Machine learning-based alpine steep slope ecological restoration scheme recommendation method and system

Abstract

The application provides a high-cold steep slope ecological restoration scheme recommendation method and system based on machine learning, historical restoration case data sets and regional geographic feature data of a high-cold steep slope region are collected, the historical restoration case data sets and the regional geographic feature data are associated and fused to generate a regional ecological restoration feature map with spatio-temporal correlation, a pre-trained restoration scheme optimization model is called to perform multi-dimensional adaptability analysis and processing on the regional ecological restoration feature map to obtain an ecological restoration scheme adaptability score matrix of the high-cold steep slope region, and based on the ecological restoration scheme adaptability score matrix and the regional ecological restoration feature map, an ecological restoration scheme spatial recommendation distribution map of the high-cold steep slope region is generated. The application improves the accuracy and practicability of the ecological restoration scheme recommendation.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a method and system for intelligent recommendation of ecological restoration schemes for high-altitude and steep slopes based on machine learning. Background Technology

[0002] Currently, the development of ecological restoration plans for high-altitude and steep slopes largely relies on the manual summarization of historical restoration cases by technical personnel. For example, by compiling implementation records of restoration measures in different regions and combining them with statistical analysis of individual geographical parameters such as topography and slope, recommendations for restoration measures tailored to specific areas can be formed. However, empirical data such as vegetation growth status and the effectiveness of restoration measures in historical restoration cases lack structured integration with geographical feature data such as regional topography, soil, and climate. This fragmented information makes it difficult to form a systematic decision-making basis through quantitative matching relationships, resulting in insufficient consideration of the complex relationships among multiple factors in the recommendation process. Summary of the Invention

[0003] In view of this, the present invention provides an intelligent recommendation method and system for ecological restoration schemes on high-altitude and steep slopes based on machine learning.

[0004] The technical solution of this invention is implemented as follows: On one hand, embodiments of the present invention provide an intelligent recommendation method for ecological restoration schemes on high-altitude and steep slopes based on machine learning. The method includes: collecting a set of historical restoration case data and regional geographic feature data of the high-altitude and steep slope region; performing correlation and fusion processing on the set of historical restoration case data and the regional geographic feature data to generate a regional ecological restoration feature map with spatiotemporal correlation, wherein each node in the regional ecological restoration feature map corresponds to a restoration case unit, and the edge weight between nodes represents the matching degree value between the case unit and the regional geographic features; calling a pre-trained restoration scheme optimization model to perform multi-dimensional adaptability analysis processing on the regional ecological restoration feature map to obtain an ecological restoration scheme adaptability scoring matrix for the high-altitude and steep slope region, wherein the ecological restoration scheme adaptability scoring matrix contains the adaptability quantification value of different restoration measure combinations in each sub-region of the region; and generating a spatial recommendation distribution map of ecological restoration schemes for the high-altitude and steep slope region based on the ecological restoration scheme adaptability scoring matrix and the regional ecological restoration feature map, wherein each geographic grid unit in the spatial recommendation distribution map of the ecological restoration scheme is labeled with the corresponding optimal restoration measure combination and implementation priority ranking.

[0005] On the other hand, embodiments of the present invention provide a computer system including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the program to implement the steps in the above-described method.

[0006] This invention provides an intelligent recommendation method for ecological restoration schemes on high-altitude and steep slopes based on machine learning. It collects historical restoration case data and regional geographical feature data for these areas, and then performs correlation and fusion processing on these two types of data to generate a regional ecological restoration feature map with spatiotemporal correlation. Each node corresponds to a restoration case unit, and the edge weights between nodes represent the matching degree between the case unit and the regional geographical features. This data processing method organically integrates scattered historical restoration experience and regional geographical features through quantitative matching relationships. By calling a pre-trained restoration scheme optimization model, a multi-dimensional adaptability analysis is performed on the regional ecological restoration feature map to obtain an ecological restoration scheme adaptability scoring matrix. This matrix contains quantitative adaptability values ​​for different combinations of restoration measures in each sub-region within the region. This allows the analysis process to directly utilize the inherent connection between historical cases and geographical features, while simultaneously considering the adaptability of both the measure combination and sub-region dimensions. This solves the problems of insufficient utilization of correlation information from multi-source heterogeneous data and evaluation based solely on a single measure or the entire region. Based on the ecological restoration scheme adaptability scoring matrix and regional ecological restoration feature map, a spatial distribution map of recommended ecological restoration schemes for high-altitude and steep slope areas is generated using GIS spatial visualization technology. The abstract adaptability quantification value is bound to specific geographic grid units, and the corresponding optimal combination of restoration measures and implementation priority ranking are marked, forming a three-dimensional decision support system of measures-space-time. This solves the limitations of conventional numerical results, which are difficult to intuitively guide spatial implementation and only output a list of measures. It is specifically designed for the actual needs of ecological restoration projects, where measures need to be adapted to local conditions, implemented in different regions, and implemented in stages due to limited resources, thus improving the accuracy and practicality of ecological restoration scheme recommendations. Attached Figure Description

[0007] Figure 1 This is a schematic diagram illustrating the implementation process of an intelligent recommendation method for ecological restoration schemes on steep, cold slopes based on machine learning, provided in an embodiment of the present invention.

[0008] Figure 2 This is a schematic diagram of the hardware entity of a computer system provided in an embodiment of the present invention. Detailed Implementation

[0009] This invention provides an intelligent recommendation method for ecological restoration schemes on steep, cold slopes based on machine learning. This method can be executed by a processor of a computer system. The computer system can refer to devices with data processing capabilities, such as servers, laptops, tablets, and desktop computers.

[0010] Figure 1 This is a schematic diagram illustrating the implementation process of an intelligent recommendation method for ecological restoration schemes on steep, cold slopes based on machine learning, as provided in an embodiment of the present invention. Figure 1 As shown, the method includes: Step S100: Collect historical restoration case data set and regional geographical feature data for high-altitude and steep slope areas. The historical restoration case data set includes records of vegetation growth status and corresponding restoration measures implementation records for different restoration periods. The regional geographical feature data includes records of topographic slope distribution, soil composition distribution, and climate condition changes.

[0011] The historical restoration case dataset is a comprehensive record of past ecological restoration work in high-altitude, steep slope areas. The vegetation growth status records for different restoration periods detail the vegetation's growth at each stage, including information such as height, coverage, and the number and types of vegetation. This information reflects the vegetation's growth status and health at different points in time. The corresponding restoration implementation records document the specific restoration methods used at each restoration period, such as the types of vegetation planted, the type and frequency of fertilization, and the method and amount of irrigation. Regional geographic feature data is a crucial dataset describing the natural geographic conditions of high-altitude, steep slope areas. Topographic slope distribution records, through measurement and analysis, clarify the slope magnitude and distribution at different locations within the region. Soil composition distribution records provide a detailed analysis of the content and distribution of various soil components, including the content of nutrients such as organic matter, nitrogen, phosphorus, and potassium, as well as information on soil pH and texture. This information directly affects vegetation growth and the living environment. Climate condition change records document long-term climate changes within the region, such as the patterns of change in meteorological elements like temperature, precipitation, and sunlight.

[0012] To collect this data, satellite remote sensing technology can be used to periodically acquire vegetation imagery data for recording vegetation growth at different restoration periods. Image processing and analysis algorithms can then be used to extract relevant vegetation information. Simultaneously, monitoring points can be set up in the field for regular manual measurement and recording. For records of corresponding restoration measures, accurate information can be obtained by reviewing historical archives and interviewing relevant personnel. For records of topographic slope distribution, GPS and GIS technologies, combined with topographic surveying instruments, can be used to accurately measure and analyze the topography of the area. For records of soil composition distribution, soil samples can be collected for laboratory analysis to obtain detailed soil composition information. For records of climate change, meteorological data can be collected from weather stations and then statistically analyzed.

[0013] Step S200: Perform correlation and fusion processing on the historical restoration case dataset and regional geographic feature data to generate a regional ecological restoration feature map with spatiotemporal correlation. Each node in the regional ecological restoration feature map corresponds to a restoration case unit, and the edge weight between nodes represents the matching degree value between the case unit and the regional geographic features.

[0014] Association and fusion processing organically combines historical restoration case datasets with regional geographic feature data to uncover their inherent connections and interactions. A regional ecological restoration feature map with spatiotemporal correlation is a graphical representation that displays the relationship between restoration case units and regional geographic features in the form of nodes and edges. Each node corresponds to a restoration case unit, representing a specific restoration case and containing relevant information such as restoration measures and vegetation growth status. The edge weights between nodes represent the matching degree between the case unit and the regional geographic features. This value reflects the applicability and effectiveness of the restoration case under the regional geographic conditions. A higher matching degree indicates a greater fit between the restoration case and the regional geographic features, and a higher likelihood of achieving good restoration results in the region.

[0015] As one implementation method, step S200 can be specifically implemented as the following steps S210~S250: Step S210: Perform time-series interpolation on the vegetation growth status records in the historical restoration case dataset to obtain a continuous time series vegetation cover change curve. The time interval of the vegetation cover change curve is consistent with the sampling period of the climate condition change records.

[0016] Temporal interpolation is used to estimate missing data values ​​between discrete time points, thus obtaining continuous time-series data. In historical restoration case datasets, vegetation growth records may be collected at discontinuous time points, resulting in gaps in the data. Temporal interpolation allows for the estimation of vegetation cover values ​​in these gaps using known vegetation cover data and appropriate interpolation algorithms, yielding a continuous vegetation cover change curve. This curve visually illustrates how vegetation cover changes over time, reflecting the growth and development trend of vegetation. Setting the time interval of the vegetation cover change curve to match the sampling period of climate condition change records facilitates subsequent analysis of the relationship between vegetation growth and climate conditions. Various interpolation algorithms, such as linear interpolation and spline interpolation, can be used for temporal interpolation.

[0017] Step S220: Perform contour density analysis on the topographic slope distribution records in the regional geographic feature data to generate a topographic zoning vector map based on slope intervals. The boundary lines of the topographic zoning vector map are correlated with the sampling point distribution density of the soil composition distribution records.

[0018] Contour density analysis is a method for in-depth analysis of topographic slope distribution records. By calculating the density of contour lines, the undulation of the terrain can be understood. Contour lines are curves formed by connecting points of the same elevation on a topographic map. The density of contour lines reflects the steepness of the terrain; the higher the density, the steeper the terrain. By analyzing the contour line density, the terrain can be divided into different slope intervals, each representing different terrain features. A topographic zoning vector map based on slope intervals is a graphical representation that divides the terrain according to slope intervals and displays it in vector graphic form. The boundary lines of the topographic zoning vector map are correlated with the sampling point distribution density of soil composition records to better combine topographic and soil information for ecological restoration analysis. Soil composition may vary on terrains with different slopes.

[0019] When performing contour density analysis, relevant tools in Geographic Information System (GIS) software can be used. The terrain slope distribution records are imported into the GIS software to generate a contour map. Then, using the analysis functions of the GIS software, the density of the contour lines is calculated and classified according to preset slope intervals. Finally, the classification results are converted into vector graphics to generate a terrain zoning vector map. When associating the boundary lines of the terrain zoning vector map with the sampling point distribution density of the soil composition distribution records, spatial analysis methods can be used to overlay the locations of soil sampling points with the terrain zoning vector map, adjusting the boundary lines of the terrain zoning vector map based on the distribution density of the sampling points.

[0020] Step S230: Input the vegetation cover change curve and the topographic zoning vector map into the spatiotemporal correlation module, calculate the spatial correlation coefficient between the vegetation cover change rate and the soil composition distribution record in each topographic zoning, and generate a spatiotemporal correlation matrix containing the correlation coefficient.

[0021] The spatiotemporal correlation module is used to analyze the spatiotemporal relationships between different data. After inputting the vegetation cover change curve and topographic zoning vector map into the module, it calculates the spatial correlation between the vegetation cover change rate and soil composition within each topographic zone, in conjunction with soil composition distribution records. The vegetation cover change rate reflects the degree of change in vegetation cover over a certain period of time, and it may have a close relationship with soil composition. For example, factors such as soil nutrient content and pH may affect vegetation growth and development, thus leading to changes in vegetation cover. By calculating the spatial correlation coefficient, the strength and direction of this relationship can be quantified. The spatial correlation coefficient is a numerical value ranging from -1 to 1, with positive values ​​indicating positive correlation and negative values ​​indicating negative correlation. The closer the absolute value is to 1, the stronger the correlation. The generated spatiotemporal correlation matrix containing the correlation coefficient is a two-dimensional matrix, where rows represent different topographic zones and columns represent different soil composition indicators. Each element in the matrix represents the spatial correlation coefficient between the vegetation cover change rate and the corresponding soil composition indicator within the corresponding topographic zone.

[0022] In the spatiotemporal correlation module, various statistical methods can be used to calculate spatial correlation coefficients, such as the Pearson correlation coefficient and the Spearman correlation coefficient. The Pearson correlation coefficient is obtained by calculating the covariance and standard deviation between variables. The Spearman correlation coefficient, on the other hand, is a non-parametric correlation coefficient that does not depend on the distribution of variables but is calculated based on the rank of the variables, making it suitable for handling non-linear data. When choosing a correlation coefficient calculation method, a comprehensive consideration should be given to the characteristics of the data and the actual needs.

[0023] Step S240: Construct a set of node attributes for the regional ecological restoration feature map based on the spatiotemporal correlation matrix.

[0024] The node attribute set includes vegetation recovery rate, soil organic matter content change trend and climate adaptability index. The edge weights between nodes are obtained by normalizing the element values ​​of the spatiotemporal correlation matrix. The vegetation recovery rate is calculated by the slope of the vegetation cover change curve. The soil organic matter content change trend is extracted by the temporal fluctuation characteristics of the soil component distribution record. The climate adaptability index is obtained by the correlation analysis between the vegetation cover change curve and the climate condition change record.

[0025] The node attribute set of the regional ecological restoration feature map is a crucial information set describing the characteristics of restoration case units, including key attributes such as vegetation recovery rate, soil organic matter content change trend, and climate adaptability index. Vegetation recovery rate reflects the growth speed of vegetation during the ecological restoration process. By calculating the slope of the vegetation cover change curve, the change in vegetation cover per unit time can be obtained, thus measuring the vegetation recovery speed. Soil organic matter content change trend refers to the change in soil organic matter content over time. By performing time-series analysis on soil component distribution records and extracting fluctuation characteristics, such as upward trends, downward trends, or stable states, the trend of soil organic matter content change can be understood. The climate adaptability index is an indicator for assessing the vegetation's adaptability to local climatic conditions. By performing correlation analysis between the vegetation cover change curve and climatic condition change records, such as calculating their correlation coefficients and regression coefficients, the growth response of vegetation under different climatic conditions can be obtained, thereby determining the vegetation's climate adaptability index.

[0026] The edge weights between nodes represent the matching degree between the case unit and the regional geographical features, obtained by normalizing the element values ​​of the spatiotemporal correlation matrix. Normalization is the process of converting the original data into values ​​within a certain range, such as normalizing the data to the [0,1] interval. This is to eliminate the dimensional differences between different variables, making the edge weights comparable. The normalized edge weights can more accurately reflect the degree of matching between the case unit and the regional geographical features.

[0027] When constructing the node attribute set, numerical differentiation methods, such as the finite difference method, can be used to calculate the slope of the vegetation cover change curve at different time points to determine the vegetation recovery rate. For extracting the trend of soil organic matter content changes, time series analysis methods, such as the moving average method and the autoregressive integral moving average model (ARIMA), can be used to model and analyze soil component distribution records, thereby extracting trend characteristics. For calculating the climate adaptability index, statistical analysis methods, such as correlation analysis and regression analysis, can be used to analyze the relationship between the vegetation cover change curve and climate condition change records, thus obtaining the climate adaptability index.

[0028] Step S250: Optimize the graph structure of the regional ecological restoration feature map, delete redundant connections with edge weights below a preset threshold, retain the core node connection paths with significant spatiotemporal correlation, and generate a simplified regional ecological restoration feature map.

[0029] Map structure optimization is a process of further optimizing and refining regional ecological restoration feature maps, aiming to improve the map's quality and usability. In the regional ecological restoration feature map, edge weights represent the matching degree between case units and regional geographical features. Connections with edge weights below a preset threshold indicate that these case units have a low matching degree with regional geographical features, contribute little to the formulation and analysis of ecological restoration plans, and are considered redundant connections. By removing these redundant connections, the complexity of the map can be reduced, improving its readability and analytical efficiency.

[0030] Core node connection paths with significant spatiotemporal correlations are those remaining after removing redundant connections with edge weights below a preset threshold. These paths are of significant importance in the graph, representing a strong spatiotemporal correlation between case units and regional geographical features. Retaining these core node connection paths highlights key information in the graph, allowing it to focus more on important ecological restoration factors and relationships.

[0031] A preset threshold is a pre-defined value used to determine whether the edge weights are below a certain standard. In practical applications, the specific preset threshold can be determined based on the specific research objectives and data characteristics. By adjusting the size of the preset threshold, the degree of simplification of the graph can be controlled.

[0032] When optimizing the graph structure, graph algorithms and data processing techniques can be used. First, all edges in the regional ecological restoration feature graph are traversed, and their weights are compared with a preset threshold. Edges with weights below the threshold are removed from the graph. Then, graph search algorithms, such as depth-first search and breadth-first search, are used to identify core node connection paths with significant spatiotemporal correlations and retain them in the graph. Finally, a simplified regional ecological restoration feature graph is generated.

[0033] Step S300: Call the pre-trained restoration scheme optimization model to perform multi-dimensional adaptability analysis on the regional ecological restoration feature map, and obtain the ecological restoration scheme adaptability score matrix of the high-altitude and steep slope area. The ecological restoration scheme adaptability score matrix contains the adaptability quantification value of different restoration measures combinations in each sub-region of the region.

[0034] The pre-trained restoration scheme optimization model is an artificial intelligence model that has been trained to perform in-depth analysis and evaluation of regional ecological restoration feature maps. Multi-dimensional adaptability analysis involves analyzing the regional ecological restoration feature maps from multiple perspectives, considering the impact of various factors on the adaptability of ecological restoration schemes. These factors include vegetation restoration rate, soil organic matter content trends, climate adaptability index, and regional geographical features such as topographic slope, soil composition, and climate conditions. By comprehensively considering these factors, the adaptability of different restoration measure combinations in various sub-regions within the region can be fully assessed.

[0035] The ecological restoration scheme suitability scoring matrix is ​​a two-dimensional matrix, where rows represent different combinations of restoration measures and columns represent sub-regions within the main area. Each element in the matrix represents a quantitative value of the suitability of the corresponding combination of restoration measures within that sub-region. This value reflects the feasibility and effectiveness of implementing the combination of restoration measures in that sub-region. The higher the quantitative value of the suitability, the better the suitability of the combination of restoration measures within that sub-region, and the more likely it is to achieve good ecological restoration results.

[0036] As one implementation method, step S300 can be specifically implemented as the following steps S310~S350: Step S310: Input the regional ecological restoration feature map into the map embedding layer of the restoration scheme optimization model, and perform weighted aggregation processing on the node attribute set through the graph attention mechanism to generate a map embedding vector containing node importance weights. The node importance weights are positively correlated with the historical fluctuation range of vegetation restoration rate. The node attribute set includes vegetation restoration rate, soil organic matter content change trend and climate adaptability index.

[0037] The graph embedding layer of the restoration scheme optimization model is used to convert the regional ecological restoration feature map into a low-dimensional vector representation for subsequent processing and analysis. The graph attention mechanism can assign different attention weights to each node based on the relationships and attributes between nodes, thereby highlighting important nodes and relationships.

[0038] After inputting the regional ecological restoration feature map into the graph embedding layer, a weighted aggregation process is performed on the node attribute set based on a graph attention mechanism. Specifically, the attention weight of each node to other nodes is calculated based on the edge weights between nodes and the node attributes. The attention weight reflects the importance of one node to another; a larger weight indicates a greater influence of that node on another node. By weighted summing of the node attributes, the attribute information of each node is aggregated into a vector, generating a graph embedding vector containing the importance weights of the nodes. Nodes with greater historical fluctuations in vegetation restoration rate have higher importance weights in the graph embedding vector. This is because the historical fluctuations in vegetation restoration rate reflect the instability of vegetation during the ecological restoration process; greater fluctuations indicate that the restoration case unit corresponding to that node may face more challenges and uncertainties, requiring more attention and resource investment.

[0039] When calculating the graph attention mechanism, the Graph Attention Network (GAT) algorithm can be used. This algorithm calculates the attention weights between nodes by defining an attention function. Specifically, for each node, it linearly transforms the attribute information of its neighbors and then calculates the attention coefficients between that node and its neighbors using the attention function. After normalizing the attention coefficients, the attention weights are obtained. Finally, the attribute information of the neighbors is weighted and summed according to the attention weights to obtain the aggregated attribute information of that node. Through multiple iterations and updates, a graph embedding vector containing the importance weights of the nodes can be obtained.

[0040] Step S320: Calculate the cosine similarity between the graph embedding vector and the preset repair measure feature library through the feature interaction layer of the model optimized by the repair scheme, and generate an initial matching degree matrix between the repair measures and node attributes.

[0041] The feature library of remediation measures contains feature vectors of different vegetation type configuration patterns, soil improvement technologies, and water conservation measures. The feature vectors of vegetation type configuration patterns are constructed by statistical features of vegetation configuration ratios in historical cases. The feature vectors of soil improvement technologies are constructed by quantitative indicators of soil composition regulation effects. The feature vectors of water conservation measures are constructed by features of changes in water infiltration rate.

[0042] The feature interaction layer of the remediation scheme optimization model is a module used to analyze the relationship between the map embedding vector and the preset remediation measure feature library. By calculating the cosine similarity between the map embedding vector and each feature vector in the remediation measure feature library, the initial matching degree value between the remediation measure and the node attribute can be obtained. The preset remediation measure feature library is a collection containing feature vectors of various remediation measures, covering information on different vegetation type configuration patterns, soil improvement technologies, and water conservation measures. The feature vectors of vegetation type configuration patterns are constructed based on statistical features of vegetation configuration ratios in historical cases, such as the planting ratio of different vegetation species and the spatial distribution of vegetation. These statistical features reflect the characteristics and advantages of different vegetation type configuration patterns. The feature vectors of soil improvement technologies are constructed based on quantitative indicators of soil composition regulation effects, such as changes in soil pH and improvements in soil fertility. These quantitative indicators can accurately measure the effectiveness of soil improvement technologies. The feature vectors of water conservation measures are constructed based on changes in water infiltration rate, such as the magnitude of decrease in water infiltration rate and the extension of water retention time. These features reflect the effectiveness of water conservation measures.

[0043] The generated initial matching degree matrix between restoration measures and node attributes is a two-dimensional matrix, where rows represent different combinations of restoration measures and columns represent nodes in the regional ecological restoration feature map. Each element in the matrix represents the initial matching degree value between the corresponding restoration measure and node attributes, which reflects the applicability and matching degree of the restoration measure in the restoration case unit represented by the node.

[0044] When calculating cosine similarity, the feature vectors in the graph embedding vector and the feature library of the repair measures are first standardized to eliminate dimensional differences. Then, their dot product is calculated and divided by the product of their magnitudes to obtain the cosine similarity value. The cosine similarity value is used as the initial matching degree value to construct the initial matching degree matrix.

[0045] As one implementation method, step S320 can be specifically implemented as the following steps S321~S325: Step S321: Standardize the feature vector of each repair measure in the repair measure feature library so that the value of each dimension of the feature vector is within a preset numerical range.

[0046] Standardization can employ a maximum-minimum scaling method, with the scaling benchmark being the extreme value range of the corresponding features in the historical remediation case dataset. The remediation measure feature vectors include feature vectors for vegetation type configuration patterns, soil amendment techniques, and water conservation measures. Standardization unifies data from different dimensions to the same scale for subsequent calculations and analysis. In the remediation measure feature library, each remediation measure feature vector may have different value ranges and dimensions, which can affect the accuracy of cosine similarity calculations. Standardization eliminates these differences, ensuring that each dimension of the feature vector falls within a preset numerical range, such as [0,1]. Standardizing these feature vectors makes feature vectors from different types of remediation measures comparable, improving the accuracy of cosine similarity calculations.

[0047] Step S322: Extract the node attribute feature sub-vectors from the map embedding vector. The node attribute feature sub-vectors contain the feature values ​​of vegetation restoration rate, soil organic matter content change trend and climate adaptability index. The weight of each dimension is determined by the analytic hierarchy process. The weight of vegetation restoration rate is set according to its influence on the restoration effect. The weight of soil organic matter content change trend is set according to the degree of soil improvement demand. The weight of climate adaptability index is set according to the regional climate fluctuation range.

[0048] The node attribute feature sub-vector is extracted from the graph embedding vector and is used to describe the key feature vectors of node attributes. It contains feature values ​​of vegetation restoration rate, soil organic matter content change trend, and climate adaptability index. These feature values ​​reflect the important characteristics of the restoration case unit in the ecological restoration process.

[0049] The Analytic Hierarchy Process (AHP) is used to determine the weights of different factors. In this step, the AHP is used to determine the weight of each dimension in the feature sub-vectors of node attributes. Specifically, for vegetation restoration rate, its weight is set according to its influence on the restoration effect. If the vegetation restoration rate has a significant impact on the ecological restoration effect, such as rapidly increasing vegetation coverage and improving the quality of the ecological environment, it is assigned a higher weight; conversely, it is assigned a lower weight. For the trend of soil organic matter content change, its weight is set according to the degree of soil improvement needs. If the organic matter content in the soil is low, requiring a large amount of soil improvement work, it is assigned a higher weight; conversely, it is assigned a lower weight. For the climate adaptability index, its weight is set according to the regional climate fluctuation range. If the regional climate fluctuation range is large, vegetation needs better adaptability to survive and grow, it is assigned a higher weight; conversely, it is assigned a lower weight.

[0050] When determining weights, a hierarchical model needs to be constructed to organize the objectives, criteria, and solutions hierarchically. Then, the relative importance of the criteria is judged through pairwise comparisons, constructing a judgment matrix. The eigenvectors and eigenvalues ​​of the judgment matrix are calculated to obtain the weight vectors of the criteria. Finally, based on the weight vectors of the criteria, the weights of each dimension in the node attribute feature sub-vectors are determined.

[0051] Step S323: Calculate the cosine similarity value between the node attribute feature sub-vector and the standardized repair measure feature vector. The calculation range of the cosine similarity value is mapped to a preset numerical range through translation and scaling transformation as the initial matching degree value.

[0052] After obtaining the node attribute feature vectors and the standardized repair measure feature vectors, it is necessary to calculate their cosine similarity value. Specifically, the node attribute feature vectors and the standardized repair measure feature vectors are multiplied by a dot product, and then divided by the product of their magnitudes to obtain the cosine similarity value. Since the cosine similarity value ranges from -1 to 1, it needs to be mapped to a preset numerical interval, typically [0, 1], for ease of subsequent processing and analysis.

[0053] Translation and scaling transformations are used to map values ​​from one interval to another. Specifically, the minimum and maximum cosine similarity values ​​are first determined, then the minimum value is subtracted from the maximum value, and then divided by the difference between the minimum and maximum values ​​to obtain a value in the interval [0,1]. Finally, this value is used as the initial matching score.

[0054] This is done to standardize the range of initial matching scores, making the matching scores between different repair measures and node attributes comparable. Mapping the initial matching scores to the [0,1] range also facilitates subsequent sorting and filtering operations.

[0055] Step S324: Construct an initial matching degree matrix based on the initial matching degree values. The row dimension of the initial matching degree matrix corresponds to the number of the combination of restoration measures, the column dimension corresponds to the node number of the regional ecological restoration feature map, and the matrix element value is the initial matching degree value between the corresponding node and the restoration measures.

[0056] The initial matching degree matrix is ​​a two-dimensional matrix used to store the initial matching degree values ​​between different combinations of restoration measures and each node in the regional ecological restoration feature map. The row dimension of the matrix corresponds to the index of the restoration measure combination, with each row representing a different combination. The column dimension of the matrix corresponds to the node index in the regional ecological restoration feature map, with each column representing a different node. Each element in the matrix is ​​the initial matching degree value between the corresponding node and the restoration measure, reflecting the applicability and matching degree of the restoration measure within the restoration case unit represented by that node.

[0057] When constructing the initial matching degree matrix, the calculated initial matching degree values ​​need to be filled into the matrix sequentially according to the order of the remediation measure combination number and the node number. Ensure that the rows and columns of the matrix correspond one-to-one with the numbers of the remediation measure combination and the node.

[0058] Step S325: Perform row standardization on the initial matching degree matrix so that the sum of the elements in each row is a preset standard value. The processed matrix is ​​used to represent the relative compatibility ratio of the same repair measure combination at different nodes.

[0059] Row standardization is a further processing step of the initial matching degree matrix. Its purpose is to adjust the sum of each row's elements to a preset standard value, such as 1. This is to eliminate the total difference in initial matching degree values ​​between different combinations of remediation measures, ensuring that each row reflects the relative fit ratio of the same remediation measure combination at different nodes.

[0060] Specifically, for each row of the initial matching degree matrix, the sum of the elements in that row is calculated. Then, each element in that row is divided by the sum of the elements in that row to obtain the standardized element values. After row standardization, the sum of the elements in each row is 1, which means that each element represents the proportion of the fit of the remediation combination at the corresponding node to the total fit of the remediation combination. After row standardization, the processed matrix is ​​used to represent the relative fit proportion of the same remediation combination at different nodes.

[0061] Step S330: The spatiotemporal constraint module of the repair scheme optimization model performs constraint filtering on the initial matching degree matrix, filtering out repair measure combinations that do not conform to the maximum slope threshold in the terrain slope distribution record, and obtaining a candidate matching degree matrix that meets the basic constraints.

[0062] The spatiotemporal constraint module of the remediation scheme optimization model is used to screen and filter the initial matching degree matrix. It considers spatiotemporal constraints such as terrain slope distribution records to ensure the feasibility of the selected remediation measure combinations in practical applications. The terrain slope distribution records the slope magnitude at different locations within the region, and the maximum slope threshold is a pre-set slope value used to limit the applicability of remediation measure combinations. Some remediation measure combinations may not be feasible in areas with large slopes, or their implementation effect may be poor. Therefore, it is necessary to filter out remediation measure combinations that do not meet the maximum slope threshold.

[0063] This is to improve the feasibility and effectiveness of ecological restoration plans, avoid using inappropriate combinations of restoration measures in unsuitable areas, and thus improve the efficiency and quality of ecological restoration.

[0064] As one implementation method, step S330 can be specifically implemented as the following steps S331~S335: Step S331: Extract the slope value sequence of the terrain slope distribution record from the regional geographic feature data, and obtain the maximum slope value of each sub-region by filtering the maximum value of the sliding window. The size of the filtering window is proportional to the area of ​​the sub-region.

[0065] The slope value sequence recorded by the terrain slope distribution data records the slope magnitude at different locations within a region. Detailed terrain slope information can be obtained by extracting the slope value sequence from regional geographic feature data. Sliding window maximum value filtering is used to find the maximum value within each window in a continuous data sequence. In this step, the slope value sequence recorded by the terrain slope distribution data is used as input data, and a sliding window is set to find the maximum slope value within the window. The size of the filtering window is proportional to the area of ​​the sub-region to ensure that the maximum slope value can be reasonably calculated within sub-regions of different sizes. A larger filtering window is used for larger sub-regions to cover more slope information; a smaller filtering window is used for smaller sub-regions to improve calculation accuracy. Through sliding window maximum value filtering, the maximum slope value of each sub-region can be obtained, reflecting the maximum steepness of the terrain within each sub-region.

[0066] When performing maximum value filtering in a sliding window, various algorithms can be used, such as the monotonic queue algorithm. This algorithm maintains a monotonically decreasing queue to quickly find the maximum value within the window as it slides. Specifically, when the window slides to the right, the newly entered element is compared with the element at the end of the queue. If the new element is greater than the element at the end of the queue, the element at the end of the queue is popped from the queue until the new element is less than the element at the end of the queue or the queue is empty. Then, the new element is added to the queue. Simultaneously, if the element at the head of the queue is no longer in the window, it is popped from the queue.

[0067] Step S332: Obtain a preset set of terrain slope constraint thresholds, which includes the maximum allowable slope value corresponding to different types of restoration measures.

[0068] The preset set of terrain slope constraint thresholds is a pre-defined set of thresholds that sets a maximum allowable slope value for each type of remediation measure, based on its characteristics and requirements. Different types of remediation measures have different requirements for terrain slope. For example, some vegetation planting measures may be better implemented in areas with gentler slopes, while some engineering measures may be implemented within a certain slope range. By obtaining the preset set of terrain slope constraint thresholds, the limitations of each type of remediation measure regarding terrain slope can be clearly defined.

[0069] The set of terrain slope constraint thresholds can be determined through expert experience, historical data, and experimental studies. In practical applications, the set of terrain slope constraint thresholds needs to be reasonably adjusted according to the specific ecological restoration project and regional characteristics to ensure that the selected combination of restoration measures is feasible in terms of terrain slope.

[0070] Step S333: Compare the maximum slope value of the sub-region with the constraint threshold set element by element to generate a binary constraint matrix. A value of 1 in the matrix element indicates that the corresponding combination of repair measures meets the slope constraint in the sub-region, and a value of 0 indicates that it does not meet the constraint.

[0071] After obtaining the maximum slope value of the sub-region and the preset set of terrain slope constraint thresholds, they are compared element-by-element to determine whether each combination of remediation measures satisfies the slope constraints in each sub-region. Specifically, for the maximum slope value of each sub-region, it is compared with the maximum allowable slope value of the corresponding remediation measure type in the constraint threshold set. If the maximum slope value of the sub-region is less than or equal to the maximum allowable slope value of the remediation measure type, the remediation measure combination is considered to satisfy the slope constraints in the sub-region, and the corresponding matrix element value is 1; otherwise, it is considered not to satisfy the slope constraints, and the corresponding matrix element value is 0.

[0072] In this way, a binary constraint matrix is ​​generated. The row dimension of this matrix corresponds to the number of the remediation measure combination, and the column dimension corresponds to the number of the sub-region. Each element value in the matrix indicates whether the corresponding remediation measure combination satisfies the slope constraint in that sub-region. The binary constraint matrix provides the basis for subsequent filtering processing. By multiplying it element-wise with the initial matching degree matrix, remediation measure combinations that do not meet the slope constraint can be filtered out.

[0073] Step S334: Multiply the initial matching degree matrix and the binary constraint matrix element by element, set the matrix element values ​​that do not meet the constraints to 0, and obtain the matching degree matrix after preliminary filtering.

[0074] Element-wise multiplication of the initial matching degree matrix and the binary constraint matrix is ​​a matrix operation. By multiplying corresponding elements of the two matrices, a new matrix is ​​obtained. In this step, the initial matching degree matrix stores the initial matching degree values ​​between different repair measure combinations and each node, while the binary constraint matrix stores information on whether each repair measure combination satisfies the slope constraint in each sub-region. By performing element-wise multiplication, if an element in the binary constraint matrix is ​​0, it indicates that the corresponding repair measure combination does not satisfy the slope constraint in that sub-region, and the corresponding element in the initial matching degree matrix is ​​set to 0; if an element in the binary constraint matrix is ​​1, it indicates that the corresponding repair measure combination satisfies the slope constraint in that sub-region, and the corresponding element in the initial matching degree matrix is ​​retained. After element-wise multiplication, a pre-filtered matching degree matrix is ​​obtained. The matrix elements that do not satisfy the slope constraint have been set to 0, meaning that these repair measure combinations are no longer considered in the corresponding sub-regions, thus achieving preliminary filtering of the initial matching degree matrix.

[0075] Step S335: Check the non-zero elements in the column direction of the initially filtered matching degree matrix, delete all row vectors with all element values ​​of 0, and obtain a candidate matching degree matrix containing only combinations of repair measures that satisfy the basic constraints.

[0076] After obtaining the initial matching degree matrix, further screening and optimization are performed. This involves checking for non-zero elements in the column directions of the initially filtered matching degree matrix, specifically checking if any non-zero element exists in each row vector. If all elements in a row vector are 0, it indicates that the combination of remediation measures does not meet the slope constraints in any sub-region and has no practical application value; therefore, this row vector needs to be deleted.

[0077] After checking and deleting non-zero elements along the column direction, a candidate matching degree matrix is ​​obtained, containing only combinations of remediation measures that satisfy the basic constraints. This matrix retains only remediation measure combinations that satisfy the slope constraint in at least one sub-region. These remediation measure combinations are feasible in terms of terrain slope and can be used as candidate schemes for subsequent ecological restoration planning.

[0078] Step S340: Construct a multi-objective optimization function based on the candidate matching degree matrix, with the optimization objectives of minimizing the vegetation restoration cycle, maximizing the soil erosion control effect, and enhancing climate adaptability. Solve the Pareto optimal solution set of the multi-objective optimization function using the particle swarm optimization algorithm.

[0079] A multi-objective optimization function is a function that comprehensively considers multiple optimization objectives, such as minimizing the vegetation restoration cycle, maximizing soil erosion control, and enhancing climate adaptability. In ecological restoration, these three objectives are interrelated and mutually restrictive, requiring a balanced solution. By constructing a multi-objective optimization function, these objectives can be transformed into a mathematical expression for solution and optimization.

[0080] In particle swarm optimization (PSO), each particle represents a possible solution. It flies through the solution space, constantly adjusting its position and velocity to find the optimal solution. In this step, PSO is used to find the Pareto optimal solution set for a multi-objective optimization function. A Pareto optimal solution is a solution that achieves a balance among multiple objectives, where further optimization of a particular objective is impossible without compromising other objectives. By solving for the Pareto optimal solution set, a series of solutions that achieve a good balance among multiple objectives can be obtained, providing more possibilities for selecting ecological restoration solutions.

[0081] When constructing a multi-objective optimization function, it is necessary to assign appropriate weights to each optimization objective to reflect their importance in the overall optimization. Simultaneously, a specific functional expression for each optimization objective needs to be defined. For example, the vegetation restoration cycle can be represented by the time required for vegetation cover to reach a certain threshold, the soil erosion control effect can be represented by the degree of reduction in soil loss, and climate adaptability can be represented by vegetation growth indicators under different climatic conditions.

[0082] As one implementation method, step S340 can be specifically implemented as the following steps S341~S346: Step S341: Determine the weight configuration of each objective in the multi-objective optimization function using the analytic hierarchy process (AHP).

[0083] The judgment matrix element values ​​of the analytic hierarchy process (AHP) are obtained based on the statistical analysis of the actual contribution of each objective in historical restoration case data. The weight configuration of the vegetation restoration cycle is determined by the average convergence speed of the vegetation cover change curve. The weight configuration of the soil erosion control effect is determined by the increase in organic matter content in the soil composition distribution record. The weight configuration of climate adaptability is determined by the absolute value of the correlation coefficient between the climate condition change record and the vegetation cover change curve.

[0084] The Analytic Hierarchy Process (AHP) constructs a judgment matrix to compare and evaluate the relative importance of different objectives. The values ​​of the judgment matrix elements are obtained based on the statistical analysis of the actual contribution of each objective in historical restoration case data. This means that the actual impact of each objective on the ecological restoration effect in historical data is fully considered when determining the weights.

[0085] The weighting of the vegetation restoration cycle is determined by the average convergence rate of the vegetation cover change curve. The average convergence rate of the vegetation cover change curve reflects the growth rate and recovery capacity of vegetation during the ecological restoration process. A faster average convergence rate indicates a shorter vegetation restoration cycle and a greater contribution to the ecological restoration effect, thus assigning a higher weight; conversely, a slower convergence rate assigns a lower weight.

[0086] The weighting of soil erosion control effectiveness is determined by the increase in organic matter content in the soil composition distribution record. The increase in soil organic matter content reflects the effectiveness of soil erosion control measures. A larger increase in organic matter content indicates better control of soil erosion and a greater contribution to ecological restoration, thus assigning a higher weight; conversely, a smaller increase assigns a lower weight.

[0087] The weighting of climate adaptability is determined by the absolute value of the correlation coefficient between climate condition change records and vegetation cover change curves. The absolute value of the correlation coefficient reflects the growth response of vegetation under different climatic conditions. If the absolute value of the correlation coefficient is large, it indicates that the vegetation is more adaptable to climatic conditions and contributes more to the ecological restoration effect, and therefore is given a higher weight; conversely, it is given a lower weight.

[0088] When performing the analytic hierarchy process (AHP), the first step is to construct a judgment matrix, comparing each objective pairwise and assigning corresponding values ​​based on their relative importance. Then, by calculating the eigenvectors and eigenvalues ​​of the judgment matrix, the weight allocation for each objective is obtained. During the calculation process, it is crucial to ensure the consistency of the judgment matrix; that is, the element values ​​of the judgment matrix must satisfy certain logical relationships to guarantee the rationality and accuracy of the weight allocation.

[0089] Step S342: Construct a multi-objective optimization function containing the objective weight configuration. The expression of the multi-objective optimization function is constructed by the sum of the products of each objective function value and the corresponding weight configuration.

[0090] Among them, the objective function for vegetation restoration cycle adopts the reciprocal form to achieve the shortest optimization, the objective function for soil erosion control effect adopts the linear form to achieve the maximum optimization, and the objective function for climate adaptability adopts the exponential form to strengthen the optimization tendency of highly adaptable areas.

[0091] Specifically, after determining the weight configurations for each objective, a multi-objective optimization function containing these weight configurations needs to be constructed. The expression of the multi-objective optimization function is constructed by summing the products of each objective function value and its corresponding weight configuration. For the vegetation restoration cycle objective function, a reciprocal form is used to achieve the shortest possible value. This is because the shorter the vegetation restoration cycle, the larger its reciprocal; maximizing the reciprocal allows for minimizing the vegetation restoration cycle. For example, assuming the vegetation restoration cycle is T, the vegetation restoration cycle objective function can be expressed as f1 = 1 / T.

[0092] For the objective function of soil erosion control, a linear form is adopted to maximize optimization. The linear form directly reflects the magnitude of the soil erosion control effect; by maximizing this objective function, the soil erosion control effect can be maximized. For example, assuming that the soil erosion control effect can be represented by the degree of reduction in soil loss, and let the degree of reduction in soil loss be E, then the objective function of soil erosion control can be expressed as f2=E.

[0093] For the climate adaptability objective function, an exponential form is adopted to enhance the optimization tendency of highly adaptable areas. The exponential form can amplify the advantages of highly adaptable areas, making the optimization process more inclined to select areas and remediation measures with high climate adaptability. For example, assuming that climate adaptability can be represented by the correlation coefficient between climate condition change records and vegetation cover change curves, and let the correlation coefficient be C, then the climate adaptability objective function can be expressed as f3=e C Adding the products of each objective function value and its corresponding weight configuration, we obtain the expression for the multi-objective optimization function: F = w1f1 + w2f2 + w3f3, where w1, w2, and w3 are the weight configurations for vegetation restoration cycle, soil erosion control effect, and climate adaptability, respectively.

[0094] Step S343: Initialize the configuration items of the particle swarm optimization algorithm. The dimension of the particle position vector is consistent with the number of rows of the candidate matching degree matrix. The value range of each dimension is dynamically determined by the extreme value range of the corresponding column in the candidate matching degree matrix. The initial range of the particle velocity vector is set proportionally according to the value range of the particle position vector.

[0095] The particle position vector represents the particle's position in the solution space, and its dimension matches the number of rows in the candidate matching degree matrix. This is because each row of the candidate matching degree matrix represents a combination of repair measures, and each dimension of the particle position vector corresponds to the selection of a combination of repair measures. The range of values ​​for each dimension is dynamically determined by the extreme value range of the corresponding column in the candidate matching degree matrix. Each column in the candidate matching degree matrix represents a sub-region, and the element values ​​in the column represent the matching degree of different combinations of repair measures in that sub-region. By determining the minimum and maximum values ​​of each column, the range of values ​​for each dimension can be obtained. This ensures that the particle does not exceed a reasonable range when searching in the solution space.

[0096] The particle velocity vector represents the particle's movement speed in the solution space, and its initial range is set proportionally to the range of values ​​for the particle position vector. Generally, the initial range of the particle velocity vector is directly proportional to the range of values ​​for the particle position vector. By appropriately setting the initial range of the particle velocity vector, the particle can be made to search in the solution space at an appropriate speed, avoiding particles moving too fast or too slow, thus improving the algorithm's search efficiency.

[0097] When initializing the configuration options for the Particle Swarm Optimization (PSO) algorithm, parameters such as the number of particles and the maximum number of iterations need to be set. The number of particles determines the algorithm's search capability; a larger number of particles allows for a wider search range, but also increases the computational load. The maximum number of iterations determines the algorithm's termination condition; the algorithm stops searching when the maximum number of iterations is reached. By setting these parameters appropriately, the PSO algorithm can achieve better results in solving Pareto optimal solutions for multi-objective optimization functions.

[0098] Step S344: In each iteration, the diversity of the Pareto optimal solution set is maintained by sorting by crowding distance. The calculation of crowding distance takes into account the distribution density of particles in the target space, assigning a smaller selection probability to particles in dense regions and a larger selection probability to particles in sparse regions.

[0099] During the iterative process of particle swarm optimization (PSO), it is necessary to maintain the diversity of the Pareto optimal solution set to ensure that the searched solutions can cover the entire Pareto front of the multi-objective optimization function. Crowding distance is an indicator used to measure the distribution density of particles in the target space. By calculating the crowding distance, we can understand the distribution of particles in the target space.

[0100] To calculate the crowding distance, the particles in the Pareto optimal solution set are first sorted according to each objective function value. For each objective function, the difference in objective function values ​​between adjacent particles is calculated, and these differences are summed to obtain the crowding distance of that particle on that objective function. Then, the crowding distances of that particle on all objective functions are summed to obtain the total crowding distance of that particle.

[0101] In each iteration, the diversity of the Pareto optimal solution set is maintained by sorting by crowding distance. Specifically, particles in dense regions are assigned a lower selection probability, while particles in sparse regions are assigned a higher selection probability. This is because in dense regions, the particle distribution is relatively concentrated, and there may be a large number of similar solutions; selecting these particles may reduce the diversity of solutions. In sparse regions, the particle distribution is more dispersed, and selecting these particles can explore more different solutions, increasing the diversity of solutions. In this way, the particles in the Pareto optimal solution set can be continuously adjusted in each iteration to make them more evenly distributed in the target space, thereby maintaining the diversity of the Pareto optimal solution set.

[0102] Step S345: Adopt an adaptive crossover and mutation strategy to adjust the particle evolution process. The crossover probability is dynamically adjusted according to the non-dominated level of the particles. The higher the non-dominated level of the particles, the greater the crossover probability. The mutation probability decreases exponentially with the number of iterations.

[0103] The adaptive crossover and mutation strategy is used to adjust the particle evolution process. It dynamically adjusts the crossover and mutation probabilities based on the particle's characteristics and the number of iterations. The crossover operation involves exchanging some information between two particles to generate a new particle; the mutation operation involves randomly changing the value of a certain dimension of a particle to increase particle diversity.

[0104] The crossover probability is dynamically adjusted based on the particle's non-dominated level; particles with higher non-dominated levels have a greater crossover probability. Non-dominated level refers to a particle's dominance within the Pareto optimal solution set; a higher non-dominated level indicates better performance on multiple objectives. By assigning a larger crossover probability to particles with high non-dominated levels, these superior particles are more likely to generate new offspring, thereby accelerating the algorithm's convergence speed.

[0105] The mutation probability decreases exponentially with the number of iterations. In the early stages of the algorithm, the mutation probability is relatively high, which helps particles to conduct a broad search in the solution space and explore more possibilities. As the number of iterations increases, the mutation probability gradually decreases because in the later stages of the algorithm, it is necessary to focus more on the already found better solutions and avoid excessive mutation that could cause particles to jump out of the optimal solution region. By adopting an adaptive crossover and mutation strategy, the crossover and mutation probabilities can be dynamically adjusted according to the characteristics of the particles and the number of iterations during the particle evolution process. This allows the algorithm to achieve a better balance between search efficiency and search accuracy, improving its performance in solving the Pareto optimal solution set of multi-objective optimization functions.

[0106] Step S346: Obtain the preset dual judgment criteria as the iteration termination condition. When the change in the average objective function value of the Pareto optimal solution set within a consecutive preset number of algebras is lower than the preset threshold, and the global optimal position of the particle swarm has not been updated for a consecutive preset number of algebras, terminate the iteration and output the current Pareto optimal solution set.

[0107] The pre-defined dual criterion is used to determine the termination condition of the particle swarm optimization algorithm iteration. It comprehensively considers the change in the average objective function value of the Pareto optimal solution set and the update of the global optimal position of the particle swarm. The change in the average objective function value of the Pareto optimal solution set reflects the degree of optimization of the solution during the iteration process. If the change in the average objective function value is lower than the pre-defined threshold within a pre-defined number of consecutive iterations, it indicates that the algorithm has converged to a relatively stable state, and further iterations may not bring significant optimization effects.

[0108] The global optimal position of a particle swarm refers to the position of the particle with the best objective function value among all particles. If the global optimal position of the particle swarm remains unchanged for a preset number of algebras, it indicates that the algorithm has not found a better solution during this period, and it also suggests that the algorithm may have converged.

[0109] The iteration terminates and the current Pareto optimal solution set is output when both conditions are met simultaneously: the change in the average objective function value of the Pareto optimal solution set within a consecutive preset number of algebras is lower than a preset threshold, and the global optimal position of the particle swarm has not been updated within a consecutive preset number of algebras. This ensures that the algorithm stops promptly after achieving a certain optimization effect, avoiding unnecessary waste of computational resources. The preset number of algebras and the threshold need to be set reasonably based on the specific problem and experimental results. Generally, the preset number of algebras can be adjusted according to the complexity of the problem and the search capability of the algorithm, while the threshold can be set according to the accuracy requirements of the objective function. By reasonably setting dual judgment criteria, the particle swarm optimization algorithm can achieve better results when solving for the Pareto optimal solution set of multi-objective optimization functions.

[0110] Step S350: Map the Pareto optimal solution set to an ecological restoration scheme suitability score matrix. Each element value in the matrix represents the comprehensive suitability score of the corresponding restoration measure combination in each sub-region of the region. The comprehensive suitability score is obtained by weighted summation of the objective function of the optimal solution set.

[0111] The Pareto optimal solution set contains a series of solutions that achieve a good balance among multiple objectives, with each solution corresponding to a selection scheme of restoration measures. Mapping the Pareto optimal solution set to an ecological restoration scheme suitability score matrix allows for the representation of these solutions in matrix form. Each element in the matrix represents the comprehensive suitability score of the corresponding restoration measure combination within each sub-region of the region. The comprehensive suitability score is obtained by weighted summation of the objective functions of the optimal solution set. When constructing the multi-objective optimization function, corresponding weights have been assigned to each objective function. By multiplying each objective function value by its corresponding weight and summing the results, the comprehensive suitability score of the restoration measure combination in that sub-region is obtained. By mapping the Pareto optimal solution set to an ecological restoration scheme suitability score matrix, the comprehensive suitability of different restoration measure combinations in various sub-regions can be visually displayed.

[0112] Step S400: Based on the ecological restoration scheme adaptability scoring matrix and regional ecological restoration feature map, generate a spatial recommendation distribution map of ecological restoration schemes for high-altitude and steep slope areas. Each geographic grid unit in the spatial recommendation distribution map of ecological restoration schemes is labeled with the corresponding optimal combination of restoration measures and the implementation priority ranking.

[0113] The spatial distribution map of recommended ecological restoration solutions is a visualization tool that combines information about ecological restoration solutions with geospatial information to intuitively display suitable combinations of restoration measures and their implementation priorities for different regions. Generated based on an ecological restoration solution suitability scoring matrix and regional restoration feature maps, this distribution map fully considers the differences in geographical features and the suitability of restoration measures within a region, providing precise guidance for actual ecological restoration work.

[0114] As one implementation method, step S400 can be specifically implemented as the following steps S410~S460: Step S410: Extract the maximum row value from the ecological restoration scheme suitability score matrix to determine the combination of restoration measures corresponding to the highest suitability score for each sub-region, and generate a preliminary recommended scheme matrix. The preliminary recommended scheme matrix includes the sub-region number and the optimal restoration measure combination number.

[0115] The ecological restoration scheme suitability score matrix is ​​a two-dimensional matrix. Each row represents the suitability score of a combination of restoration measures in different sub-regions, and each column represents the suitability score of different combinations of restoration measures in a sub-region. The maximum value extraction process involves finding the element with the largest value in each row of the matrix. The restoration measure combination corresponding to this element is the combination with the highest suitability in that sub-region.

[0116] In practice, the maximum value is determined by traversing each row of the matrix and using a comparison algorithm. For example, starting with the first element of each row, it can be compared sequentially with subsequent elements. If a larger element is found, the maximum value and its corresponding column index are updated. The column index corresponds to the number of the combination of repair measures.

[0117] The preliminary recommended scheme matrix is ​​a matrix containing sub-region numbers and optimal restoration measure combination numbers. Sub-region numbers can typically be obtained directly from the node numbers of the regional ecological restoration feature map; these numbers uniquely identify each sub-region within the high-altitude, steep slope region. Recording the optimal restoration measure combination number corresponding to each sub-region forms the preliminary recommended scheme matrix. This matrix provides the foundational data for subsequent geospatial correlation and visualization.

[0118] As one implementation method, step S410 can be specifically implemented as the following steps S411~S415: Step S411: Traverse each column vector of the ecological restoration scheme suitability score matrix. Each column vector corresponds to the suitability score of all combinations of restoration measures for a sub-region.

[0119] In this step, the ecological restoration scheme suitability score matrix is ​​processed column-wise. Each column vector contains the suitability scores for different combinations of restoration measures within the same sub-region. By traversing each column vector, the suitability score data for each sub-region can be analyzed sequentially. During the traversal, a loop structure can be used, starting from the first column of the matrix and operating column by column. For each column vector, subsequent steps will sort and filter them to determine the optimal combination of restoration measures for that sub-region.

[0120] Step S412: Sort each column vector in descending order, extract the first element value after sorting as the highest fitness score, and record the row index corresponding to the first element as the optimal repair measure combination number. The row index and the repair measure combination number correspond one-to-one.

[0121] The descending sorting process arranges the elements in each column vector in descending order. Various sorting algorithms can be used, such as quicksort and bubble sort. After sorting, the first element of the vector is the highest fitness score among all remediation combinations for that sub-region. The row index corresponding to this highest fitness score is recorded. Since there is a one-to-one correspondence between row indices and remediation combination numbers, this row index represents the optimal remediation combination number for that sub-region. In this way, the most suitable remediation combination can be determined for each sub-region.

[0122] Step S413: If there are multiple identical highest fitness scores in the column vector, a random number generator is used to randomly select one from the corresponding row index as the optimal combination number of restoration measures. The random number seed is associated with the geographic coordinates of the sub-region, and the geographic coordinates are obtained from the node spatial coordinates of the regional ecological restoration feature map.

[0123] In some cases, a subregion may have multiple combinations of remediation measures with the same highest fit score. To select from these combinations, a random number generator is used. The random number generator randomly selects one from the row indices with the same highest fit score as the optimal remediation measure combination number.

[0124] To ensure the stability and repeatability of randomness, the random number seed is associated with the geographic coordinates of the sub-region. These geographic coordinates can be obtained from the spatial coordinates of nodes in the regional ecological restoration feature map, with each node in each sub-region having unique spatial coordinates. Thus, for the same sub-region, under the same conditions, the random selection result is consistent.

[0125] Step S414: Construct a preliminary recommended scheme matrix. The preliminary recommended scheme matrix includes sub-region numbers and optimal restoration measure combination numbers. The sub-region numbers correspond to the node numbers of the regional ecological restoration feature map.

[0126] Based on the optimal remediation combination numbers determined in the previous steps for each sub-region, we begin constructing a preliminary recommended scheme matrix. Each row of the matrix contains two elements: the first element is the sub-region number, and the second element is the optimal remediation combination number corresponding to that sub-region.

[0127] The sub-region numbers are directly taken from the node numbers of the regional ecological restoration feature map, ensuring the consistency between the sub-region information in the matrix and the geospatial information. Recording each sub-region and its corresponding optimal restoration measure combination number sequentially in the matrix completes the construction of the preliminary recommended scheme matrix.

[0128] Step S415: Perform a uniqueness check on the preliminary recommended scheme matrix. If duplicate numbers exist, they are distinguished by adding a number suffix, so that each sub-region corresponds to a unique combination number of remediation measures. The number suffix is ​​generated based on the geographical location characteristics of the sub-region.

[0129] After the initial recommended scheme matrix is ​​constructed, a uniqueness check is required. Due to possible random selection or other factors in the previous steps, multiple sub-regions may correspond to the same remediation measure combination number. To ensure that each sub-region has a unique remediation measure combination number, number differentiation is necessary.

[0130] If duplicate numbers are found, they are distinguished by adding a suffix. The suffix is ​​generated based on the geographical location characteristics of the sub-region, such as its north-south or east-west orientation on the map. This ensures that each sub-region corresponds to a unique combination of restoration measures, while also reflecting the geographical characteristics of the sub-region in the number.

[0131] Step S420: Associate and map the preliminary recommended scheme matrix with the node spatial coordinates of the regional ecological restoration feature map to obtain the geographic coordinate boundary of each sub-region and the corresponding recommended restoration measure combination. The node spatial coordinates are extracted from the node attribute table of the regional ecological restoration feature map.

[0132] The node attribute table of the regional ecological restoration feature map contains the spatial coordinate information of each node, which defines the geographical location of each sub-region. By associating the preliminary recommended scheme matrix with the node spatial coordinates, the recommended restoration measure combination for each sub-region can be linked to its geographical coordinate boundary.

[0133] In the association mapping process, the spatial coordinates of nodes are first extracted from the node attribute table of the regional ecological restoration feature map. Then, based on the correspondence between sub-region numbers and node numbers in the preliminary recommended scheme matrix, the recommended restoration measure combinations for each sub-region are matched with the corresponding node spatial coordinates. This yields the geographic coordinate boundaries of each sub-region and the corresponding recommended restoration measure combinations, providing crucial data for subsequent geospatial analysis and visualization.

[0134] As one implementation method, step S420 can be specifically implemented as the following steps S421~S425: Step S421: Extract the node attribute table from the regional ecological restoration feature map. The node attribute table includes the node number, spatial coordinate X value, spatial coordinate Y value, and sub-region boundary coordinate string field. The node number corresponds to the sub-region number in the preliminary recommended scheme matrix. The spatial coordinate X value and spatial coordinate Y value are the coordinates of the node's center point. The sub-region boundary coordinate string field consists of multiple latitude and longitude coordinate pairs.

[0135] The node attribute table of the regional ecological restoration feature map is an important data structure for storing node-related information. The node number uniquely identifies each node and corresponds to the sub-region number in the preliminary recommended scheme matrix, providing a foundation for subsequent association operations.

[0136] The X and Y coordinates represent the coordinates of the node's center point, which accurately pinpoint the location of each sub-region in geographic space. The sub-region boundary coordinate string field consists of multiple latitude and longitude coordinate pairs, which are connected sequentially to form the boundary of the sub-region, accurately depicting the geographic extent of each sub-region.

[0137] By extracting node attribute tables from regional ecological restoration feature maps, key information needed for geospatial analysis and visualization was obtained.

[0138] Step S422: Associate and match the sub-region numbers of the preliminary recommended scheme matrix with the node numbers of the node attribute table, and merge the recommended remedial measure combination numbers into the node attribute table through a left join operation to generate a spatial attribute table containing recommended measures. The recommended remedial measure combination numbers are extracted from the preliminary recommended scheme matrix.

[0139] A left join is a database operation used to associate two tables based on specified fields. In this step, the sub-region numbers of the preliminary recommendation matrix are matched with the node numbers of the node attribute table.

[0140] Specifically, using the node attribute table as the main table, the recommended remediation measure combination numbers in the preliminary recommended scheme matrix are merged into the node attribute table according to the correspondence between sub-region numbers and node numbers. In this way, the node attribute table contains the geographic coordinate information of each sub-region and the corresponding recommended remediation measure combination number, generating a spatial attribute table containing the recommended measures.

[0141] This spatial attribute table integrates geospatial information and remediation measures information, providing a unified data source for subsequent geospatial analysis and visualization.

[0142] Step S423: Parse the sub-region boundary coordinate string field in the spatial attribute table, and convert the coordinate pairs into geometric polygon objects supported by the GIS software. The spatial reference system of the geometric polygon objects is consistent with the regional geographic feature data.

[0143] The sub-region boundary coordinate string field in the spatial attribute table stores the boundary coordinate information of the sub-region. However, these coordinate pairs exist in the form of strings and need to be parsed and converted before they can be recognized and processed by GIS software.

[0144] During the parsing process, the sub-region boundary coordinate string field is first split according to a preset delimiter (such as commas, spaces, etc.) to obtain latitude and longitude coordinate pairs. Then, using tools or libraries provided by the GIS software, these coordinate pairs are converted into geometric polygon objects.

[0145] To ensure the accurate representation of geometric polygonal objects in geospace, their spatial reference system needs to be consistent with the regional geographic feature data. The spatial reference system defines the benchmark and projection method of geographic coordinates, ensuring consistency and comparability between different geographic data.

[0146] Step S424: Perform spatial topology repair on the generated geometric polygon objects to eliminate self-intersecting boundaries and overlapping areas, so that each polygon is a simple polygon and does not overlap with each other. The topology repair process is performed using a GIS topology inspection tool.

[0147] During the generation of geometric polygon objects, self-intersecting boundaries or overlapping areas may occur. Self-intersecting boundaries refer to the boundary lines of polygons intersecting each other, which can cause the geometry of the polygons to deviate from reality; overlapping areas refer to the overlapping parts between multiple polygons, which can affect the accuracy of geospatial analysis and visualization.

[0148] To eliminate these problems, spatial topology repair is needed for the generated geometric polygon objects. GIS topology inspection tools can be used, which can automatically detect and repair topological errors in polygons.

[0149] The specific operations of topology repair processing include identifying self-intersecting boundaries and overlapping regions, and then adjusting the boundary lines of polygons or segmenting overlapping parts to make each polygon a simple polygon (i.e., the boundaries do not self-intersect) and to ensure that the polygons do not overlap with each other. In this way, the resulting geometric polygon objects can accurately represent the geographical extent of each sub-region, providing a reliable data foundation for subsequent analysis and visualization.

[0150] Step S425: Construct a spatial database table. The spatial field of the spatial database table is the restored geometric polygon object. The attribute fields include node number, recommended restoration measure combination number, and suitability score. The node number corresponds to the node number in the node attribute table. The recommended restoration measure combination number is consistent with the optimal restoration measure combination number in the preliminary recommended scheme matrix. The suitability score is extracted from the ecological restoration scheme suitability score matrix.

[0151] Spatial database tables are important data structures used to store geospatial data and related attribute information. In this step, a spatial database table is constructed, storing the repaired geometric polygon objects as spatial fields within the table.

[0152] The attribute fields include node number, recommended restoration measure combination number, and suitability score. The node number corresponds to the node number in the node attribute table, ensuring the consistency of geospatial information; the recommended restoration measure combination number is consistent with the optimal restoration measure combination number in the preliminary recommended scheme matrix, clarifying the restoration measures corresponding to each sub-region; the suitability score is extracted from the ecological restoration scheme suitability score matrix, reflecting the degree of suitability of the restoration measure combination in the corresponding sub-region.

[0153] By constructing spatial database tables, geospatial information and remediation measures information were integrated, providing complete data support for subsequent GIS visualization operations.

[0154] Step S430: Use GIS spatial analysis tools to check the topological relationship of geographic coordinate boundaries, merge adjacent sub-regions with the same recommended measures, and generate a simplified spatial partition polygon set. The attribute fields of the spatial partition polygon set include the repair measure combination number and the suitability score.

[0155] GIS spatial analysis tools possess powerful geospatial data processing and analysis capabilities. In this step, these tools are used to perform topological relationship checks on geographic coordinate boundaries. Topological relationship checks identify adjacent sub-regions—those whose boundaries touch or are close to each other. Then, based on the recommended remediation measure combination number, adjacent sub-regions with the same recommended measures are filtered out. These sub-regions are then merged into a larger polygon, thereby reducing the number of polygons and simplifying the representation of geospatial data.

[0156] The attribute fields of the generated simplified spatial partition polygon set include the combination number of restoration measures and the suitability score. This attribute information can be directly inherited from the spatial database table, ensuring that the merged polygons still retain important restoration measures and suitability information. In this way, not only is the geospatial data simplified, but key ecological restoration information is also preserved, providing more concise and effective data for subsequent visualization and analysis.

[0157] As one implementation method, step S430 can be specifically implemented as the following steps S431~S435: Step S431: Call the buffer analysis function of the GIS spatial analysis tool to generate a buffer polygon based on the average width ratio of the sub-region for each geographic coordinate boundary polygon. The buffer polygon is used to eliminate small gaps. The average width of the sub-region is calculated by the spatial span of the geographic coordinate boundary.

[0158] Buffer analysis is a common method in GIS spatial analysis that creates buffers around geographic objects at a certain distance. In this step, the buffer analysis function of a GIS spatial analysis tool is used to generate buffer polygons for each geographic coordinate boundary polygon.

[0159] The width of the buffer is determined based on the average width ratio of the sub-regions. The average width of the sub-regions is obtained by calculating the spatial span of the geographic coordinate boundaries, that is, the maximum distance between the geographic coordinate boundaries in the horizontal and vertical directions. The width of the buffer is obtained by multiplying the average width of the sub-regions by a preset scaling factor.

[0160] The generated buffer polygons can eliminate tiny gaps between geographic coordinate boundary polygons. In real-world geographic data, due to measurement errors or data accuracy issues, there may be small gaps between adjacent polygons, which can affect subsequent analysis and visualization. By creating buffer polygons, these tiny gaps can be filled, making the geospatial data more continuous and complete.

[0161] Step S432: Perform a fusion operation on all buffer polygons, merging adjacent polygons with the same recommended repair measure combination number into a single polygon. The tolerance of the fusion operation is determined based on the coordinate precision, and the recommended repair measure combination number is extracted from the attribute field of the spatial database table.

[0162] The merging operation is the process of combining multiple adjacent polygons into a larger polygon. In this step, the merging operation is performed on all buffer polygons. First, the recommended remediation measure combination number is extracted from the attribute fields of the spatial database table, and polygons that are adjacent and have the same recommended measure are selected based on this number.

[0163] The tolerance for the fusion operation is determined based on the coordinate precision. Coordinate precision refers to the accuracy of the geographic coordinate data. The tolerance needs to be adjusted according to the coordinate precision to ensure the accuracy of the fusion operation. When the boundary distance between two polygons is within the tolerance range, they are considered adjacent and can be fused.

[0164] By merging polygons that are adjacent and have the same recommended remediation measure combination number, the representation of geospatial data is further simplified while retaining the remediation measure information for each polygon.

[0165] Step S433: Perform boundary smoothing on the merged polygons and use the Bézier curve fitting algorithm to optimize the boundary lines of the polygons, wherein the fitting error is controlled within the accuracy range of the original boundary coordinates.

[0166] The boundaries of the merged polygons may contain irregular jagged edges or abrupt changes. These rough boundaries can affect the visualization and analytical accuracy of geospatial data. To improve this, the boundaries of the merged polygons are smoothed.

[0167] Bézier curve fitting is a commonly used curve fitting method that generates smooth curves using control points. In this step, the Bézier curve fitting algorithm is used to optimize the boundary lines of the polygon. By adjusting the position and number of control points, the fitted curve can be made as close as possible to the original boundary lines.

[0168] To ensure the accuracy of the fit, the fitting error needs to be controlled within the accuracy range of the original boundary coordinates. This means that the deviation between the fitted boundary line and the original boundary line cannot exceed the accuracy of the coordinate data, thus ensuring that important geographic information is not lost during the smoothing process.

[0169] Step S434: Calculate the area and perimeter of the smoothed polygon, delete polygons with areas smaller than the preset minimum threshold, and merge them into the adjacent polygon with the largest area. The minimum threshold is determined according to the proportion of the total area of ​​the region. The total area of ​​the region is calculated by summing the areas of all sub-regions.

[0170] After smoothing the boundaries of the polygon, further filtering and optimization are needed. First, the area and perimeter of the smoothed polygon are calculated. Area and perimeter are important parameters describing the geometric features of a polygon; they help assess its size and shape.

[0171] Then, polygons are filtered according to a preset minimum threshold. The minimum threshold is determined based on the proportion of the total area of ​​the region, which is calculated by summing the areas of all sub-regions. If the area of ​​a polygon is less than the minimum threshold, it indicates that the polygon may be a small and unimportant area, and it is deleted to simplify the geospatial data.

[0172] After deleting polygons with areas smaller than the minimum threshold, they are merged into the adjacent polygon with the largest area. This ensures the continuity and integrity of geospatial data while reducing the number of polygons and improving the efficiency of data processing and analysis.

[0173] Step S435: Extract the vertex coordinate sequence of the processed polygon set and store it as a spatial partitioned polygon set containing polygon ID, recommended measure number, number of vertices and coordinate pair list. The polygon ID is a unique identifier, the recommended measure number is consistent with the recommended repair measure combination number, and the coordinate pair list is arranged in a clockwise direction to ensure the directionality of the polygon.

[0174] After the preceding series of processes, a simplified and optimized set of polygons is obtained. In this step, the vertex coordinate sequences of these polygons need to be extracted and stored as a canonical spatially partitioned polygon set.

[0175] The spatial partition polygon set contains multiple fields, among which the polygon ID is a unique identifier used to distinguish each polygon. The recommended measure number is consistent with the recommended remediation measure combination number, recording the remediation measure information corresponding to each polygon. The vertex count indicates the number of vertices of the polygon, and the coordinate pair list is arranged in a clockwise direction, storing the latitude and longitude coordinates of all vertices of the polygon.

[0176] Arranging the coordinate pairs in a clockwise direction ensures the orientation of the polygons, which is crucial for the analysis and visualization of geospatial data. Storing them as a standardized set of spatially partitioned polygons provides a clear and accurate data structure for subsequent symbolic rendering and annotation operations.

[0177] Step S440: The spatial partition polygon set is symbolically rendered. Different combinations of repair measures are represented by different color schemes. The brightness value of the color scheme is positively correlated with the adaptability score, and the width of the boundary line is proportional to the area of ​​the sub-region.

[0178] Symbolic rendering is a crucial step in presenting geospatial data in a visual manner. This step involves symbolic rendering of a collection of spatially partitioned polygons.

[0179] Different combinations of remediation measures are represented by different color schemes, making it easy to distinguish different remediation plans on the map. For example, different colors can be assigned to different combinations of remediation measures, such as green for a vegetation planting plan and blue for a water conservancy project remediation plan.

[0180] The brightness value of a color scheme is positively correlated with its suitability score. A higher suitability score corresponds to a higher brightness value for the corresponding color scheme. This allows for a visual representation of the suitability of remediation measures for each area on the map. For example, areas with higher suitability scores are brighter, indicating that the remediation measure is likely to be more effective in that area.

[0181] The width of the boundary line is proportional to the area of ​​the sub-region. Larger sub-regions have wider boundary lines, while smaller sub-regions have narrower boundary lines. This highlights the differences between regions of different sizes, making the map clearer and easier to understand.

[0182] As one implementation method, step S440 can be specifically implemented as the following steps S441~S445: Step S441: Construct a repair measure-color mapping table, mapping each repair measure combination number to a unique color scheme. The repair measure combination number is extracted from the attribute field of the spatial partition polygon set.

[0183] The repair measure-color map table is a key data structure for implementing symbolic rendering. In this step, the repair measure combination number is first extracted from the attribute fields of the spatial partition polygon set. Then, a unique color scheme is assigned to each repair measure combination number.

[0184] A color mapping table can be built using predefined color schemes, such as different color categories like reds, greens, and blues. Each color category can be further subdivided into different brightness and saturation levels to differentiate more combinations of remediation measures. By constructing a remediation measure-color mapping table, a clear correspondence is provided for subsequent color rendering.

[0185] Step S442: Extract the adaptability score value corresponding to each polygon from the adaptability score matrix of the ecological restoration scheme, normalize the score value to a preset numerical range, and use it as a color brightness adjustment parameter.

[0186] To ensure a positive correlation between color brightness values ​​and suitability scores, the suitability scores need to be processed. First, the suitability score for each polygon is extracted from the ecological restoration scheme suitability score matrix. Since different regions may have different ranges of suitability scores, these scores are normalized to a preset numerical range, typically [0,1]. Normalization can be achieved using linear normalization, which involves subtracting the minimum value from the score and then dividing by the difference between the maximum and minimum values. The normalized score can then be used as a color brightness adjustment parameter to adjust the color brightness of each polygon.

[0187] Step S443: Determine the basic color value of the polygon according to the repair measures - color mapping table, adjust the brightness component of the basic color value through the HSL color model, the brightness value is determined by the product relationship between the basic brightness and the normalized score value, the adjustment coefficient is dynamically set according to the color system type, and the basic brightness is preset according to the visual recognition of the color system.

[0188] After determining the remediation measures - color map and color brightness adjustment parameters, color rendering begins. First, the base color value for each polygon is determined according to the remediation measures - color map. The base color value is the initial color of the color scheme assigned based on the remediation measure combination number. Then, the brightness component of the base color value is adjusted using the HSL (Hue, Saturation, Brightness) color model. The brightness value is determined by multiplying the base brightness by the normalized score value. The base brightness is preset based on the visual recognizability of the color scheme; different color schemes may have different base brightness values. The adjustment coefficient is dynamically set according to the color scheme type; different color schemes may require different adjustment coefficients to achieve the optimal brightness effect. In this way, the color brightness of each polygon can be dynamically adjusted based on the adaptability score, allowing the map to intuitively reflect the degree of adaptability of remediation measures in different areas.

[0189] Step S444: Calculate the area value of each polygon, normalize the area value to a preset numerical range as the boundary line width. The normalization formula uses logarithmic transformation to compress area differences. The area value is consistent with the sub-region area field in the attribute table of the recommended distribution map of the ecological restoration scheme. The color of the boundary line is uniformly set to white, and the transparency is set to a fixed value.

[0190] To ensure that the width of the boundary line is proportional to the area of ​​the sub-region, the area values ​​of the polygons need to be processed. First, the area value of each polygon is calculated, which can be obtained from the sub-region area field in the attribute table of the spatial recommendation distribution map of the ecological restoration scheme.

[0191] Then, the area values ​​are normalized to a preset numerical range, typically [0,1]. To reduce area differences, a logarithmic transformation is used as the normalization formula. The logarithmic transformation can compress larger area values, making the differences in boundary line widths between polygons of different areas more reasonable.

[0192] The normalized area value is used as the boundary line width, and the boundary line color is uniformly set to white with a fixed transparency value. This makes the map boundary lines clearer and more uniform, improving the visualization effect.

[0193] Step S445: Input the processed color parameters and boundary line parameters into the GIS rendering engine, perform layered rendering of the spatial partition polygon set, and arrange the rendering order according to the polygon area from large to small to ensure that small polygons are visible. The rendering parameters of the GIS rendering engine are adjusted according to the display characteristics of the output device.

[0194] After processing the color and boundary parameters, these parameters are input into the GIS rendering engine for layered rendering. The GIS rendering engine is a software tool specifically designed for geospatial data visualization; it can render polygons based on the input parameters.

[0195] The rendering order is based on polygon area, from largest to smallest. This ensures that smaller polygons are not obscured by larger polygons during rendering, thus guaranteeing that all polygons are visible on the map. The rendering parameters of the GIS rendering engine need to be adjusted according to the display characteristics of the output device. Different output devices may have different resolutions, color modes, and display capabilities. By adjusting the rendering parameters, the map can achieve optimal display results on different devices.

[0196] Step S450: Overlay priority sorting labels onto the symbolized GIS map. The labels are located at the geometric center of the polygon. The label content includes the name of the combination of remediation measures and the recommended implementation order number. The smaller the number, the higher the priority.

[0197] After completing the symbolic rendering, to further provide guidance on the implementation of ecological restoration plans, priority ranking labels need to be overlaid on the GIS map. These labels can help users quickly understand the combination of restoration measures for each area and the order in which they are implemented.

[0198] The label's position is set to the geometric center of the polygon, ensuring it is centered within the polygon for easy viewing. The geometric center can be determined by calculating the weighted average of the polygon's vertex coordinates, with the weights typically allocated based on the edge lengths corresponding to the vertices.

[0199] The label includes the name of the remediation measure combination and the recommended implementation sequence number. The name of the remediation measure combination can be obtained by looking up the remediation measure combination number in a pre-defined measure name lookup table, which clearly shows the specific remediation measures used in each area. The smaller the recommended implementation sequence number, the higher the priority of the remediation measures in that area, and the more quickly they need to be remediated.

[0200] As one implementation method, step S450 can be specifically implemented as the following steps S451~S457: Step S451: Construct an implementation priority evaluation index system. The evaluation index system includes adaptability score, sub-region area, topographic complexity index, and soil erosion risk level. The adaptability score is extracted from the adaptability score matrix of the ecological restoration scheme. The sub-region area is obtained through spatial calculation of the geographic coordinate boundary. The topographic complexity index is obtained through contour line density analysis of the topographic zoning vector map. The soil erosion risk level is obtained through comprehensive analysis of clay content and slope value in the soil composition distribution record.

[0201] The implementation priority evaluation index system is an important basis for determining the priority of restoration implementation in each sub-region. This system includes multiple evaluation indicators, each reflecting the ecological restoration needs and feasibility of the sub-region from different perspectives.

[0202] The suitability score is extracted from the ecological restoration scheme suitability score matrix. It reflects the degree of suitability of each combination of restoration measures in the sub-region. The higher the suitability score, the better the restoration measures may be in the region.

[0203] The area of ​​a sub-region is calculated spatially based on its geographic coordinate boundaries. Larger sub-regions may require more resources and time for restoration, thus area is a factor influencing implementation priority. The terrain complexity index is obtained through contour line density analysis of the terrain zoning vector map. Contour line density reflects the degree of terrain undulation; denser contour lines indicate more complex terrain, potentially posing greater challenges to restoration efforts. Therefore, the terrain complexity index is also an important indicator for evaluating implementation priority. The soil erosion risk level is obtained through a comprehensive analysis of clay content and slope values ​​in the soil composition distribution record. Clay content and slope values ​​are crucial factors influencing soil erosion. By comprehensively considering these two factors, the soil erosion risk level of each sub-region can be assessed. A higher soil erosion risk level indicates a more fragile ecological environment in the area, requiring priority for restoration.

[0204] Step S452: Standardize each evaluation index. The suitability score is positively standardized, the sub-region area is positively standardized, the topographic complexity index is negatively standardized, and the soil erosion risk level is positively standardized. The standardization method is the maximum-minimum method to eliminate dimensional differences.

[0205] Since different evaluation indicators may have different dimensions and value ranges, they need to be standardized in order to make them comparable in the evaluation process.

[0206] Adaptability scores, sub-region area, and soil erosion risk levels are positively standardized, meaning that higher values ​​indicate a greater likelihood of prioritizing implementation. Positive standardization is achieved by subtracting the minimum value from the index value and then dividing by the difference between the maximum and minimum values.

[0207] The terrain complexity index uses negative standardization because higher terrain complexity corresponds to lower implementation priority. Negative standardization can be achieved by subtracting the index value from the maximum value and then dividing by the difference between the maximum and minimum values.

[0208] The standardization method uses the maximum-minimum method, which is simple and easy to implement and can effectively eliminate dimensional differences, allowing all evaluation indicators to be compared on a uniform scale.

[0209] Step S453: Calculate the objective weight of each evaluation index using the entropy weight method. The entropy weight is obtained by inversely deducing the information entropy value of each index. The lower the information entropy value, the greater the weight of the index. The information entropy value of the fit score is calculated by the degree of dispersion of its distribution in all sub-regions, and the information entropy value of the sub-region area is calculated by the probability density function of the area distribution.

[0210] Entropy weighting is an objective weighting method that determines the weight of each indicator based on its information entropy value. Information entropy is an indicator that measures the uncertainty of data; the lower the information entropy value, the more information the indicator provides, and the higher its weight should be assigned in the evaluation. For the fit score, its information entropy value is calculated by the dispersion of its distribution across all sub-regions. The greater the dispersion, the greater the difference in the fit score across different sub-regions, the more information the indicator provides, and the lower the information entropy value. For the sub-region area, its information entropy value is calculated using the probability density function of the area distribution. The probability density function describes the probability distribution of area values ​​across different intervals; by analyzing the probability density function, the information entropy value of the sub-region area can be calculated.

[0211] Calculating the objective weights of each evaluation indicator using the entropy weight method can avoid the influence of subjective factors and make the weight allocation more reasonable and accurate.

[0212] Step S454: Construct a weighted comprehensive evaluation model. The priority comprehensive score is obtained by multiplying the adaptability score by the entropy weight, the sub-region area by the entropy weight, the terrain complexity index after inverse transformation by the entropy weight, and the soil erosion risk level by the entropy weight. The terrain complexity index inverse transformation is achieved by reversing the value range.

[0213] After determining the objective weights of each evaluation indicator, a weighted comprehensive evaluation model is constructed to calculate the comprehensive score for implementation priority. This model weights and sums each evaluation indicator with its corresponding entropy weight to obtain the comprehensive score for implementation priority for each sub-region. Specifically, the comprehensive score for implementation priority is obtained by summing the products of the adaptability score and the entropy weight, the sub-region area and the entropy weight, the inversely transformed terrain complexity index and the entropy weight, and the soil erosion risk level and the entropy weight. The inverse transformation of the terrain complexity index is achieved by reversing its value range, because higher terrain complexity corresponds to lower implementation priority; the inverse transformation ensures that the index plays a correct role in the weighted summation.

[0214] The weighted comprehensive evaluation model can take into account the impact of multiple evaluation indicators and accurately assess the implementation priority of each sub-region.

[0215] Step S455: Sort the comprehensive scores of all sub-regions in descending order, and assign recommended implementation sequence numbers according to the sorting results. The sub-region with the highest comprehensive score is the starting number, and the numbers are increased sequentially. If the comprehensive scores are the same, they are sorted again by soil erosion risk level, and the sub-regions with higher risk levels are assigned smaller numbers.

[0216] After obtaining the comprehensive score for implementation priority of each sub-region, the comprehensive scores of all sub-regions are sorted in descending order. The sorting result reflects the implementation priority of each sub-region; the higher the comprehensive score, the more urgent the repair need of that sub-region, and the higher its implementation priority.

[0217] Based on the ranking results, a recommended implementation order number is assigned to each sub-region. The sub-region with the highest overall score is assigned the starting number, typically 1, and then increments sequentially. Thus, sub-regions with smaller numbers indicate higher implementation priority.

[0218] If multiple sub-regions have the same overall score, they are then ranked a second time based on their soil erosion risk level. Sub-regions with a higher soil erosion risk level indicate that their ecological environment is more fragile and requires priority for restoration; therefore, their numbers are assigned to lower levels.

[0219] This sorting and numbering method clearly determines the implementation priority order of each sub-region.

[0220] Step S456: Generate label content containing the recommended implementation sequence number. The name of the remediation measure combination in the label content is obtained by looking up the remediation measure combination number in the preset measure name lookup table.

[0221] After determining the recommended implementation sequence number for each sub-region, a tag containing that number is generated. In addition to the recommended implementation sequence number, the tag content also includes the name of the remediation measure combination. The name of the remediation measure combination is obtained by looking up the remediation measure combination number in a pre-defined measure name lookup table. This pre-defined lookup table stores the correspondence between remediation measure combination numbers and names; by consulting this table, the accurate name of the remediation measure combination can be added to the tag content for each sub-region. In this way, the generated tag content contains both implementation priority information and specific remediation measure information, providing detailed guidance for the implementation of the ecological restoration plan.

[0222] Step S457: Locate the label to the geometric center of the polygon using a GIS labeling tool. The geometric center is calculated by weighted average of the coordinates of the polygon vertices, with the weight being the length of the side corresponding to the vertex.

[0223] After generating the labels containing the recommended implementation sequence number and the name of the remediation measure combination, the labels need to be accurately located on the GIS map. This task can be accomplished using a GIS annotation tool.

[0224] The labels are positioned at the geometric center of the polygon, calculated using a weighted average of the polygon's vertex coordinates. The weight is the length of the side corresponding to the vertex; the longer the side, the greater the weight given to the vertex in the geometric center calculation. This method ensures the labels are centered within the polygon and takes into account its shape, resulting in more reasonable and accurate label placement. Thus, the implementation priority labels overlaid on the GIS map clearly display the restoration measures and implementation sequence information for each sub-area, providing intuitive guidance for ecological restoration work.

[0225] Step S460: Generate a spatial distribution map of recommended ecological restoration schemes containing spatial reference information. The spatial reference information adopts a coordinate system consistent with the regional geographic feature data. The attribute table of the spatial distribution map of recommended ecological restoration schemes includes polygon number, restoration measure combination number, suitability score, implementation priority, and sub-region area fields. The polygon number corresponds to the identifier of the polygon set of spatial partitions. The restoration measure combination number is consistent with the optimal restoration measure combination number in the preliminary recommended scheme matrix. The suitability score is extracted from the ecological restoration scheme suitability score matrix. The implementation priority is obtained by weighted calculation of the suitability score and the sub-region area. The sub-region area is obtained by spatial calculation of the geographic coordinate boundary.

[0226] After completing operations such as symbolic rendering and label overlay, a spatial distribution map of recommended ecological restoration schemes containing spatial reference information is finally generated. The spatial reference information uses a coordinate system consistent with the regional geographic feature data, which ensures the accuracy and consistency of the map in geographic space.

[0227] The attribute table of the spatial recommendation distribution map of ecological restoration schemes contains several important fields. The polygon number corresponds to the identifier of the polygon set for spatial partitioning, uniquely identifying each polygon. The restoration measure combination number is consistent with the optimal restoration measure combination number in the preliminary recommendation scheme matrix, clearly defining the restoration measure corresponding to each polygon. The suitability score is extracted from the ecological restoration scheme suitability score matrix, reflecting the degree of suitability of the restoration measures in the region.

[0228] The implementation priority is calculated by weighting the adaptability score with the sub-region area, comprehensively considering the impact of remediation effectiveness and regional scale on the implementation sequence. The sub-region area is obtained through spatial calculation of geographic coordinate boundaries and is an important parameter for assessing the workload and resource requirements of remediation.

[0229] By generating such a spatial distribution map of recommended ecological restoration solutions, key information such as geospatial information, restoration measures information, suitability scores, and implementation priorities are integrated, providing comprehensive and intuitive decision support for ecological restoration work in high-altitude and steep slope areas. Relevant personnel can use this map to clearly understand the best restoration measures and implementation sequence for each area, rationally allocate resources and time, and improve the efficiency and effectiveness of ecological restoration work.

[0230] Figure 2 A hardware entity diagram of a computer system provided as an embodiment of the present invention, such as... Figure 2 As shown, the hardware entity of the computer system 1000 includes a processor 1001 and a memory 1002, wherein the memory 1002 stores a computer program that can run on the processor 1001, and the processor 1001 executes the program to implement the steps in the method of any of the above embodiments.

[0231] The memory 1002 stores computer programs that can run on the processor. The memory 1002 is configured to store instructions and applications that can be executed by the processor 1001. It can also cache data to be processed or already processed (e.g., image data, audio data, voice communication data, and video communication data) of the processor 1001 and various modules in the computer system 1000. It can be implemented by flash memory or random access memory (RAM).

[0232] When processor 1001 executes a program, it implements the steps of the intelligent recommendation method for ecological restoration schemes on steep, cold slopes based on machine learning, as described above. Processor 1001 typically controls the overall operation of computer system 1000.

[0233] The above description is merely an embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A machine learning-based intelligent recommendation method for ecological restoration schemes on steep, cold slopes, characterized in that, The method includes: Collect historical restoration case data and regional geographical feature data for high-altitude, cold, and steep slope areas; The historical restoration case dataset and the regional geographic feature data are correlated and fused to generate a regional ecological restoration feature map with spatiotemporal correlation. Each node in the regional ecological restoration feature map corresponds to a restoration case unit, and the edge weight between nodes represents the matching degree value between the case unit and the regional geographic features. The pre-trained restoration scheme optimization model is invoked to perform multi-dimensional adaptability analysis on the regional ecological restoration feature map, resulting in an ecological restoration scheme adaptability score matrix for the high-altitude and steep slope region. The ecological restoration scheme adaptability score matrix contains quantitative values ​​of the adaptability of different restoration measures combinations in each sub-region of the region. Based on the ecological restoration scheme adaptability scoring matrix and the regional ecological restoration feature map, a spatial recommendation distribution map of ecological restoration schemes for high-altitude and steep slope areas is generated. Each geographic grid unit in the spatial recommendation distribution map of ecological restoration schemes is labeled with the corresponding optimal combination of restoration measures and the implementation priority ranking.

2. The method according to claim 1, characterized in that, The process of associating and fusing the historical restoration case dataset with the regional geographic feature data to generate a regional ecological restoration feature map with spatiotemporal correlation includes: Temporal interpolation processing is performed on the vegetation growth status records in the historical restoration case dataset to obtain a continuous time series vegetation cover change curve. The time interval of the vegetation cover change curve is consistent with the sampling period of the climate condition change records. Contour line density analysis is performed on the topographic slope distribution records in the regional geographic feature data to generate a topographic zoning vector map based on slope intervals. The boundary lines of the topographic zoning vector map are correlated with the sampling point distribution density of the soil composition distribution records. The vegetation cover change curve and the topographic zoning vector map are input into the spatiotemporal correlation module to calculate the spatial correlation coefficient between the vegetation cover change rate and the soil composition distribution record in each topographic zoning, and generate a spatiotemporal correlation matrix containing the correlation coefficient. Based on the spatiotemporal correlation matrix, a set of node attributes is constructed for the regional ecological restoration feature map. The set of node attributes includes vegetation restoration rate, soil organic matter content change trend, and climate adaptability index. The edge weights between nodes are obtained by normalizing the element values ​​of the spatiotemporal correlation matrix. The vegetation restoration rate is calculated by the slope of the vegetation cover change curve. The soil organic matter content change trend is extracted by the temporal fluctuation characteristics of soil component distribution records. The climate adaptability index is obtained by the correlation analysis between the vegetation cover change curve and the climate condition change records. The regional ecological restoration feature map is optimized by removing redundant connections with edge weights below a preset threshold and retaining the connection paths of core nodes with significant spatiotemporal correlations to generate a simplified regional ecological restoration feature map.

3. The method according to claim 1, characterized in that, The pre-trained restoration scheme optimization model is invoked to perform multi-dimensional adaptability analysis on the regional ecological restoration feature map, resulting in an ecological restoration scheme adaptability scoring matrix for the high-altitude, steep slope region, including: The regional ecological restoration feature map is input into the graph embedding layer of the restoration scheme optimization model. The node attribute set is weighted and aggregated through the graph attention mechanism to generate a graph embedding vector containing the importance weight of the nodes. The importance weight of the nodes is positively correlated with the historical fluctuation range of the vegetation restoration rate. The feature interaction layer of the optimization model of the repair scheme calculates the cosine similarity between the graph embedding vector and the preset repair measure feature library to generate an initial matching degree matrix between the repair measures and node attributes. The spatiotemporal constraint module of the repair scheme optimization model performs constraint filtering on the initial matching degree matrix, filtering out repair measure combinations that do not meet the maximum slope threshold in the terrain slope distribution record, and obtaining a candidate matching degree matrix that meets the basic constraints. Based on the candidate matching degree matrix, a multi-objective optimization function is constructed, with the optimization objectives being to minimize the vegetation restoration cycle, maximize the soil erosion control effect, and maximize the climate adaptability. The Pareto optimal solution set of the multi-objective optimization function is solved by the particle swarm optimization algorithm. The Pareto optimal solution set is mapped to an ecological restoration scheme suitability score matrix. Each element value in the matrix represents the comprehensive suitability score of the corresponding restoration measure combination in each sub-region of the region. The comprehensive suitability score is obtained by weighted summation of the objective function of the optimal solution set.

4. The method according to claim 3, characterized in that, The step of performing cosine similarity calculation between the graph embedding vector and the preset feature library of repair measures through the feature interaction layer of the repair scheme optimization model to generate an initial matching degree matrix between repair measures and node attributes includes: Each feature vector of a remediation measure in the remediation measure feature library is standardized so that the value of each dimension of the feature vector is within a preset numerical range. Extract the node attribute feature sub-vectors from the map embedding vector, wherein the node attribute feature sub-vectors include feature values ​​of vegetation recovery rate, soil organic matter content change trend and climate adaptability index; Calculate the cosine similarity value between the node attribute feature subvector and the standardized repair measure feature vector. The calculation range of the cosine similarity value is mapped to a preset numerical range through translation and scaling transformation as the initial matching degree value. An initial matching degree matrix is ​​constructed based on the initial matching degree value. The row dimension of the initial matching degree matrix corresponds to the number of the combination of restoration measures, the column dimension corresponds to the node number of the regional ecological restoration feature map, and the matrix element value is the initial matching degree value between the corresponding node and the restoration measure. The initial matching degree matrix is ​​subjected to row standardization so that the sum of the elements in each row is a preset standard value. The processed matrix is ​​used to represent the relative compatibility ratio of the same repair measure combination at different nodes.

5. The method according to claim 3, characterized in that, The spatiotemporal constraint module of the repair scheme optimization model performs constraint filtering on the initial matching degree matrix, filtering out repair measure combinations that do not meet the maximum slope threshold in the terrain slope distribution record, to obtain a candidate matching degree matrix that satisfies the basic constraints, including: The slope value sequence of the terrain slope distribution record is extracted from the geographic feature data of the region, and the maximum slope value of each sub-region is obtained by the maximum value filtering process of the sliding window. The size of the filtering window is proportional to the area of ​​the sub-region. Obtain a preset set of terrain slope constraint thresholds, wherein the set of terrain slope constraint thresholds includes the maximum allowable slope value corresponding to different types of restoration measures; The maximum slope value of the sub-region is compared element by element with the set of constraint thresholds to generate a binary constraint matrix. A value of 1 in the matrix element indicates that the corresponding combination of repair measures meets the slope constraint in the sub-region, and a value of 0 indicates that it does not meet the constraint. The initial matching degree matrix is ​​multiplied element by element with the binary constraint matrix, and the matrix elements that do not meet the constraints are set to 0 to obtain the matching degree matrix after preliminary filtering. The matching degree matrix after initial filtering is checked for non-zero elements in the column direction, and all row vectors with all element values ​​of 0 are deleted to obtain a candidate matching degree matrix containing only combinations of remediation measures that satisfy the basic constraints.

6. The method according to claim 1, characterized in that, The process of generating a spatial distribution map of recommended ecological restoration schemes for high-altitude, steep slope areas based on the ecological restoration scheme suitability scoring matrix and the regional ecological restoration feature map includes: The row maximum value is extracted from the ecological restoration scheme suitability score matrix to determine the combination of restoration measures corresponding to the highest suitability score of each sub-region, and a preliminary recommended scheme matrix is ​​generated. The preliminary recommended scheme matrix includes the sub-region number and the optimal restoration measure combination number. The preliminary recommended scheme matrix is ​​associated with the node spatial coordinates of the regional ecological restoration feature map to obtain the geographic coordinate boundary of each sub-region and the corresponding recommended restoration measure combination. The node spatial coordinates are extracted from the node attribute table of the regional ecological restoration feature map. The topological relationship of the geographic coordinate boundary is checked by calling the GIS spatial analysis tool, and adjacent sub-regions with the same recommended measures are merged to generate a simplified spatial partition polygon set. The attribute fields of the spatial partition polygon set include the remediation measure combination number and the suitability score. The set of polygons in the spatial partition is rendered using symbolic rendering, and different combinations of repair measures are represented by different color schemes. Priority sorting labels are overlaid on the symbolized and rendered GIS map. The labels are located at the geometric center of the polygon. The label content includes the name of the combination of remediation measures and the recommended implementation order number. The smaller the number, the higher the priority. Generate a spatial distribution map of recommended ecological restoration schemes containing spatial reference information, which uses a coordinate system consistent with the regional geographic feature data.

7. The method according to claim 6, characterized in that, The process of extracting the row maximum value from the ecological restoration scheme suitability score matrix determines the combination of restoration measures corresponding to the highest suitability score for each sub-region, generating a preliminary recommended scheme matrix, including: Traverse each column vector of the ecological restoration scheme suitability score matrix, where each column vector corresponds to the suitability score of all combinations of restoration measures for a sub-region; Each column vector is sorted in descending order, and the first element value after sorting is extracted as the highest fitness score. The row index corresponding to the first element is recorded as the optimal repair measure combination number. The row index and the repair measure combination number are in one-to-one correspondence. If there are multiple identical highest fit scores in the column vector, a random number generator will randomly select one from the corresponding row index as the optimal combination of restoration measures number. The random number seed is associated with the geographic coordinates of the sub-region, and the geographic coordinates are obtained from the node spatial coordinates of the regional ecological restoration feature map. A preliminary recommended scheme matrix is ​​constructed, which includes sub-region numbers and optimal restoration measure combination numbers, wherein the sub-region numbers correspond to the node numbers of the regional ecological restoration feature map; The initial recommended scheme matrix is ​​checked for uniqueness. If duplicate numbers exist, they are distinguished by adding a number suffix, so that each sub-region corresponds to a unique combination number of remediation measures. The number suffix is ​​generated based on the geographical location characteristics of the sub-region.

8. The method according to claim 6, characterized in that, The step of associating and mapping the preliminary recommended scheme matrix with the node spatial coordinates of the regional ecological restoration feature map to obtain the geographical coordinate boundaries of each sub-region and the corresponding recommended restoration measure combination includes: Node attribute tables are extracted from the regional ecological restoration feature map. The node attribute tables include node number, spatial coordinate X value, spatial coordinate Y value, and sub-region boundary coordinate string field. The node number corresponds to the sub-region number of the preliminary recommended scheme matrix. The spatial coordinate X value and spatial coordinate Y value are the center point coordinates of the node. The sub-region boundary coordinate string field is composed of multiple latitude and longitude coordinate pairs. The sub-region numbers of the preliminary recommended scheme matrix are associated and matched with the node numbers of the node attribute table. The recommended repair measure combination numbers are merged into the node attribute table through a left join operation to generate a spatial attribute table containing the recommended measures. Parse the sub-region boundary coordinate string field in the spatial attribute table, and convert the coordinate pairs into geometric polygon objects supported by GIS software. The spatial reference system of the geometric polygon objects is consistent with the regional geographic feature data. Perform spatial topology repair on the generated geometric polygon objects to eliminate self-intersecting boundaries and overlapping areas; A spatial database table is constructed. The spatial field of the spatial database table is the repaired geometric polygon object. The attribute fields include node number, recommended repair measure combination number and adaptability score. The node number corresponds to the node number in the node attribute table. The recommended repair measure combination number is consistent with the optimal repair measure combination number in the preliminary recommended scheme matrix.

9. The method according to claim 6, characterized in that, The step involves using GIS spatial analysis tools to perform topological relationship checks on the geographic coordinate boundaries, merging adjacent sub-regions with the same recommended measures, and generating a simplified set of spatial partition polygons, including: The buffer analysis function of the GIS spatial analysis tool is invoked to generate buffer polygons based on the average width ratio of the sub-region for each geographic coordinate boundary polygon; Perform a merging operation on all buffer polygons, merging adjacent polygons with the same recommended repair measure combination number into a single polygon; The merged polygons are subjected to boundary smoothing, and the boundary lines of the polygons are optimized using a Bézier curve fitting algorithm, wherein the fitting error is controlled within the accuracy range of the original boundary coordinates. Calculate the area and perimeter of the smoothed polygons, delete polygons with areas smaller than a preset minimum threshold, and merge them into the adjacent polygon with the largest area. The minimum threshold is determined based on the proportion of the total area of ​​the region, and the total area of ​​the region is calculated by summing the areas of all sub-regions. Extract the vertex coordinate sequence of the processed polygon set and store it as a spatially partitioned polygon set containing polygon ID, recommended measure number, number of vertices, and a list of coordinate pairs.

10. A computer system comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 9.

Images