be carried out while ensuring ecological impacts; for low-priority development zones, the highest level of ecological protection requirement is set, indicating that ecological protection should be given priority in this area, and development activities should be restricted or reduced.

[0052] Optionally, the S6 specifically includes:

[0053] S61: After each new sample data is collected, resource characteristic data and ecological impact data are extracted from the new data set, and data preprocessing is performed on the new sample data, including normalization and denoising, to ensure consistency with the original data characteristics in the collaborative evaluation model;

[0054] S62: Calculate the difference between the new sample data and the existing data set in the collaborative evaluation model, and evaluate the error between the new sample data and the current prediction value of the model through the mean square error;

[0055] S63: Based on the mean square error result, the gradient descent algorithm is used to gradually update the weight parameters of the collaborative evaluation model to reduce the error between the new sample data and the model prediction value. The weight update formula is: Among them, θ t+1 is the model parameter after the t+1th iteration, θ t is the model parameter at the tth iteration, η is the learning rate, which controls the step size of parameter update, is the gradient of the mean square error with respect to the model parameter θ, which is used to indicate the direction in which the error decreases;

[0056] S64: Apply the updated model parameters to the collaborative evaluation model and recalculate the resource characteristic score and ecological impact score.

[0057] Optionally, the S7 specifically includes:

[0058] S71: Based on the resource development priority and ecological protection demand data output by the collaborative evaluation model, the scoring results of each monitoring point are spatially interpolated according to the geographic coordinates of the target area to generate a resource distribution map and an ecological impact distribution map of the target area;

[0059] S72: Integrate the generated resource distribution map and ecological impact distribution map into one, and represent the resource development priority and environmental risk of the target area in the form of multi-level layers; each layer represents a high-priority development area, a medium-priority development area, a low-priority development area and the corresponding ecological risk distribution.

[0060] Beneficial effects of the present invention:

[0061] The present invention, by constructing a collaborative evaluation model based on machine learning, can not only efficiently and accurately extract the depth and density characteristics of deep-sea crust resources, but also integrate multi-source environmental data to generate resource distribution and ecological impact distribution maps, thereby improving the accuracy of resource assessment and the comprehensiveness of data. Through the identification of ecologically sensitive areas and real-time dynamic feedback, it realizes the accurate classification of resource development priorities and ecological protection needs.

[0062] The present invention simulates the ecological impacts under different resource exploitation scenarios through generative adversarial networks, can predict the potential impact of mining activities on water quality and biological communities, and provide a data basis for setting clear ecological protection measures for high-, medium- and low-priority development areas. This multi-level collaborative evaluation mechanism effectively balances the needs of resource development and ecological protection, ensuring that the scientific protection of the ecological environment is achieved while meeting the value of resource development. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0064] Figure 1 A schematic diagram of a method for evaluating and predicting deep-sea crust resources according to an embodiment of the present invention;

[0065] Figure 2Schematic diagram of a method for identifying ecologically sensitive areas according to an embodiment of the present invention. DETAILED DESCRIPTION

[0066] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.

[0067] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).

[0068] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0069] like Figure 1-Figure 2 As shown, a deep-sea crust resource evaluation and prediction method based on machine learning includes the following steps:

[0070] S1: Deploy multimodal sensors in the target deep sea area to collect multi-band acoustic data, magnetic field data, and seawater chemical parameter data (such as oxygen content, pH value, metal ion concentration), and record marine biological activity data and water flow rate data to generate an initial environmental data set;

[0071] S2: Based on the initial environmental data set, the K-means clustering algorithm is used to classify the environmental characteristics of the target area to identify ecologically sensitive areas, and the identified ecologically sensitive areas are marked as priority protection areas in resource development to form an environmental baseline database;

[0072] S3: Use convolutional neural networks to process the acoustic and magnetic field data in the initial environmental data set, extract the depth and density features related to the distribution of crust resources, and generate a crust resource feature data set;

[0073] S4: Based on the crust resource characteristic dataset and environmental baseline database, a generative adversarial network (GAN) is used to simulate the ecological impact changes under different resource extraction scenarios, generate environmental impact data in ecologically sensitive areas, including changes in water quality and biological community structure, and form an ecological impact feedback dataset;

[0074] S5: Based on the crust resource characteristic data set and the ecological impact feedback data set, a collaborative evaluation model is constructed. By setting weights to calculate the impact of different data characteristics, the crust resource distribution prediction value and the ecological impact prediction value are output to form a resource and ecological collaborative scoring result;

[0075] S6: Dynamically update the collaborative evaluation model, match and adjust the new sample data collected each time with the parameters in the collaborative evaluation model to improve the prediction accuracy and adaptability of the collaborative evaluation model under different time periods and environmental conditions;

[0076] S7: Based on the output of the collaborative evaluation model, a resource and ecological impact distribution map of the target area is generated, and the resource development priority and environmental risk distribution are displayed through a graphical user interface to provide decision support for subsequent deep-sea crust development and ecological protection.

[0077] S1 specifically includes:

[0078] S11: In the target deep-sea area, determine the deployment points and depths of multimodal sensors based on the terrain features and crust resource distribution; use remotely operated vehicles (ROVs) or autonomous underwater vehicles (AUVs) to deploy multimodal sensor arrays at different depths of 50 meters, 100 meters, 200 meters and 300 meters, respectively, and maintain a horizontal spacing of 100 meters between deployment points at each depth interval to cover the changes in environmental characteristics of the target area; through the built-in automatic leveling device, each sensor can be parallel to the seabed terrain after deployment to improve the stability and accuracy of data collection;

[0079] S12: The acoustic acquisition module of the multimodal sensor is adjusted to a multi-band acquisition mode, and three acquisition channels of low frequency (1-5kHz), medium frequency (5-15kHz) and high frequency (15-30kHz) are set to respectively acquire acoustic signals of each frequency band, and the acquired acoustic data is transmitted to the host computer through a real-time data transmission interface to form a multi-band acoustic data set;

[0080] S13: using the three-axis magnetometer module in the multimodal sensor, based on the magnetic field distribution characteristics of the sensor deployment location, continuously collect magnetic field data, record the magnetic field strength, direction change and interference factors of each deployment point, and generate a magnetic field data set of the target area;

[0081] S14: The chemical parameter data of oxygen content, pH value and metal ion concentration in seawater are collected in sequence through the water quality detection module arranged on the multimodal sensor, and the collected data are time-stamped to ensure the time synchronization of the chemical parameter data, and the processed data are used to generate a seawater chemical parameter data set;

[0082] S15: Set up biological monitoring modules at the multimodal sensor deployment points to identify and record the characteristics of marine biological activities based on acoustic and optical detection technologies, record the species, frequency and intensity of active organisms, and generate a marine biological activity data set;

[0083] S16: Continuously collect water flow velocity data in the target area through the flow velocity measurement module, record the water flow direction, flow velocity change and water pressure information, and integrate the flow velocity data of each deployment point into a water flow velocity data set;

[0084] S17: Integrate multi-band acoustic data sets, magnetic field data sets, seawater chemical parameter data sets, marine biological activity data sets and water flow rate data sets, perform time and space synchronization processing on each data set, generate an initial environmental data set, and provide reliable environmental basic data for subsequent crust resource evaluation and ecological protection prediction; the above steps adopt a multi-depth and multi-spacing deployment scheme to enable sensors to fully cover environmental characteristics at different depths and locations; the generated initial environmental data set includes comprehensive and accurate multiple environmental parameters, which provides a data basis for the accuracy of subsequent resource evaluation and ecological protection decision-making.

[0085] S2 specifically includes:

[0086] S21: Extract multi-band acoustic data, magnetic field data, seawater chemical parameter data, marine biological activity data and water flow velocity data of the target area from the initial environmental data set as input feature variables of the K-means clustering algorithm, and normalize each feature variable to ensure the consistency of the dimension of the data features;

[0087] S22: setting the number of cluster centers of the K-means clustering algorithm, determining the K value according to the ecological complexity of the target area, preferably 5 to 10 categories, which are used to represent areas with different ecological environmental characteristics, so as to subdivide the types of areas with different environmental sensitivities;

[0088] S23: using the normalized feature variable data, calculating through the K-means clustering algorithm, dividing each data point in the initial environmental data set into the closest cluster center, forming environmental feature areas of different categories, and generating classification labels;

[0089] S24: Analyze the generated environmental characteristic areas, identify ecologically sensitive areas based on seawater chemical parameters, marine biological activity density, and water flow rate change characteristics, mark them as priority protection areas in resource development, and record the spatial coordinates, characteristic data, and classification labels of the priority protection areas;

[0090] S25: Import the data of the marked priority protection areas and various environmental characteristic areas into the environmental baseline database, and save the characteristic classification results and environmental parameters of each area as baseline data for reference in subsequent resource evaluation and ecological protection steps; through the above steps, it is possible to achieve accurate classification of the environmental characteristics of the target area, and identify ecologically sensitive areas as priority protection objects. The generated environmental baseline database provides a clear environmental reference basis for the subsequent coordination of resource development and ecological protection.

[0091] S24 specifically includes:

[0092] S241: Based on the seawater chemical parameters in the environmental baseline database, calculate the mean and standard deviation of oxygen content, pH value and metal ion concentration to identify chemical anomaly areas; set the chemical sensitivity score of each monitoring point as S c , the formula is: Among them, S c is the chemical sensitivity score, O represents the oxygen content at the monitoring point, O avg is the average oxygen content of all monitoring points, σ O is the standard deviation of oxygen content; P represents the pH value of the monitoring point, P avg is the average pH value of all monitoring points, σ P is the standard deviation of pH value; M represents the metal ion concentration at the monitoring point, M avg is the average metal ion concentration of all monitoring points, σ M is the standard deviation of metal ion concentration; for chemical sensitivity score S c Monitoring points above the set threshold are identified as sensitive areas with significant chemical characteristics;

[0093] S242: Based on the marine biological activity density data, normalize the biological activity frequency F at each monitoring point and calculate the biological activity density score S b , the formula is: Among them, S b is the biological activity density score, F is the biological activity frequency at the monitoring point, and F max is the maximum biological activity frequency of all monitoring points in the area; when S b When the biological activity density is higher than the set threshold, the monitoring point is marked as an ecologically sensitive point with active organisms;

[0094] S243: Based on the water flow velocity variation characteristics, calculate the velocity variation rate R of each monitoring point v , to identify sensitive areas where water flow fluctuates violently; the flow velocity change rate is calculated by the following formula: Among them, R v is the flow rate change rate, V c represents the current flow velocity, V p represents the flow velocity at the previous moment, Δt is the time interval; if R v When the rate of change is higher than the set threshold, the monitoring point is identified as a sensitive area with significant water flow fluctuations;

[0095] S244: Combined chemical sensitivity score S c , biological activity density score S b and the flow rate change rate R v , the monitoring points that meet at least two of the above three conditions are marked as ecologically sensitive areas; at the same time, their spatial coordinates, comprehensive feature scores and classification labels are recorded, and the ecologically sensitive areas are marked as priority protection areas in resource development; through the above steps, based on the quantitative indicators of multiple environmental parameters, ecologically sensitive areas can be accurately identified and marked as priority protection areas, providing a clear basis for ecological protection in resource development.

[0096] S3 specifically includes:

[0097] S31: Preprocess the acoustic data and magnetic field data in the initial environmental data set, remove abnormal data and background noise through data normalization and noise filtering methods to obtain high-quality input data, and perform temporal and spatial sampling smoothing on each set of data to ensure data consistency and feature extractability;

[0098] S32: Construct a convolutional neural network (CNN) model, including an input layer, a convolution layer, a pooling layer, and a fully connected layer, to extract depth and density features from acoustic and magnetic field data; the convolution layer extracts features from the input data through multiple 3×3 convolution kernels to identify density changes and depth distribution characteristics of crust resources;

[0099] S33: After the convolution layer, a pooling layer is set to reduce the dimension of the extracted features through the maximum pooling operation, reduce the data dimension and retain the main information of the features to improve the data processing efficiency and the accuracy of feature recognition; the pooling operation is implemented by the following formula: pool (x)=max i,j∈W x i,j , where f pool (x) is the output feature value after pooling, x i,j represents the element value in the pooling window W, i and j are the indexes in the window; the pooling operation is performed on all x in the pooling window W.i,j The value is taken as the maximum value, thus achieving dimensionality reduction;

[0100] S34: Input the feature vector output by the pooling layer to the fully connected layer, and generate a comprehensive feature vector of depth and density features by weighted summation; the output calculation formula of the fully connected layer is: Among them, y is the output feature value of the fully connected layer, w k is the weight coefficient, x k is the feature value output by the pooling layer, b is the bias term, and n is the number of features output by the pooling layer;

[0101] S35: Associating the extracted comprehensive feature vector with the corresponding geographic coordinates and timestamp to generate the depth and density characteristics of the crust resources at each monitoring point, and integrating the feature data of all monitoring points to form a crust resource feature data set to provide input data for subsequent resource distribution prediction and evaluation; the above steps use the convolutional neural network model to achieve efficient extraction of depth and density features in acoustic and magnetic field data, and the generated crust resource feature data set can accurately reflect the resource distribution characteristics of the target area, providing comprehensive and reliable data support for resource evaluation.

[0102] S4 specifically includes:

[0103] S41: Extract characteristic data of ecologically sensitive areas from the crust resource characteristic dataset and the environmental baseline database, including water quality chemical parameters (oxygen content, pH value, metal ion concentration), biological activity data and flow rate information, as the initial input dataset of the generative adversarial network (GAN) to simulate the baseline data of ecological environment changes;

[0104] S42: Construct a generative adversarial network (GAN) model, which includes a generator and a discriminator. The generator is used to simulate the environmental impacts of different resource extraction scenarios, including changes in water quality and changes in the structure of biological communities. The discriminator is used to compare the generated simulated data with the actual historical data to optimize the output effect of the generator. The goal of the GAN model is to make the generated simulated data statistically match the actual data to ensure the credibility of the simulation.

[0105] S43: Input different mining scenario parameters into the generator, including mining depth, mining intensity and mining time, use the noise vector z combined with the mining scenario parameters as the input of the generator, and generate environmental impact data under these mining conditions through the generator; the output expression of the generator is: G(z, θ G )=X sim , where G represents the generator model, z is the noise vector, and θ G is the parameter of the generator, X sim To generate simulated environmental impact data;

[0106] S44: Input the generated simulated data X into the discriminator sim and the actual baseline data X real , the discriminator calculates the similarity between the two and feeds back the error to the generator; the output formula of the discriminator is: D(X) = σ(W D ·X+b D ), where D(X) is the output of the discriminator, indicating the probability that the input data X belongs to the actual data; W D is the weight matrix of the discriminator, X is the input data, b D is the bias term, σ is the activation function; the discriminator updates the parameter θ by optimizing D To maximize the probability of the discriminator recognizing the actual data, the generator continuously adjusts the parameter θ according to the feedback G To optimize the accuracy of the generated data;

[0107] S45: Through multiple rounds of iterative training of the generator and the discriminator, the generator is gradually able to simulate the changes in water quality and biological community structure under different mining scenarios; the final generated simulation data includes the changes in oxygen content, pH value, metal ion concentration and biological activity frequency environmental parameters;

[0108] S46: The environmental impact data output by the optimized generator is associated and integrated with the geographic coordinates of the ecologically sensitive areas and the mining scenario parameters to generate an ecological impact feedback dataset; this feedback dataset is used to provide ecological impact assessments under different resource mining scenarios, and provide data support for subsequent resource development and ecological protection; through the above steps, the generative adversarial network is used to realize the simulation of ecological impacts under different resource mining scenarios. The generated ecological impact feedback dataset can comprehensively reflect the potential environmental impact of mining activities on ecologically sensitive areas, and provide a key reference for environmental protection decisions in resource development.

[0109] The collaborative evaluation model constructed in S5 specifically includes:

[0110] S51: Extract depth characteristics and density characteristics from the crust resource characteristic data set, and extract environmental impact data from the ecological impact feedback data set, including water quality parameters (oxygen content, pH value, metal ion concentration) and biological activity density data, as input variables of the collaborative evaluation model; and normalize each input variable to eliminate the impact of dimensional differences on the model;

[0111] S52: Construct a collaborative evaluation model to combine crust resource characteristic data and ecological impact data by weighted summation to form a resource-ecological score; the collaborative evaluation model uses a linear weighting method, where the resource characteristic score is recorded as R and the ecological impact score is recorded as E; the comprehensive score function S of the collaborative evaluation model istotal Expressed as: S total =αR+βE, where S total is the comprehensive score of the collaborative evaluation model; α and β are the weight coefficients of resource characteristic score and ecological impact score, respectively, which are set according to the specific resource development priority and ecological protection needs; the resource characteristic score R and ecological impact score E are defined as: R = w d D+w m M and E = w o O+w p P+w i I+w b B; where d is the depth characteristic value of the crust resource, M is the density characteristic value, and w d and w m are the weight coefficients of depth and density respectively; O is oxygen content, P is pH value, I is metal ion concentration, B is biological activity density, and w o ,w p ,w i and w b are the weight coefficients of each environmental impact parameter respectively; through the above definition, the expression of the constructed collaborative evaluation model is: S total =α(w d D+w m M)+β(w o O+w p P+w i I+w b B);

[0112] S53: Comprehensive score S calculated based on the collaborative evaluation model total , classify the resource development priorities and ecological protection needs of the target area; specifically, by setting different scoring intervals, divide the scoring results into high-priority development areas, medium-priority development areas and low-priority development areas, and set corresponding ecological protection needs; through the above steps, the output of the collaborative evaluation model can provide a clear priority basis for resource development and ecological protection decisions in the target area.

[0113] S53 specifically includes:

[0114] S531: Comprehensive score S calculated based on the collaborative evaluation model total ; Set high priority threshold T high and low priority threshold T low :Specifically when S total ≥T high When T low ≤S total <T high When Stotal <T low When , the area is marked as a low priority development area;

[0115] S532: For high-priority development zones, because the resource development value of the area is high and the ecological sensitivity is low, the ecological protection demand is set to the lowest level, indicating that development activities should be given priority; for medium-priority development zones, the resource development value and ecological sensitivity are relatively balanced, and a medium-level ecological protection demand is set, indicating that development activities must be carried out while ensuring ecological impacts, requiring appropriate ecological protection measures; for low-priority development zones, because the resource development value of the area is low and the ecological sensitivity is high, the highest level of ecological protection demand is set, indicating that ecological protection should be given priority in the area, and development activities should be restricted or reduced to ensure the stability of the ecological environment; through the above steps, the regional development priorities can be reasonably divided based on the comprehensive score of the collaborative evaluation model, and corresponding ecological protection demands can be set to ensure effective protection of the ecological environment during resource development.

[0116] S6 specifically includes:

[0117] S61: Each time new sample data is collected, resource characteristic data (including depth and density characteristics) and ecological impact data (including oxygen content, pH value, metal ion concentration and biological activity density) are extracted from the new data set, and the new sample data are preprocessed, including normalization and denoising, to ensure consistency with the original data characteristics in the collaborative evaluation model;

[0118] S62: Calculate the difference between the new sample data and the existing data set in the collaborative evaluation model, and evaluate the error between the new sample data and the current prediction value of the model through the mean square error (MSE); the calculation formula of the mean square error is: Among them, MSE is the mean square error, N is the number of samples, and y i is the actual observed value of the ith sample, is the model prediction value of the i-th sample; the mean square error is used to measure the adaptability of the current parameters of the model;

[0119] S63: Based on the mean square error result, the gradient descent algorithm is used to gradually update the weight parameters of the collaborative evaluation model to reduce the error between the new sample data and the model prediction value. The weight update formula is: Among them, θ t+1 is the model parameter after the t+1th iteration, θ t is the model parameter at the tth iteration, η is the learning rate, which controls the step size of parameter update, is the gradient of the mean square error with respect to the model parameter θ, which is used to indicate the direction in which the error decreases;

[0120] S64: Apply the updated model parameters to the collaborative evaluation model, and recalculate the resource characteristic scores and ecological impact scores to ensure that the model maintains high prediction accuracy in new environmental conditions and time periods. After each update, verify the prediction effect of the model on the new sample data to ensure that the error is within an acceptable range. Through the above steps, the collaborative evaluation model can dynamically adapt to new environmental conditions and data changes, continuously optimize prediction accuracy, and provide more accurate decision-making support for resource development and ecological protection.

[0121] S7 specifically includes:

[0122] S71: Based on the resource development priority and ecological protection demand data output by the collaborative evaluation model, the scoring results of each monitoring point are spatially interpolated according to the geographic coordinates of the target area to generate a resource distribution map and ecological impact distribution map of the target area. The spatial interpolation is calculated using the inverse distance weighted (IDW) algorithm, and the interpolation formula is: Among them, V p is the predicted value of the interpolation point p, V i is the score value of the i-th monitoring point, d i is the distance between the interpolation point and the i-th monitoring point, k is the distance weight index, and n1 is the number of monitoring points involved in the interpolation. This formula ensures a smooth transition of the score in space and forms a coherent distribution map;

[0123] S72: Integrate the generated resource distribution map and ecological impact distribution map into one, and represent the resource development priority and environmental risk of the target area in the form of multi-level layers; each layer represents a high-priority development area, a medium-priority development area, a low-priority development area and the corresponding ecological risk distribution, so as to realize the spatial visualization of resource and ecological information; through the above steps, the resource distribution map and the ecological impact distribution map can intuitively present the resource development priority and ecological protection needs of the target area, and provide data support for subsequent resource development planning and ecological protection decision-making.

[0124] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.

[0125] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for evaluating and predicting deep-sea crust resources based on machine learning, characterized in that: The following steps are involved: S1: Deploy multimodal sensors in the target deep sea area to collect multi-band acoustic data, magnetic field data, and seawater chemical parameter data, and record marine biological activity data and water flow rate data to generate an initial environmental data set; S2: Based on the initial environmental data set, the K-means clustering algorithm is used to classify the environmental characteristics of the target area to identify ecologically sensitive areas, and the identified ecologically sensitive areas are marked as priority protection areas in resource development to form an environmental baseline database; S3: Use convolutional neural networks to process the acoustic and magnetic field data in the initial environmental data set, extract the depth and density features related to the distribution of crust resources, and generate a crust resource feature data set; S4: Based on the crust resource characteristic dataset and environmental baseline database, a generative adversarial network is used to simulate the ecological impact changes under different resource mining scenarios, generate environmental impact data in ecologically sensitive areas, and form an ecological impact feedback dataset; S5: Based on the crust resource characteristic data set and the ecological impact feedback data set, a collaborative evaluation model is constructed. By setting weights to calculate the impact of different data characteristics, the crust resource distribution prediction value and the ecological impact prediction value are output to form a resource and ecological collaborative scoring result; S6: Dynamically update the collaborative evaluation model, match and adjust the new sample data collected each time with the parameters in the collaborative evaluation model to improve the prediction accuracy and adaptability of the collaborative evaluation model under different time periods and environmental conditions; S7: Based on the output of the collaborative assessment model, a resource and ecological impact distribution map of the target area is generated, and resource development priorities and environmental risk distribution are displayed through a graphical user interface.

2. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 1, characterized in that: The S1 specifically includes: S11: In the target deep sea area, deploy multimodal sensor arrays at different depths of 50 m, 100 m, 200 m and 300 m using remotely operated submersibles or autonomous underwater vehicles, with a horizontal spacing of 100 m between deployment points at each depth interval to cover the changes in environmental characteristics of the target area; S12: adjusting the acoustic acquisition module of the multimodal sensor to a multi-band acquisition mode, setting three acquisition channels of low frequency, medium frequency and high frequency, respectively acquiring acoustic signals of each frequency band, and transmitting the acquired acoustic data to a host computer through a real-time data transmission interface to form a multi-band acoustic data set; S13: using the three-axis magnetometer module in the multimodal sensor, based on the magnetic field distribution characteristics of the sensor deployment location, continuously collecting magnetic field data, recording the magnetic field strength, direction changes and interference factors of each deployment point, and generating a magnetic field data set of the target area; S14: using a water quality detection module disposed on the multimodal sensor, chemical parameter data of oxygen content, pH value and metal ion concentration in seawater are collected in sequence, and timestamps are marked on the collected data, and the processed data are used to generate a seawater chemical parameter data set; S15: Set up biological monitoring modules at the multimodal sensor deployment points to identify and record the characteristics of marine biological activities based on acoustic and optical detection technologies, record the species, frequency and intensity of active organisms, and generate a marine biological activity data set; S16: Continuously collect water flow velocity data in the target area through the flow velocity measurement module, record the water flow direction, flow velocity change and water pressure information, and integrate the flow velocity data of each deployment point into a water flow velocity data set; S17: Integrate multi-band acoustic datasets, magnetic field datasets, seawater chemical parameter datasets, marine biological activity datasets, and water flow velocity datasets, perform time and space synchronization on each dataset, and generate an initial environmental dataset.

3. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 1, characterized in that: The S2 specifically includes: S21: Extract multi-band acoustic data, magnetic field data, seawater chemical parameter data, marine biological activity data and water flow velocity data of the target area from the initial environmental data set as input feature variables of the K-means clustering algorithm, and normalize each feature variable to ensure the consistency of the dimension of the data features; S22: Set the number of cluster centers of the K-means clustering algorithm, and determine the K value according to the ecological complexity of the target area, where the K value is 5 to 10 categories, which are used to represent areas with different ecological environmental characteristics; S23: using the normalized feature variable data, calculating through the K-means clustering algorithm, dividing each data point in the initial environmental data set into the closest cluster center, forming environmental feature areas of different categories, and generating classification labels; S24: Analyze the generated environmental characteristic areas, identify ecologically sensitive areas based on seawater chemical parameters, marine biological activity density, and water flow rate change characteristics, mark them as priority protection areas in resource development, and record the spatial coordinates, characteristic data, and classification labels of the priority protection areas; S25: Import the marked data of the priority protection areas and each environmental characteristic area into the environmental baseline database, and save the characteristic classification results and environmental parameters of each area as baseline data.

4. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 3, characterized in that: The S24 specifically includes: S241: Based on the seawater chemical parameters in the environmental baseline database, calculate the mean and standard deviation of oxygen content, pH value and metal ion concentration to identify chemical anomaly areas; set the chemical sensitivity score of each monitoring point as S c ; For chemical sensitivity score S c Monitoring points above the set threshold are identified as sensitive areas with significant chemical characteristics; S242: Based on the marine biological activity density data, normalize the biological activity frequency F at each monitoring point and calculate the biological activity density score S b ; When S b When the biological activity density is higher than the set threshold, the monitoring point is marked as an ecologically sensitive point with active organisms; S243: Based on the water flow velocity variation characteristics, calculate the velocity variation rate R of each monitoring point v , to identify sensitive areas where water flow fluctuates violently; if R v When the rate of change is higher than the set threshold, the monitoring point is identified as a sensitive area with significant water flow fluctuations; S244: Combined with the chemical sensitivity score S c , biological activity density score S b and the flow rate change rate R v , the monitoring points that meet at least two of the above three conditions are marked as ecologically sensitive areas; at the same time, their spatial coordinates, comprehensive feature scores and classification labels are recorded, and the ecologically sensitive areas are marked as priority protection areas in resource development.

5. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 1, characterized in that: The S3 specifically includes: S31: Preprocess the acoustic data and magnetic field data in the initial environmental data set, remove abnormal data and background noise through data normalization and noise filtering methods, and obtain high-quality input data; S32: construct a convolutional neural network model, including an input layer, a convolution layer, a pooling layer and a fully connected layer, for extracting depth and density features from acoustic data and magnetic field data; the convolution layer extracts features from the input data through multiple 3×3 convolution kernels to identify density changes and depth distribution characteristics of crust resources; S33: a pooling layer is set after the convolution layer to reduce the dimension of the extracted features through a maximum pooling operation, thereby reducing the data dimension and retaining the main information of the features; S34: Input the feature vector output by the pooling layer to the fully connected layer, and generate a comprehensive feature vector of depth and density features by weighted summation; S35: Associating the extracted comprehensive feature vector with the corresponding geographic coordinates and timestamp to generate the depth and density characteristics of the crust resources at each monitoring point, and integrating the feature data of all monitoring points to form a crust resource feature data set.

6. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 1, characterized in that: The S4 specifically includes: S41: Extract characteristic data of ecologically sensitive areas from the crust resource characteristic dataset and the environmental baseline database, including water quality chemical parameters, biological activity data, and flow rate information, as the initial input dataset of the generative adversarial network to simulate the baseline data of ecological environment changes; S42: Construct a generative adversarial network model, which includes a generator and a discriminator. The generator is used to simulate the environmental impact of different resource extraction scenarios, including changes in water quality and changes in the structure of biological communities. The discriminator is used to compare the generated simulated data with the actual historical data to optimize the output effect of the generator. S43: Input different mining scenario parameters into the generator, including mining depth, mining intensity and mining time, use the noise vector z combined with the mining scenario parameters as the input of the generator, and generate environmental impact data under these mining conditions through the generator; the output expression of the generator is: G(z, θ G )=X sim , where G represents the generator model, z is the noise vector, and θ G is the parameter of the generator, X sim To generate simulated environmental impact data; S44: Input the generated simulated data X into the discriminator sim and the actual baseline data X real , the discriminator calculates the similarity between the two and feeds back the error to the generator; the output formula of the discriminator is: D(X) = σ(W D ·X+b D ), where D(X) is the output of the discriminator, indicating the probability that the input data X belongs to the actual data; W D is the weight matrix of the discriminator, X is the input data, b D is the bias term, σ is the activation function; the discriminator updates the parameter θ by optimizing D To maximize the recognition probability of the discriminator for the actual data, the generator continuously adjusts the parameter θ according to the feedback G To optimize the accuracy of the generated data; S45: Through multiple rounds of iterative training of the generator and the discriminator, the generator is gradually able to simulate the changes in water quality and biological community structure under different mining scenarios; the final generated simulation data includes the changes in oxygen content, pH value, metal ion concentration and biological activity frequency environmental parameters; S46: The environmental impact data output by the optimized generator is associated and integrated with the geographic coordinates of the ecologically sensitive areas and the mining scenario parameters to generate an ecological impact feedback data set.

7. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 1, characterized in that: The S5 specifically includes: S51: Extract depth features and density features from the crust resource feature dataset, and extract environmental impact data, including water quality parameters and biological activity density data, from the ecological impact feedback dataset as input variables of the collaborative evaluation model; and normalize each input variable to eliminate the impact of dimensional differences on the model; S52: Construct a collaborative evaluation model, and synthesize the crust resource characteristic data and ecological impact data by weighted summation to form a resource-ecological score; the collaborative evaluation model adopts a linear weighting method, where the resource characteristic score is recorded as R and the ecological impact score is recorded as E; then the comprehensive score function S of the collaborative evaluation model is total Expressed as: S total =αR+βE, where S total is the comprehensive score of the collaborative evaluation model; α and β are the weight coefficients of the resource characteristic score and the ecological impact score, respectively, which are set according to the specific resource development priority and ecological protection needs; the resource characteristic score R and the ecological impact score E are defined as: R = w d D+w m M and E = w o O+w p P+w i I+w b B; where D is the depth characteristic value of the crust resource, M is the density characteristic value, and w d and w m are the weight coefficients of depth and density respectively; O is oxygen content, P is pH value, I is metal ion concentration, B is biological activity density, and w o ,w p ,w i and w b are the weight coefficients of each environmental impact parameter; S53: Comprehensive score S calculated based on the collaborative evaluation model total , classify the resource development priorities and ecological protection needs of the target areas; specifically, by setting different scoring intervals, divide the scoring results into high-priority development areas, medium-priority development areas and low-priority development areas, and set corresponding ecological protection needs.

8. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 7, characterized in that: The S53 specifically includes: S531: Comprehensive score S calculated based on the collaborative evaluation model total ; Set high priority threshold T high and low priority threshold T low :Specifically when S total ≥T high When T low ≤S total <T high When S total <T low When , the area is marked as a low priority development area; S532: For high-priority development zones, the ecological protection requirement is set at the lowest level, indicating that development activities should be given priority; for medium-priority development zones, a medium-level ecological protection requirement is set, indicating that development activities must be carried out while ensuring ecological impacts; for low-priority development zones, the highest level of ecological protection requirement is set, indicating that ecological protection should be given priority in this area, and development activities should be restricted or reduced.

9. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 1, characterized in that: The S6 specifically includes: S61: After each new sample data is collected, resource characteristic data and ecological impact data are extracted from the new data set, and data preprocessing is performed on the new sample data, including normalization and denoising, to ensure consistency with the original data characteristics in the collaborative evaluation model; S62: Calculate the difference between the new sample data and the existing data set in the collaborative evaluation model, and evaluate the error between the new sample data and the current prediction value of the model through the mean square error; S63: Based on the mean square error result, the gradient descent algorithm is used to gradually update the weight parameters of the collaborative evaluation model to reduce the error between the new sample data and the model prediction value. The weight update formula is: Among them, θ t+1 is the model parameter after the t+1th iteration, θ t is the model parameter at the tth iteration, η is the learning rate, which controls the step size of parameter update, is the gradient of the mean square error with respect to the model parameter θ, which is used to indicate the direction in which the error decreases; S64: Apply the updated model parameters to the collaborative evaluation model and recalculate the resource characteristic score and ecological impact score.

10. The method for evaluating and predicting deep-sea crust resources based on machine learning according to claim 1, characterized in that: The S7 specifically includes: S71: Based on the resource development priority and ecological protection demand data output by the collaborative evaluation model, the scoring results of each monitoring point are spatially interpolated according to the geographic coordinates of the target area to generate a resource distribution map and an ecological impact distribution map of the target area; S72: Integrate the generated resource distribution map and ecological impact distribution map into one, and represent the resource development priority and environmental risk of the target area in the form of multi-level layers; each layer represents a high-priority development area, a medium-priority development area, a low-priority development area and the corresponding ecological risk distribution.

Images