Geospatial data management method and system based on artificial intelligence

Abstract

The invention provides a geographic spatial data management method and system based on artificial intelligence, and the method comprises the steps: firstly obtaining a geographic spatial data set of a target region, geographic entity observation records containing time sequence changes, and corresponding spatial position identifiers, and then carrying out the feature extraction of the geographic spatial data set, generating time sequence correlation features and spatial topological features, then calling a pre-trained geographic space analysis model to perform dynamic feature fusion processing on the time sequence correlation features and the spatial topological features, and generating a fused feature set; and generating a geographic space optimization strategy containing a geographic entity state prediction result and resource allocation priority configuration based on the fused feature set, and finally triggering a geographic information system updating operation of the target area according to the geographic space optimization strategy, thereby realizing dynamic management and timely updating of geographic space data. The accuracy, timeliness and effectiveness of geographic space data management are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a geospatial data management method and system based on artificial intelligence. Background Art

[0002] The management of geospatial data is crucial for numerous fields, including urban planning, resource allocation, environmental monitoring, and disaster warning. With the rapid development of information technology, the methods for acquiring geospatial data are becoming increasingly diverse, and the amount of data is exploding. However, existing geospatial data management methods have numerous limitations. On the one hand, traditional data management approaches often focus on the storage and querying of static geospatial data, making it difficult to effectively handle the temporal evolution of geographic entity observations and capture the patterns of their evolution over time, resulting in a lag in understanding geospatial state. On the other hand, when processing geospatial data, existing methods often view temporal and spatial features in isolation, lacking in-depth exploration and comprehensive analysis of the correlation between the two. This makes it difficult to generate comprehensive and accurate geospatial optimization strategies, and thus fails to provide a scientific and effective basis for operations such as resource allocation and geographic information system updates. This, in turn, impacts the effectiveness of geospatial data management and the quality of decision-making in various fields. Summary of the Invention

[0003] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a geospatial data management method based on artificial intelligence, the method comprising:

[0004] Acquire a geospatial data set of a target area, wherein the geospatial data set includes time-series-changing geographical entity observation records and corresponding spatial location identifiers;

[0005] Extracting features from the geographic spatial data set to generate temporal correlation features and spatial topological features of the geographic entity observation records;

[0006] Calling a pre-trained geospatial analysis model to perform dynamic feature fusion processing on the temporal correlation features and the spatial topology features to generate a fused feature set;

[0007] generating a geospatial optimization strategy based on the fused feature set, wherein the geospatial optimization strategy includes a geographic entity state prediction result and a resource allocation priority configuration;

[0008] Triggering a geographic information system update operation for the target area according to the geospatial optimization strategy. In another aspect, an embodiment of the present invention further provides an artificial intelligence-based geospatial data management system, comprising a processor and a machine-readable storage medium, the machine-readable storage medium being connected to the processor, the machine-readable storage medium being used to store programs, instructions, or codes, and the processor being used to execute the programs, instructions, or codes in the machine-readable storage medium to implement the above-mentioned method.

[0009] Based on the above aspects, by obtaining the geospatial data set of the target area, covering the temporal changes in the geographical entity observation records and the corresponding spatial location identifiers, the feature extraction of the geospatial data set is performed to generate temporal correlation features and spatial topological features, accurately characterizing the characteristics of the geographical entities from the two key dimensions of time and space, calling the pre-trained geospatial analysis model to perform dynamic feature fusion processing on the temporal correlation features and spatial topological features, fully considering the intrinsic connection between temporal and spatial features, making the fused feature set more comprehensive and representative, and generating a geospatial optimization strategy based on the fused feature set that includes the geographical entity status prediction results and resource allocation priority configuration. Finally, according to the geospatial optimization strategy, the geographic information system update operation of the target area is triggered, realizing the dynamic management and timely update of geospatial data, improving the accuracy, timeliness and effectiveness of geospatial data management, and helping to better serve practical application scenarios such as urban planning and resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a schematic diagram of the execution flow of the artificial intelligence-based geospatial data management method provided by an embodiment of the present invention.

[0011] Figure 2 Schematic diagram of the hardware architecture of the artificial intelligence-based geospatial data management system provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0012] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1 This is a flow chart of an artificial intelligence-based geospatial data management method provided by an embodiment of the present invention. The artificial intelligence-based geospatial data management method is introduced in detail below.

[0013] Step S110: Acquire a geographic spatial data set of a target area, wherein the geographic spatial data set includes time-series-changing geographic entity observation records and corresponding spatial location identifiers.

[0014] In this embodiment, the target area can be a new urban area to be developed, a large mining area, or a nature reserve that requires ecological monitoring. The sources of geographic spatial data sets are relatively wide, such as satellite remote sensing images, drone aerial survey data, data collected by ground surveying and mapping equipment, etc. For satellite remote sensing images, they can periodically shoot the target area to obtain geographic information at different time points. This information can reflect the temporal changes of geographic entities, such as changes in land use types, increases and decreases in vegetation coverage, etc. Ground surveying and mapping equipment, such as total stations, GPS receivers, etc., can accurately measure the spatial position of geographic entities and obtain corresponding spatial position identifiers, which are generally presented in the form of latitude and longitude coordinates or plane rectangular coordinates.

[0015] Assume that satellite remote sensing observes a target area at different times, such as t1, t2, and t3. The resulting geographic entity observation records are represented by a set R, where R = {r1, r2, r3, …}, where each ri contains the observed values ​​of various attributes of the geographic entity at a specific point in time, such as the reflectivity and elevation of the feature. The corresponding spatial location identifiers are represented by a set P, where P = {p1, p2, p3, …}, where pi is the specific location coordinate of the geographic entity in space. Thus, the geospatial data set D can be represented as D = (R, P), which combines the temporal variation information and spatial location information of the geographic entity.

[0016] Step S120: extracting features from the geographic spatial data set to generate temporal correlation features and spatial topological features of the geographic entity observation records.

[0017] After acquiring a geospatial data set, in order to deeply explore the useful information in the data, feature extraction operations are required to generate temporal correlation features and spatial topological features. Temporal correlation features can reveal the changing patterns and relationships of geographic entity observation records over time, while spatial topological features can reflect the spatial connectivity and distribution characteristics of geographic entities.

[0018] Step S121: Perform data integrity check on the geographic entity observation records in the geospatial data set, detect the timestamps and spatial location identifiers of missing data, generate supplementary data based on the interpolation algorithm of adjacent observation records, and form a preprocessed geographic entity data set with continuous spatiotemporal coverage.

[0019] During the geospatial data collection process, due to factors such as equipment failure and weather conditions, there may be data missing in the geographic entity observation records. In order to ensure the integrity and continuity of the data, a data integrity check operation is required.

[0020] First, each observation record in the geospatial data set is checked for each timestamp and spatial location identifier in the record to determine whether a corresponding observation exists. This can be done by traversing the set of observation records R and the set of spatial location identifiers P in the geospatial data set D. For each timestamp t and spatial location p, a corresponding observation r is checked. If no observation exists for a timestamp t and spatial location p, it is marked as missing data, and the timestamp and spatial location identifier of the missing data are recorded.

[0021] Suppose that in the time series T = {t1, t2, ..., tn}, it is found that the observation value of the spatial position p5 at the time point t3 is missing. Next, the supplementary data is generated based on the interpolation algorithm of adjacent observation records. Taking linear interpolation as an example, for the missing data point, find the adjacent time points t2 and t4 before and after it and the corresponding observation values ​​r2 and r4 of the spatial position p5. Let the missing data be r3. According to the linear interpolation formula, r3 = r2 + (r4-r2) * ((t3-t2) / (t4-t2)). In this way, all missing data are supplemented, and finally a pre-processed geographic entity data set D' with continuous spatiotemporal coverage is formed, so that the data is continuous in both time and space.

[0022] Step S122: Perform outlier detection processing on the preprocessed geographic entity data set, calculate the numerical distribution range of each geographic entity observation record, remove data points that exceed the preset distribution range, and perform sliding window mean filling processing on the vacant positions after removal to generate a standardized geographic entity data set.

[0023] After obtaining the preprocessed geographic entity data set, there may be some outliers, which may be caused by measurement errors, data transmission errors, etc. In order to ensure the quality of the data, outlier detection processing is required.

[0024] Calculate the numerical distribution range of each geographic entity observation record. For each attribute feature, its distribution range can be determined by statistical methods. For example, calculate the mean μ and standard deviation σ of each attribute feature, then set a preset multiple k and take the mean plus or minus k times the standard deviation as the normal distribution range of the data, that is, [μ-kσ, μ+kσ]. For each observation record r in the preprocessed geographic entity data set D', check whether the values ​​of its various attribute features are within the corresponding normal distribution range. If the value of an attribute feature exceeds this range, the data point is marked as an outlier.

[0025] Suppose that for a certain attribute feature A, its mean μA and standard deviation σA are calculated, with a preset multiplier k = 3. If the value of attribute feature A in observation r is not within the range [μA - 3σA, μA + 3σA], the observation is marked as an outlier. Removing these data points outside the preset distribution range will leave gaps in the data. To fill these gaps, a sliding window mean filling process is used.

[0026] Taking a sliding window of length m as an example, for each vacant position, take the m / 2 adjacent normal data points before and after it (if there are fewer than m / 2 data points before and after, take as many data points as possible) and calculate the mean of these data points as the filling value for the vacant position. Assuming the vacant position is rj, and its adjacent normal data points before and after it are rj-i, ..., rj-1, rj+1, ..., rj+i (i = m / 2), then the filling value is the mean of the attribute feature values ​​corresponding to these data points. In this way, all vacant positions are filled, and finally a standardized geographic entity data set D" is generated, making the data more standardized and reliable.

[0027] Step S123: Perform a time series segmentation operation on the standardized geographic entity data set, divide the geographic entity observation records into multiple time window subsequences according to a preset period length, each time window subsequence contains a fixed number of continuous observation records, and perform overlapping cutting processing on the observation records at the time window boundary to avoid data truncation.

[0028] In order to better analyze the temporal characteristics of geographic entity observation records, it is necessary to perform time series segmentation on the standardized geographic entity data set. The geographic entity observation records are divided into multiple time window subsequences according to the preset period length.

[0029] Step S1231: obtaining a timestamp sequence of all geographic entity observation records in the standardized geographic entity data set, sorting the observation records in the order of the timestamps, and generating an ordered time series data stream.

[0030] For the standardized geographic entity data set D', the timestamps of all observation records are first extracted to form a timestamp sequence T = {t1, t2, ..., tn}. Then, the observation records are sorted according to the order of the timestamps, so that the observation records are arranged in chronological order, forming an ordered time series data stream S. Common sorting algorithms such as quick sort and merge sort can be used for sorting. Through sorting, the chronological order of the geographic entity observation records can be clearly presented.

[0031] Step S1232: Determine the time window length and sliding step parameters according to a preset division rule, and divide the ordered time series data stream into a plurality of initial time window subsequences in a sliding window manner.

[0032] The preset division rules are determined according to the specific analysis requirements and the changing characteristics of the geographic entities. Assume that the time window length is w and the sliding step is s. The ordered time series data stream S is divided in a sliding window manner. Starting from the starting position of the data stream, continuous observation records of length w are selected each time as a time window subsequence. Then, the window is slid forward according to the sliding step s, and the next time window subsequence is selected until the entire ordered time series data stream is covered. For example, the first time window subsequence S1 contains observation records from the 1st to the wth, the second time window subsequence S2 contains observation records from the s+1th to the s+wth, and so on, generating multiple initial time window subsequences.

[0033] Step S1233: Detect the number of observation records in each initial time window subsequence. If the number is lower than a preset minimum record threshold, perform a window expansion operation to dynamically adjust the window length to include a sufficient number of observation records.

[0034] To ensure that each time window subsequence contains sufficient observations for effective analysis, it is necessary to check the number of observations in each initial time window subsequence. A minimum record threshold th is preset. For each initial time window subsequence Si, the number of observations ni is counted. If ni is less than th, the window is expanded.

[0035] Window expansion can be achieved by dynamically adjusting the window length. For example, the window length is increased by a fixed increment δ each time, and observations are reselected until the number of observations in the new time window subsequence reaches or exceeds the minimum record threshold th. Assuming that the number of observations n1 in the initial time window subsequence S1 is less than th, the window length is increased from w to w + δ, and the observations from the 1st to the w + δth are reselected as the new time window subsequence. The number of observations is rechecked until the required number of observations is met.

[0036] Step S1234: Time alignment is performed on the expanded time window subsequences, and a dynamic time warping algorithm is used to align the time offsets of observation records in different time windows to eliminate the timing misalignment caused by inconsistent data acquisition frequencies.

[0037] Since data collection frequencies may be inconsistent, observation records within different time windows may have time offsets, resulting in time sequence misalignment. To eliminate this effect, the dynamic time warping (DTW) algorithm is used to time-align the subsequences of the expanded time windows.

[0038] The DTW algorithm aligns two time series in time by finding the optimal matching path between them. For two different time window subsequences Sj and Sk, the distance matrix between them is first calculated. Each element in the distance matrix represents the distance between two observations. Then, using dynamic programming, the optimal path from the upper left corner to the lower right corner of the distance matrix is ​​found. The elements on this optimal path correspond to the optimal match between the two time series. Based on the optimal path, the observations in the time window subsequences are adjusted to align them in time. This eliminates timing misalignments caused by inconsistent data acquisition frequencies and ensures temporal consistency between the time window subsequences.

[0039] Step S1235: normalize the aligned time window subsequences, associate the normalized time window subsequences with the corresponding spatial location identifiers and store them, and establish a bidirectional mapping relationship between the time window index and the spatial region identifier.

[0040] To eliminate dimensional differences between different attribute features, the aligned time window subsequences are normalized. Common normalization methods can be used, such as min-max normalization or Z-score normalization. Taking min-max normalization as an example, for each attribute feature in each time window subsequence, its minimum value min and maximum value max are found. Each observation x of this attribute feature is normalized to the interval [0, 1]. The normalization formula is x' = (x-min) / (max-min).

[0041] Normalized time window subsequences are associated and stored with their corresponding spatial location identifiers. For each time window subsequence Si, its corresponding spatial location identifier pi is recorded. A bidirectional mapping relationship is also established between the time window index and the spatial region identifier. The time window index facilitates the search for a specific time window subsequence, while the spatial region identifier identifies the geographic location corresponding to the time window subsequence. This bidirectional mapping relationship enables rapid data query and analysis across both time and space dimensions.

[0042] Step S124: Perform trend decomposition processing on each time window subsequence, and use the seasonality and trend decomposition algorithm to separate the long-term trend component, seasonal cycle component and residual noise component of the geographic entity observation record, and extract the slope change characteristics of the long-term trend component and the amplitude fluctuation characteristics of the seasonal cycle component as time series correlation features.

[0043] After completing the division and processing of the time window subsequences, trend decomposition processing is performed on each time window subsequence to extract the long-term trend component, seasonal cycle component and residual noise component.

[0044] Use a seasonality and trend decomposition algorithm, such as the STL (Seasonal-Trend decomposition using Loess) algorithm. The STL algorithm decomposes the time series into a long-term trend component, a seasonal cycle component, and a residual noise component through local weighted regression. For each time window subsequence Si, it is input into the STL algorithm to obtain the long-term trend component Ti, the seasonal cycle component Si, and the residual noise component Ri, i.e., Si = Ti + Si + Ri.

[0045] Extract the slope variation characteristics of the long-term trend component. Slope variation reflects the rate of change of the geographic entity observation record over the long term. The slope variation characteristics can be obtained by calculating the slope of the long-term trend component at different time points. For example, for two adjacent time points ti and ti+1 in the long-term trend component Ti, calculate their slope ki = (Ti+1-Ti) / (ti+1-ti). These slope values ​​are used as the slope variation characteristics of the long-term trend component.

[0046] Extract the amplitude fluctuation characteristics of the seasonal cycle component. Amplitude fluctuation reflects the magnitude of change in the geographic entity observation record within the seasonal cycle. This amplitude fluctuation characteristic can be obtained by calculating the difference between the maximum and minimum values ​​of the seasonal cycle component within a seasonal cycle. For example, for the seasonal cycle component Si, find its maximum value maxi and minimum value mini within a seasonal cycle, and the amplitude fluctuation characteristic is ai = maxi - mini. The slope change characteristics of the long-term trend component and the amplitude fluctuation characteristics of the seasonal cycle component are used as time series correlation features. These characteristics can reflect the changing patterns and interrelationships of the geographic entity observation record over time.

[0047] Step S125: performing spatial clustering analysis based on the spatial location identifier, constructing a topological network including spatial adjacency relationships of geographic entities, and extracting spatial topological features of the topological network.

[0048] In order to analyze the spatial connection and distribution characteristics of geographic entities, spatial clustering analysis is performed based on spatial location identifiers, a topological network is constructed, and spatial topological features are extracted.

[0049] Step S1251: Parse the longitude and latitude coordinate information of the spatial location identifier of each geographic entity to construct a geographic entity coordinate set.

[0050] For each geographical entity observation record in the standardized geographical entity data set D”, parse the longitude and latitude coordinate information in its corresponding spatial location identifier. Assume that the spatial location identifier is represented by the set P = {p1, p2, …, pn}, and each pi contains the longitude and latitude coordinates (xi, yi). Extract these longitude and latitude coordinates to construct the geographical entity coordinate set C = {(x1, y1), (x2, y2), …, (xn, yn)}, which contains the specific location information of all geographical entities in space.

[0051] Step S1252: Calculate the spatial distance matrix between each pair of geographical entity coordinates, dynamically adjust the adjacency threshold based on the distribution density of geographical entity coordinates. If the spatial distance between two geographical entities is less than the dynamic adjacency threshold, establish a corresponding connection edge in the topological network.

[0052] According to the geographical entity coordinate set C, calculate the spatial distance between each pair of geographical entity coordinates. Common distance measurement methods such as Euclidean distance and Manhattan distance can be used. Assume that the Euclidean distance is adopted. For geographical entity coordinates (xi, yi) and (xj, yj), their spatial distance dij = √((xi - xj)^2 + (yi - yj)^2). Calculate the spatial distances between all pairs of geographical entities and construct the spatial distance matrix D = [dij], where i, j = 1, 2, …, n.

[0053] Dynamically adjust the adjacency threshold based on the distribution density of geographical entity coordinates. The distribution density of geographical entity coordinates can be measured by calculating the number of other geographical entities within a set range around each geographical entity. For areas with a higher distribution density, appropriately reduce the adjacency threshold; for areas with a lower distribution density, appropriately increase the adjacency threshold. Assume that the adjacency threshold is thd. For each element dij in the spatial distance matrix D, if dij < thd, establish a connection edge between geographical entities i and j in the topological network. In this way, construct a topological network containing the spatial adjacency relationships of geographical entities.

[0054] Step S1253: Perform a weight assignment operation on the connection edges. The weight value is inversely proportional to the spatial distance. The smaller the spatial distance, the larger the weight value of the connection edge.

[0055] After constructing the topological network, weights are assigned to the edges in the network. Since the closer the spatial distance between two geographic entities, the closer the connection, the weight value is inversely proportional to the spatial distance. For an edge (i, j) in the topological network, its corresponding spatial distance is dij, and the weight value is wij. The weight assignment formula can be wij = 1 / dij (to avoid the denominator being zero, a very small constant ε can be added to the denominator). In this way, each edge in the topological network is assigned a corresponding weight, reflecting the strength of the spatial connection between geographic entities.

[0056] Step S1254: traverse all geographic entity nodes, construct an undirected topological network containing weighted connection edges, and calculate the degree centrality index of each node to characterize its spatial connection density.

[0057] Traverse all geographic entity nodes in the geographic entity coordinate set C, combine the previously established connection edges and weight assignments, and construct an undirected topological network G = (V, E, W) containing weighted connection edges, where V is the node set, that is, the geographic entity set; E is the connection edge set; and W is the connection edge weight set.

[0058] Calculate the degree centrality index for each node. This index characterizes the spatial connectivity density of a node in a topological network. For node i, its degree centrality index Ci can be calculated by summing the weights of all connected edges: Ci = ∑wij, where j represents all nodes connected to node i. A larger degree centrality index indicates a more densely connected node in the topological network and a higher spatial connectivity density.

[0059] Step S1255: executing a community detection algorithm on the undirected topological network to identify geographic entity subgroups with high cohesion, and extracting the spatial coverage and internal connection strength index of each subgroup.

[0060] To further analyze the structure of the topological network, a community detection algorithm, such as the Louvain algorithm, is applied to the undirected topological network G. The Louvain algorithm iteratively optimizes modularity to partition the topological network into multiple subgroups of geographical entities with high cohesion. Nodes within each subgroup are densely connected, while connections between subgroups are relatively weak.

[0061] For each identified geographic entity subgroup, its spatial coverage and internal connectivity strength index are extracted. Spatial coverage can be determined by calculating the minimum bounding rectangle or convex hull of all geographic entity coordinates within the subgroup. Internal connectivity strength can be measured by calculating the sum of the weights of all connecting edges within the subgroup. The larger the sum of the edge weights, the stronger the internal connectivity within the subgroup.

[0062] Step S1256: Generate a spatial topology feature vector based on the degree centrality index and subgroup characteristics, wherein the feature vector includes node connection density, subgroup coverage radius, and cross-group connection strength.

[0063] In this embodiment, a spatial topological feature vector is generated based on the degree centrality index and subgroup characteristics obtained above. First, the node connection density is reflected by the degree centrality index of each node calculated above. The degree centrality index reflects the degree of connection between a node and other nodes in the topological network, that is, the spatial connection density of the node. The degree centrality indexes of all nodes are arranged in a set order to form a characteristic sequence of node connection density.

[0064] Regarding the subgroup coverage radius, after identifying the geographic entity subgroups, the spatial coverage of each subgroup has been calculated. The subgroup coverage radius can be obtained by calculating a characteristic distance of the subgroup's spatial coverage. For example, if the spatial coverage is determined by the minimum enclosing rectangle, half the length of the rectangle's diagonal can be used as the subgroup coverage radius. If the spatial coverage is determined by the convex hull, the distance to the farthest point on the convex hull boundary can be calculated based on the center of the convex hull as the subgroup coverage radius. Arrange the coverage radii of all subgroups in a certain subgroup numbering order to form a characteristic sequence of subgroup coverage radius.

[0065] Cross-group connection strength measures the closeness of connections between different subgroups. It can be calculated by summing the weights of the edges connecting nodes in different subgroups. For each pair of subgroups, the weights of the edges connecting them are summed. These cross-group connection strength values ​​are then organized into a feature sequence based on a certain order of subgroup pairs.

[0066] By concatenating the node connection density feature sequence, the subgroup coverage radius feature sequence, and the cross-group connection strength feature sequence, a spatial topology feature vector is generated. This spatial topology feature vector contains information in multiple dimensions, reflecting the spatial topology and connection characteristics of geographic entities from different aspects.

[0067] Step S126: Perform feature alignment processing on the temporal correlation features and the spatial topological features, and generate feature dimensions consistent with the number of spatial sub-regions by aggregating the temporal correlation features of all time window sub-sequences in the spatial sub-region, so that the mapping relationship between the time window sub-sequences and the spatial sub-regions is consistent, and the aligned features are combined into a multi-dimensional feature matrix.

[0068] After obtaining temporal correlation features and spatial topology features, they need to be aligned to facilitate subsequent analysis and processing. First, determine the spatial subregions. These subregions can be divided based on the geographic entity subgroups obtained from the previous spatial cluster analysis, with each subgroup corresponding to a spatial subregion.

[0069] For each spatial subregion, aggregate the temporal correlation features of all time window subsequences within that region. Assume that there are multiple time window subsequences within spatial subregion A, and each time window subsequence has corresponding temporal correlation features, such as the slope change feature of the long-term trend component and the amplitude fluctuation feature of the seasonal cycle component. Aggregate these temporal correlation features to obtain a comprehensive temporal correlation feature corresponding to the spatial subregion.

[0070] In this way, a comprehensive temporal correlation feature is generated for each spatial subregion, ensuring that the generated feature dimension matches the number of spatial subregions. This ensures a consistent mapping between the time window subsequences and spatial subregions, meaning that each spatial subregion has a corresponding comprehensive temporal correlation feature.

[0071] The aligned temporal correlation features and spatial topological features are combined into a multidimensional feature matrix. The comprehensive temporal correlation features and corresponding spatial topological feature vectors of each spatial subregion can be arranged in a set order to form the rows of the multidimensional feature matrix. For example, the first row can be the concatenation of the comprehensive temporal correlation features and spatial topological feature vectors of the first spatial subregion, the second row can be the concatenation of the corresponding features of the second spatial subregion, and so on. This multidimensional feature matrix integrates the temporal and spatial feature information of the geographic entity observation records.

[0072] Step S130: calling a pre-trained geospatial analysis model to perform dynamic feature fusion processing on the temporal correlation features and the spatial topology features to generate a fused feature set.

[0073] To fully explore the potential relationship between temporal correlation features and spatial topology features, a pre-trained geospatial analysis model is used to dynamically fuse these two features. The geospatial analysis model is trained on a large amount of data and can effectively learn the complex relationships between features.

[0074] Step S131: Input the temporal correlation features into the temporal feature encoder of the geospatial analysis model, extract local temporal patterns of different time scales through the temporal convolution layer, and use the multi-head attention mechanism to capture the global dependencies across time windows to generate a temporal encoding feature vector.

[0075] The temporal correlation features obtained above are input into the temporal feature encoder of the geospatial analysis model. The temporal feature encoder primarily consists of a temporal convolution layer and a multi-head attention mechanism. The temporal convolution layer uses convolution kernels of varying sizes to perform sliding convolutions on the temporal correlation features, extracting local temporal patterns at different time scales. For example, using a smaller convolution kernel can extract short-term temporal variation patterns, while using a larger convolution kernel can capture temporal trends over longer time scales.

[0076] The multi-head attention mechanism is used to capture global dependencies across time windows. It partitions temporally correlated features into multiple sub-features, each of which is processed by a different attention head. Each attention head calculates an attention score between sub-features, which reflects the degree of correlation between features at different time points. By taking a weighted sum of these attention scores, an attention representation for each sub-feature is obtained. Finally, the attention representations of all sub-features are concatenated to obtain a representation of global dependencies.

[0077] After processing by the temporal convolution layer and the multi-head attention mechanism, a temporal encoding feature vector is generated. This temporal encoding feature vector contains the local patterns of temporal correlation features at different time scales and the global dependency information across time windows.

[0078] Step S132: input the spatial topological features into the spatial feature encoder of the geospatial analysis model, aggregate the topological features of adjacent nodes through the graph convolution layer, and compress the feature dimensions using the spatial pooling layer to generate a spatial encoding feature vector.

[0079] The spatial topological features are input into the spatial feature encoder of the geospatial analysis model. The spatial feature encoder primarily consists of a graph convolution layer and a spatial pooling layer. The graph convolution layer aggregates the topological features of adjacent nodes based on the previously constructed topological network containing the spatial adjacency relationships of geographic entities. The features of each node are influenced by the features of its adjacent nodes. Through graph convolution, the feature information of adjacent nodes is transferred to the current node, updating the feature representation of the current node.

[0080] The spatial pooling layer is used to compress feature dimensions. Since spatial topological features can be high-dimensional, dimensionality reduction is necessary to reduce computational complexity and improve model efficiency. The spatial pooling layer can employ methods such as max pooling and average pooling to sample and compress features, preserving important feature information.

[0081] After processing by the graph convolution layer and the spatial pooling layer, a spatial encoding feature vector is generated. This spatial encoding feature vector integrates the spatial topological structure information of the geographic entity and is dimensionally compressed to facilitate subsequent feature fusion operations.

[0082] Step S133: perform feature alignment operation on the temporal coding feature vector and the spatial coding feature vector, adjust the time dimension of the temporal coding feature vector and the spatial dimension of the spatial coding feature vector to the same order of magnitude through spatial sub-region aggregation or time interpolation, and fill in the null values ​​of the feature missing positions.

[0083] Because the dimensions of the temporal encoding feature vector and the spatial encoding feature vector may be inconsistent, feature alignment is required. Spatial sub-region aggregation or temporal interpolation can be used to adjust the dimensions. If the temporal dimension of the temporal encoding feature vector differs significantly from the spatial dimension of the spatial encoding feature vector, spatial sub-region aggregation can be used to group the temporal encoding feature vector according to its spatial sub-region. The features within each group can then be aggregated, such as by averaging or summing, to bring the dimensions of the temporal encoding feature vector closer to those of the spatial encoding feature vector.

[0084] If the discrepancy between the time and space dimensions is due to uneven temporal sampling, temporal interpolation can be used. This interpolation method interpolates the time-series encoded feature vectors along the time dimension, bringing the temporal dimension and spatial dimension into the same order of magnitude. During feature alignment, there may be locations where features are missing. These missing locations can be filled using appropriate methods, such as the average or median of adjacent feature values.

[0085] Through the feature alignment operation, the temporal encoding feature vector and the spatial encoding feature vector are matched in dimension.

[0086] Step S134: construct a cross-modal attention mechanism, calculate the correlation matrix between the temporal encoding feature vector and the spatial encoding feature vector, and generate a dynamic weight coefficient based on the correlation matrix.

[0087] Step S1341: Map the temporal encoding feature vector and the spatial encoding feature vector to a high-dimensional latent space respectively to generate a temporal query vector and a spatial key vector.

[0088] To better calculate the correlation between the temporal encoding feature vector and the spatial encoding feature vector, they are first mapped to a high-dimensional latent space. A linear transformation is performed on the temporal encoding feature vector using a fully connected layer, mapping it to the high-dimensional latent space to obtain the temporal query vector. Similarly, a similar linear transformation is performed on the spatial encoding feature vector to obtain the spatial key vector. This mapping operation transforms the original feature vector into a space more suitable for calculating correlation, allowing features from different modalities to be compared and correlated in the same space.

[0089] Step S1342: Calculate the dot product similarity between the temporal query vector and the spatial key vector to generate an initial attention score matrix.

[0090] After obtaining the temporal query vector and spatial key vector, we calculate the dot product similarity between them. For each element in the temporal query vector and each element in the spatial key vector, we calculate the dot product between them and compose the initial attention score matrix with all the dot product results. The dot product similarity reflects the degree of similarity between the temporal and spatial features; a higher score indicates a closer association between the two features.

[0091] Step S1343: Normalize the initial attention score matrix, use the Softmax function to normalize the score into a probability distribution, and generate an attention weight matrix.

[0092] Because the scores in the initial attention score matrix can range widely, they need to be normalized to facilitate subsequent weighted calculations. The Softmax function is used to convert each score in the initial attention score matrix into a probability value, so that the sum of all elements in the matrix is ​​1. After processing with the Softmax function, the attention weight matrix is ​​obtained. Each element in the attention weight matrix represents the probability of association between temporal and spatial features, with larger weights indicating stronger associations.

[0093] Step S1344: Perform weighted aggregation on the spatial coding feature vector according to the attention weight matrix to generate a spatial attention feature vector.

[0094] Based on the attention weight matrix, a weighted aggregation operation is performed on the spatial encoding feature vector. Each element in the spatial encoding feature vector is multiplied by the corresponding weight value in the attention weight matrix. All weighted elements are then fused together to obtain the spatial attention feature vector. This spatial attention feature vector integrates the correlation information between the spatial encoding feature vector and the temporal encoding feature vector, highlighting spatial features that are strongly correlated with temporal features.

[0095] Step S1345: Concatenate the spatial attention feature vector and the temporal coding feature vector, and fuse the cross-modal information through a fully connected layer to generate a joint feature representation.

[0096] The spatial attention feature vector is concatenated with the temporal encoding feature vector, forming a new vector by connecting the two vectors in a predetermined order. This new vector is then fed into a fully connected layer. The fully connected layer processes the concatenated vector through a series of linear transformations and nonlinear activation functions, fusing cross-modal information. This processing generates a joint feature representation that integrates feature information from both temporal and spatial modalities.

[0097] Step S1346: perform layer normalization on the joint feature representation to eliminate scale differences between features, and use residual connections to retain the original feature information to generate the final dynamic weight coefficient.

[0098] Layer normalization is performed on the joint feature representation. This standardizes each element in the joint feature representation, eliminating scale differences between features and allowing features to be compared and processed at the same scale. Residual connections are also used to preserve the information of the original temporal and spatial encoding feature vectors. Residual connections add the original feature vector to the processed feature vector, preventing the loss of important original information during the feature fusion process. After layer normalization and residual connections, the final dynamic weight coefficients are generated. These coefficients reflect the importance of the temporal and spatial encoding feature vectors in the feature fusion process and are dynamically adjusted based on the input data.

[0099] Step S135: performing weighted fusion on the temporal coding feature vector and the spatial coding feature vector according to the dynamic weight coefficient to generate a preliminary fusion feature matrix.

[0100] Based on the previously generated dynamic weight coefficients, a weighted fusion operation is performed on the temporal coding feature vector and the spatial coding feature vector. Each element in the temporal coding feature vector is multiplied by the corresponding dynamic weight coefficient; similarly, each element in the spatial coding feature vector is multiplied by the corresponding dynamic weight coefficient. The weighted temporal coding feature vector and spatial coding feature vector are then concatenated to form a preliminary fused feature matrix. This preliminary fused feature matrix integrates temporal and spatial feature information and highlights the importance of different features based on the dynamic weight coefficients.

[0101] Step S136: performing a feature enhancement operation on the preliminary fused feature matrix, enhancing the interaction between features through a nonlinear transformation layer, and using a denoising autoencoder to remove redundant noise, thereby generating an optimized fused feature set.

[0102] To further improve the quality of the preliminary fused feature matrix, feature enhancement is performed on it. First, the preliminary fused feature matrix is ​​processed through a nonlinear transformation layer. This nonlinear transformation layer can use activation functions such as ReLU and Sigmoid to perform a nonlinear transformation on each element in the feature matrix. This nonlinear transformation can enhance the interactions between features and explore potential relationships between them.

[0103] Next, a denoising autoencoder is used to remove redundant noise from the initial fused feature matrix. A denoising autoencoder is an unsupervised learning model that removes noise by adding noise to the input data and then learning to reconstruct the original data. The initial fused feature matrix is ​​used as input to the denoising autoencoder. After processing by the encoder and decoder, a de-noised feature matrix is ​​generated. After processing by the nonlinear transformation layer and the denoising autoencoder, an optimized fused feature set is generated, which contains richer and more valuable geospatial feature information.

[0104] The training process of the above geospatial analysis model can be: using a large amount of geospatial data sets, the temporal correlation features and spatial topological features obtained after feature extraction as training data, and dividing the training data into a training set, a validation set and a test set.

[0105] Initialize the convolution kernel weights of the temporal convolution layer, the parameters of the multi-head attention mechanism, the weights of the graph convolution layer, and the relevant parameters of the spatial pooling layer. Input the training data into the geospatial analysis model, pass it through the temporal feature encoder and spatial feature encoder in sequence, and obtain the temporal encoding feature vector and the spatial encoding feature vector. Then, perform feature alignment and cross-modal attention fusion to output a preliminary fused feature matrix. During this process, define an appropriate loss function, such as the mean squared error loss function, and calculate the loss between the preliminary fused feature matrix and the true label. Based on the loss value, use an optimization algorithm (such as the stochastic gradient descent algorithm) to update the parameters of the geospatial analysis model so that the loss value gradually decreases. Evaluate the model performance on the validation set, and adjust the parameters and structure of the geospatial analysis model based on the verification results, such as adjusting the convolution kernel size and the number of attention heads. Use the test set to conduct a final test on the trained geospatial analysis model to evaluate its generalization ability.

[0106] Step S140: generating a geospatial optimization strategy based on the fused feature set, wherein the geospatial optimization strategy includes a geographic entity state prediction result and a resource allocation priority configuration.

[0107] After obtaining the fused feature set, these feature information are used to generate a geospatial optimization strategy, which aims to predict the status of geographic entities and reasonably configure resource allocation priorities.

[0108] Step S141: inputting the fused feature set into the geographic entity state prediction model, calculating the probability of state change of the geographic entity in the future time window through the multi-layer perceptron network, and generating a probability distribution prediction result.

[0109] The fused feature set is input into the geographic entity state prediction model, which primarily consists of a multilayer perceptron network. This is a feedforward neural network consisting of an input layer, hidden layers, and an output layer. The fused feature set is fed into the input layer and, after a series of nonlinear transformations and weighted summation operations in the hidden layers, ultimately yields the probability of the geographic entity's state changing in the future time window at the output layer.

[0110] Neurons in the hidden layer perform nonlinear transformations on the input signal using activation functions, increasing the model's expressive power. Each neuron in the output layer corresponds to a geographic entity's state category, and its output value represents the probability of the entity being in that state category. The state change probabilities of all geographic entities are arranged in a predefined order to form a probability distribution prediction. This probability distribution prediction reflects the possible state changes of the geographic entity within a future time window.

[0111] During the training process of the geographic entity state prediction model, the geographic entity state prediction model is mainly composed of a multi-layer perceptron network, including an input layer, multiple hidden layers and an output layer. The input layer receives the fused feature set, the hidden layer transforms and combines the input features through a nonlinear activation function (such as a ReLU function), and the output layer outputs the probability distribution of the state change of the geographic entity in the future time window. The layers are connected in a fully connected manner, that is, each neuron is connected to all neurons in the previous layer. In detail, the fused feature set can be used as input data, and the corresponding true state of the geographic entity can be used as label data, which is also divided into a training set, a validation set and a test set. After initializing the weights and biases of each layer of the multilayer perceptron network, the training data is input into the model and passes through the input layer, hidden layer, and output layer in sequence to obtain the predicted state change probability distribution. Then, the cross-entropy loss function is used to calculate the loss value between the predicted probability distribution and the true label. The optimization algorithm (such as the Adam optimization algorithm) is used to update the parameters of the geographic entity state prediction model according to the loss value. The performance of the geographic entity state prediction model is evaluated on the validation set, and the parameters such as the number of neurons in the hidden layer and the learning rate are adjusted. The final test is performed on the test set to evaluate the prediction accuracy of the geographic entity state prediction model.

[0112] Step S142: identifying risky geographic entities according to the probability distribution prediction result, calculating resource demand urgency scores of the risky geographic entities, and generating a priority list by sorting based on the resource demand urgency scores.

[0113] Based on the probability distribution prediction results, geographical entities that may be at risk are identified. For example, if the probability of a geographical entity being in a dangerous state exceeds a preset threshold, it is marked as a risky geographical entity. For each identified risky geographical entity, a resource demand urgency score is calculated. This resource demand urgency score can take into account multiple factors, such as the current state of the geographical entity, the predicted trend of state change, and the impact on the surrounding environment.

[0114] Each factor can be assigned a weight, and the scores for each factor are then multiplied by the corresponding weight and summed to produce a resource demand urgency score. For example, a geographic entity's current state score can be determined based on the degree of deviation between its current attribute value and its normal state value; the predicted state change trend score can be determined based on the probability of state change for that geographic entity as predicted by the probability distribution. After calculating the resource demand urgency scores for all risky geographic entities, they are sorted from high to low by score to generate a priority list. The geographic entities in the priority list are arranged according to the urgency of their resource needs, with higher-scoring geographic entities receiving higher priority in resource allocation.

[0115] Step S143: Construct a multi-objective optimization model with maximizing resource allocation efficiency and minimizing geographic entity protection costs as objective functions, and introduce constraints to limit the total amount of resources, allocation time window, and resource allocation order set based on the priority list.

[0116] To rationally allocate resources, a multi-objective optimization model was constructed with two objective functions: maximizing resource allocation efficiency and minimizing the cost of protecting geographic entities. Resource allocation efficiency can be measured by effective resource utilization, such as whether resources can be allocated to the required geographic entities in a timely and accurate manner. The cost of protecting geographic entities can include resource procurement, transportation, and maintenance costs.

[0117] Constraints are introduced to limit the scope of the model's solution. First, the total amount of resources is limited to ensure that allocated resources do not exceed the upper limit of available resources. Second, a time window is set to ensure that resource allocations must be completed within the specified timeframe. Finally, the resource allocation order is set based on the priority list generated earlier, ensuring that high-priority geographic entities receive resources first.

[0118] Step S144: using an evolutionary algorithm to solve the multi-objective optimization model, generating a Pareto optimal solution set, and selecting the optimal solution that meets the preset strategy through an interactive decision interface.

[0119] Step S1441: setting population initialization parameters, crossover probability parameters and mutation probability parameters of the evolutionary algorithm.

[0120] Before using an evolutionary algorithm to solve a multi-objective optimization model, several key parameters must be set. Population initialization parameters determine the size of the initial population and the encoding method for the individuals. The initial population size determines the coverage of the algorithm's search space. A larger population increases the likelihood of finding the optimal solution, but also increases the computational effort. The encoding method for the individuals represents the resource allocation solution in a form that the algorithm can process. Examples include binary encoding, real number encoding, and so on.

[0121] The crossover probability parameter controls the probability of crossover between individuals during the evolutionary process. A crossover exchange involves exchanging parts of the codes of two individuals to create a new individual, thereby increasing the diversity of the population. The mutation probability parameter controls the probability of an individual mutating during the evolutionary process. Mutation randomly modifies the codes of an individual to prevent the algorithm from falling into a local optimum.

[0122] Step S1442: Generate an initial population based on the resource allocation priority configuration and the constraints of the multi-objective optimization model, where each individual represents a resource allocation plan, and use a constraint satisfaction algorithm to verify the legitimacy of each individual, eliminate individuals that violate the total resource constraints, allocation time window constraints, and resource allocation sequence constraints, and generate new individuals that meet the constraints.

[0123] An initial population is generated based on the resource allocation priority configuration and the constraints of the multi-objective optimization model. Each individual represents a resource allocation plan that allocates resources to different geographic entities. A constraint satisfaction algorithm is used to verify the validity of each individual in the initial population. Each individual is checked for violations of the total resource constraints, allocation time window constraints, and resource allocation order constraints.

[0124] If an individual violates any of the constraints, it is removed from the population. At the same time, to maintain the size of the population, new individuals that meet the constraints are generated. This can be done randomly or by mutating existing legal individuals.

[0125] Step S1443: performing fitness evaluation on each individual in the initial population, and calculating a multi-objective fitness value based on the objective function of maximizing the resource allocation efficiency and the objective function of minimizing the geographic entity protection cost.

[0126] The fitness of each individual in the initial population is evaluated, and a multi-objective fitness value is calculated based on the objective functions of maximizing resource allocation efficiency and minimizing geographic entity protection costs. The fitness value reflects the performance of each individual in the multi-objective optimization problem. For each resource allocation solution represented by an individual, its resource allocation efficiency and geographic entity protection cost are calculated. Resource allocation efficiency can be measured by calculating the effective benefits generated by resources allocated to each geographic entity, such as the degree of improvement in the geographic entity's condition. The geographic entity protection cost comprehensively considers the costs of resource acquisition, transportation, and maintenance.

[0127] When calculating multi-objective fitness, resource allocation efficiency and geographic entity protection costs need to be comprehensively considered. A weighted summation approach can be used, assigning weights to each, then multiplying the scores by their corresponding weights and adding them together to obtain the multi-objective fitness value for that individual. Weightings are determined based on the specific application scenario and requirements. If resource allocation efficiency is more important, a higher weight can be assigned to resource allocation efficiency; if geographic entity protection costs are more important, a higher weight can be assigned to geographic entity protection costs.

[0128] Step S1444: performing genetic operations on the initial population, including a selection operation based on the fitness value, a crossover operation based on the crossover probability parameter, and a mutation operation based on the mutation probability parameter, to generate a progeny population.

[0129] Selection is based on the fitness of individuals. Individuals with higher fitness values ​​have a greater probability of being selected for the next generation. Roulette wheel selection and tournament selection methods can be used for selection. Roulette wheel selection uses each individual's fitness value as the area of ​​the sector it occupies on the wheel. Higher fitness values ​​correspond to larger sectors. Individuals are selected based on the sector indicated by the pointer by randomly rotating the wheel. Tournament selection randomly selects a set number of individuals from the population, compares their fitness values, and selects the individual with the highest fitness value to advance to the next generation.

[0130] The crossover operation is performed based on a crossover probability parameter. Based on a pre-set crossover probability, two individuals are randomly selected from the individuals obtained by the selection operation as parents. A crossover operation is then performed on these two parent individuals. Crossover can be performed using either single-point or multi-point crossover methods. A single-point crossover randomly selects a crossover point in the coding sequence of an individual and swaps the sections of the two parent individuals after the crossover point, generating two offspring individuals. A multi-point crossover selects multiple crossover points and swaps the sections between them.

[0131] Mutation operations are performed based on a mutation probability parameter. For each offspring individual, positions in its coding sequence are randomly selected for mutation based on the mutation probability. Mutation operations can change a value in the coding sequence, such as changing a 0 to a 1 or a 1 to a 0 in binary code, or making small adjustments to the value in real code. Mutation operations can increase population diversity and prevent the algorithm from falling into local optimal solutions.

[0132] After selection, crossover and mutation operations, the offspring population is generated. The individuals in the offspring population are generated by genetic operations on the individuals in the initial population. They inherit some characteristics of the parent individuals and may also introduce new characteristics.

[0133] Step S1445: Merge the initial population with the offspring population, perform non-dominated sorting to determine the Pareto front rank of the individuals, and calculate the crowding distance to screen diverse individuals to generate a new generation population.

[0134] The initial population and the offspring population are merged to form a larger population. A non-dominated sort is performed on the merged population. Non-dominated sorting is a sorting method for multi-objective optimization problems that classifies individuals in a population into different Pareto frontier levels. An individual is said to dominate another individual if it is not inferior to the other individual on all objective functions and is superior to the other individual on at least one objective function. Non-dominated individuals are continuously identified and assigned to the first Pareto frontier level. Non-dominated individuals are then identified from the remaining individuals and assigned to the second Pareto frontier level, and so on, until all individuals are assigned to different Pareto frontier levels.

[0135] Calculate the crowding distance for each individual. The crowding distance measures the degree of crowding within the Pareto front tier. For each Pareto front tier, sort the individuals by the value of each objective function. Then, calculate the sum of the distances between each individual's boundary individuals and their neighbors on each objective function as the crowding distance for that individual. A larger crowding distance indicates a less crowded individual within its Pareto front tier and a higher degree of diversity.

[0136] Individuals are selected based on their Pareto front rank and crowding distance to generate the next generation of the population. Individuals with lower Pareto front rank are prioritized because they perform better in multi-objective optimization problems. Within the same Pareto front rank, individuals with higher crowding distance are prioritized to ensure population diversity. In this way, a set number of individuals are selected from the merged population to form the next generation of the population.

[0137] Step S1446: Iterate the above steps until the evolution termination condition is reached and output the Pareto optimal solution set.

[0138] The population is continuously updated by iteratively executing steps such as selection, crossover, mutation, merging populations, non-dominated sorting, and individual screening. Evolutionary termination criteria can include reaching a preset maximum number of iterations or no significant improvement in the population's fitness. When these termination criteria are met, iteration ceases and the set of Pareto-optimal solutions in the current population is output. Each solution in the Pareto-optimal set represents a resource allocation scheme that strikes a balance between resource allocation efficiency and geographic entity protection costs. No single solution is superior to another in terms of all objectives.

[0139] Step S1447: Visualize the Pareto optimal solution set in the interactive decision interface, receive the selection instruction input by the user, and determine the final optimal solution based on the preset strategy matching degree.

[0140] Visualize the Pareto-optimal solution set in an interactive decision-making interface. Using two-dimensional or three-dimensional graphs, each solution's value for resource allocation efficiency and geographic entity protection costs is plotted as a point, forming a Pareto front. Users can intuitively observe the relationships and advantages and disadvantages of different solutions.

[0141] Receive user input for selection. Users can select one or more solutions from the Pareto optimal solution set based on their needs and preferences. For example, a user may be more concerned with resource allocation efficiency and wish to select a solution with higher resource allocation efficiency; or a user may be more concerned with the cost of protecting geographic entities and wish to select a solution with lower cost.

[0142] The final optimal solution is determined based on the degree of match with the preset strategy. Preset strategies can be pre-defined rules or preferences based on actual application scenarios and requirements. For example, a preset strategy might specify the weighting between resource allocation efficiency and geographic entity protection costs, or set thresholds for certain objectives. Based on the user's input selection instructions and the preset strategy, the match between each candidate solution and the preset strategy is calculated, and the solution with the highest match is selected as the final optimal solution.

[0143] Step S145: Integrate the resource allocation plan corresponding to the optimal solution with the geographic entity state prediction result to generate a geographic space optimization strategy including specific execution steps and a timetable.

[0144] The resource allocation plan corresponding to the final optimal solution is integrated with the geographic entity status prediction results. The resource allocation plan specifies which geographic entities will be allocated resources and the allocated amounts, while the geographic entity status prediction results provide information on how the geographic entities will change over the future time window. Combining these two pieces of information, specific implementation steps and a timeline are developed.

[0145] Specific implementation steps include operations such as resource acquisition, transportation, and allocation, as well as specific measures for different geographic entities. For example, if a geographic entity is predicted to be at risk, the implementation steps may include monitoring that entity and implementing protective measures. A timetable specifies the start and end times for each implementation step, ensuring the orderly allocation of resources and the protection of geographic entities.

[0146] By integrating resource allocation plans and geographic entity status prediction results, a geospatial optimization strategy with specific implementation steps and a timeline is generated. This geospatial optimization strategy provides detailed guidance for geospatial management and optimization.

[0147] Step S146: Verify the feasibility of the geospatial optimization strategy and simulate the changes in the state of geographic entities after the execution of the geospatial optimization strategy. If the simulation result exceeds the preset risk threshold, dynamically adjust the objective function weight or constraint relaxation range of the multi-objective optimization model according to the risk type and regenerate the geospatial optimization strategy.

[0148] Verify the feasibility of the generated geospatial optimization strategy. This can be done through simulation, based on the current state of the geographic entities, resource allocation plan, and execution steps. Simulate the changes in the state of the geographic entities after the geospatial optimization strategy is implemented. Consider the impact of various factors, such as the effects of resources and changes in environmental factors.

[0149] Set a preset risk threshold to measure the risk level of a geographic entity's state change. This threshold can be customized to suit different geographic entities and application scenarios. For example, a lower risk threshold can be set for certain important geographic entities. If simulation results exceed the preset risk threshold, the current geospatial optimization strategy may be risky and requires adjustment.

[0150] Dynamically adjust the objective function weights or constraint relaxation ranges of the multi-objective optimization model based on the risk type. If the risk type is low resource allocation efficiency, the weight of the resource allocation efficiency objective function can be appropriately increased, or the total resource constraint can be relaxed. If the risk type is high geographic entity protection costs, the weight of the geographic entity protection cost objective function can be appropriately increased, or the resource allocation time window constraint can be tightened.

[0151] Regenerate the geospatial optimization strategy. Based on the adjusted multi-objective optimization model, use the evolutionary algorithm again to obtain a new set of Pareto optimal solutions. Then, follow the previous steps to determine the final optimal solution, integrate the resource allocation plan and the geographic entity status prediction results, and generate a new geospatial optimization strategy. Through continuous feasibility verification and adjustment, ensure the effectiveness and reliability of the geospatial optimization strategy.

[0152] Step S150: triggering a geographic information system update operation of the target area according to the geospatial optimization strategy.

[0153] After the geospatial optimization strategy is generated, it is necessary to trigger the GIS update operation of the target area according to the geospatial optimization strategy to ensure that the data and information in the GIS can reflect the latest geospatial status and resource allocation.

[0154] Step S151: Analyze the specific execution steps and timetable in the geographic space optimization strategy, and extract the resource allocation solution parameters that match the geographic entity state prediction result and the spatial location identifier of the target geographic entity.

[0155] The specific execution steps and timelines within the geospatial optimization strategy are analyzed. The specific execution steps detail the resource allocation and processing process, while the timeline specifies the timing of each step. From this information, resource allocation plan parameters that match the predicted geographic entity status are extracted, including the amount of resources to be allocated and the target of the allocation. Furthermore, the spatial location identifiers of the target geographic entities are extracted; these identifiers are used to determine the specific location of the geographic entity within the geographic space.

[0156] For example, if the geospatial optimization strategy stipulates that a certain amount of resources should be allocated to a certain geographic entity at a certain point in time, the parsing process needs to extract parameters such as the spatial location identifier of the geographic entity and the amount of allocated resources.

[0157] Step S152: Generate an incremental update instruction for the geographic information system according to the resource allocation plan parameters. The incremental update instruction includes a spatial coordinate sequence of the resource allocation path, a timestamp constraint condition, and a corresponding geographic entity attribute modification value.

[0158] Generate incremental update instructions for the GIS based on the extracted resource allocation plan parameters. Incremental update instructions are specific instructions for updating data in the GIS, which contain the spatial coordinate sequence of the resource allocation path, timestamp constraints, and corresponding geographic entity attribute modification values.

[0159] The spatial coordinate sequence of a resource allocation path describes the path from the acquisition point to the allocation point, expressed as a series of spatial coordinates. Timestamp constraints specify the timeframe for resource allocation operations, ensuring that resource allocation is completed within the specified timeframe. The corresponding geographic entity attribute modification value refers to the changes to the geographic entity's attributes after resource allocation, such as the modified values ​​of the geographic entity's state, quantity, and other attributes.

[0160] For example, if resource allocation is to deliver materials to a geographic entity, the incremental update instruction will include the route coordinates of the material transportation, the time range of the transportation, and the modified values ​​of the attributes of the geographic entity after receiving the materials.

[0161] Step S153: Perform spatiotemporal conflict detection on the incremental update instruction to verify whether the resource allocation path has overlapping conflicts with the spatial coordinate sequence of existing tasks in the geographic information system under the timestamp constraint condition, and dynamically adjust the priority order of the resource allocation path based on the conflict detection result.

[0162] Perform spatiotemporal conflict detection on the generated incremental update instructions. In a GIS, there may already be some tasks in progress, each with its own spatial coordinate sequence and time range. It is necessary to verify whether the resource allocation path, under the timestamp constraint, overlaps with the spatial coordinate sequence of existing tasks.

[0163] Conflicts can be determined by comparing the spatial coordinate sequences of resource allocation paths with those of existing tasks, as well as their timeframes. If conflicts exist, the priority of the resource allocation paths can be dynamically adjusted based on the conflict detection results. Higher-priority resource allocation paths can be prioritized, while lower-priority resource allocation paths can have their timeframes or paths adjusted to avoid conflicts.

[0164] For example, if the resource allocation path overlaps with the path of an existing task within a certain time period, and the existing task has a higher priority, the time or path of the resource allocation path needs to be adjusted to ensure that no conflict occurs.

[0165] Step S154: Split the adjusted incremental update instruction into multiple atomic transaction operation units according to the schedule, bind each atomic transaction operation unit to the corresponding geographic entity spatial location identifier and resource allocation parameter verification rules, and add a version lock identifier to each atomic transaction operation unit so that concurrent operations on the same geographic entity are executed in priority order.

[0166] Split the adjusted incremental update instructions into multiple atomic transaction units according to the schedule. Atomic transaction units are indivisible units of operation, each corresponding to a specific resource allocation operation. For example, allocating a certain amount of resources to a geographic entity can be considered an atomic transaction unit.

[0167] Each atomic transaction operation unit is bound to a corresponding geographic entity spatial location identifier and resource allocation parameter verification rules. The geographic entity spatial location identifier is used to determine the geographic entity targeted by the operation unit, and the resource allocation parameter verification rules are used to verify the legality of the resource allocation parameters of the operation unit. For example, the resource allocation quantity is verified to be within a reasonable range.

[0168] A version lock identifier is added to each atomic transaction operation unit. This identifier is used to control concurrent operations on the same geographic entity. When multiple atomic transaction operation units simultaneously operate on a geographic entity, the version lock identifier and priority order ensure that the operations are executed sequentially, avoiding data conflicts and inconsistencies.

[0169] Step S155: triggering the asynchronous execution queue of the atomic transaction operation unit based on the timestamp constraint of the schedule, monitoring the execution status of each atomic transaction operation unit in real time and capturing resource allocation parameter verification failure events.

[0170] The asynchronous execution queue of atomic transaction units is triggered based on the timestamp constraints of the schedule. The asynchronous execution queue can process multiple atomic transaction units simultaneously, improving execution efficiency. When the execution time of an atomic transaction unit reaches, it is added to the asynchronous execution queue for execution.

[0171] Monitor the execution status of each atomic transaction operation unit in real time. This can be done by recording information such as the operation unit's start time, end time, and execution result. Furthermore, resource allocation parameter verification failure events are captured. If the resource allocation parameters of an atomic transaction operation unit do not meet the verification rules, a verification failure event is triggered.

[0172] Step S156: based on the resource allocation parameter verification failure event, trace back to the corresponding geographic entity spatial location identifier, regenerate the local incremental update instruction and insert it into the preset fault tolerance time window of the asynchronous execution queue.

[0173] When a resource allocation parameter verification failure event is detected, the event is traced back to the spatial location identifier of the corresponding geographic entity. The cause of the verification failure may be unreasonable resource allocation parameters, changes in the geographic entity status, etc.

[0174] Based on the analysis results, local incremental update instructions are regenerated. These instructions only update geographic entities that failed validation, adjusting resource allocation parameters to ensure compliance with validation rules. The regenerated local incremental update instructions are inserted into the preset fault tolerance time window of the asynchronous execution queue. The preset fault tolerance time window is a time range reserved for handling abnormal situations such as validation failures. During this time window, the operation can be re-executed to ensure a smooth GIS update.

[0175] Step S157: Rendering a resource allocation heat map corresponding to the executed atomic transaction operation unit in the visualization layer of the geographic information system, superimposing the change trajectory of the geographic entity state prediction result, and generating a dynamically updated spatial decision view.

[0176] Render a resource allocation heat map corresponding to executed atomic transaction units in the GIS visualization layer. The resource allocation heat map uses color depth to represent the density and intensity of resource allocation, with darker colors indicating greater resource allocation. Heat maps provide a visual overview of resource allocation in geographic space.

[0177] Overlay the change trajectory of the geographic entity state prediction results. The geographic entity state prediction results provide the state changes of the geographic entity in the future time window. Overlaying its change trajectory on the resource allocation heat map can provide a more comprehensive understanding of the dynamic changes in the geographic space.

[0178] Generate dynamically updated spatial decision views. As atomic transaction units are continuously executed and the status of geographic entities changes, the spatial decision view is updated in real time, providing decision makers with the latest geospatial information to help them make more reasonable decisions.

[0179] Step S158: reversely map the execution log of the atomic transaction operation unit with the geographic entity spatial location identifier, generate an incremental version snapshot, and synchronize it to the geographic spatial data copy of the edge node.

[0180] Reverse-map the execution log of each atomic transaction operation unit to the spatial location identifier of the geographic entity. The execution log records the execution status of each atomic transaction operation unit, including information such as the time, content, and results of the operation. Through reverse mapping, the execution log can be associated with the corresponding spatial location identifier of the geographic entity, facilitating subsequent query and analysis.

[0181] Generate an incremental version snapshot. An incremental version snapshot records the updates to the geographic information system at a specific point in time and contains only the incremental data associated with the atomic transaction operation unit. Generating incremental version snapshots can reduce data transmission and storage overhead.

[0182] Synchronize incremental version snapshots to the geospatial data replicas on edge nodes. Edge nodes are distributed nodes in a geographic information system that store replicas of geospatial data. Synchronizing incremental version snapshots to edge nodes ensures that data on edge nodes is consistent with that on the master node, improving the reliability and availability of the geographic information system.

[0183] Throughout the geospatial data management process, data authorization is a crucial step in ensuring the legal use of data. When acquiring geospatial data sets for a target area, explicit authorization is required from the data provider. Data providers may include government departments, research institutions, and businesses.

[0184] For data authorization, a detailed authorization management mechanism needs to be established. First, the data provider's identity and qualifications must be verified. Identity verification can be performed by verifying their registration information and relevant documents. Qualification verification involves verifying the legal ability to collect and provide data, such as whether they possess relevant surveying and mapping qualifications.

[0185] When obtaining authorization, the scope of authorization must be clearly defined, including the purpose of data use, duration of use, and usage permissions. The purpose of use should be strictly limited to geospatial data management and analysis, and must not be used for other illegal or unauthorized purposes. The duration of use should be set based on actual needs and the requirements of the data provider. If continued use of the data is required after the expiration of the term, the authorization must be renewed promptly. Regarding usage permissions, it must be clearly stated whether the data can be modified, shared, distributed, and so on.

[0186] For example, if you obtain land use data for a specific city from a government agency, the license agreement clearly stipulates that the data is only used for geospatial planning analysis of that city, with a usage period of one year, and that the data may not be shared with third parties. These license terms must be strictly adhered to during use.

[0187] At the same time, a data authorization record and audit mechanism should be established. Detailed information on each data authorization should be recorded, including the authorization time, the authorizer, and the scope of authorization. Regular audits should be conducted on the use of data authorizations to check for violations of authorization terms. If violations are discovered, timely measures should be taken, such as suspending data usage permissions and pursuing legal action.

[0188] Data collection is a fundamental part of geospatial data management, and its legitimacy and fairness must be ensured. When selecting data collection methods and equipment, relevant laws, regulations, and industry standards must be followed.

[0189] When collecting satellite remote sensing imagery, ensure that satellite operations comply with relevant international and domestic regulations. Satellite launches and operations are subject to strict approval and oversight. When using satellite remote sensing imagery, obtain it through legal channels and adhere to the data provider's usage regulations.

[0190] The use of ground surveying and mapping equipment must also comply with relevant standards. For example, total stations, GPS receivers, and other equipment must undergo quality inspection and calibration to ensure the accuracy and reliability of the collected data. During the data collection process, equipment operators must undergo training and assessment to ensure they possess professional operating skills and knowledge.

[0191] During data collection, personal privacy and sensitive information must be avoided. Geospatial data may contain personal information, such as building occupants. Necessary technical and management measures must be implemented during data collection to filter and protect this sensitive information.

[0192] For example, when using drones for aerial surveys, avoid photographing the interiors of private residences. If potentially sensitive information cannot be avoided, it should be encrypted and access rights should be strictly controlled during data use.

[0193] At the same time, the fairness of data collection must be ensured. When selecting the collection area and time, data bias due to human factors must be avoided. For example, data collection should not be limited to certain areas or specific times, while ignoring other areas and times. Data must be ensured to fully and objectively reflect the geospatial characteristics of the target area.

[0194] Figure 2 FIG. 1 shows the hardware structure of an artificial intelligence-based geospatial data management system 100 for implementing the artificial intelligence-based geospatial data management method according to an embodiment of the present invention. Figure 2 As shown, the artificial intelligence-based geospatial data management system 100 may include a processor 110 , a machine-readable storage medium 120 , a bus 130 , and a communication unit 140 .

[0195] In one possible design, the artificial intelligence-based geospatial data management system 100 can be a single server or a server group. The server group can be centralized or distributed (for example, the artificial intelligence-based geospatial data management system 100 can be a distributed system). In some embodiments, the artificial intelligence-based geospatial data management system 100 can be local or remote. For example, the artificial intelligence-based geospatial data management system 100 can access information and / or data stored in the machine-readable storage medium 120 via a network. For another example, the artificial intelligence-based geospatial data management system 100 can directly connect to the machine-readable storage medium 120 to access the stored information and / or data.

[0196] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data obtained from an external terminal. In some embodiments, the machine-readable storage medium 120 may store data and / or instructions that the artificial intelligence-based geospatial data management system 100 may execute or use to implement the exemplary methods described herein.

[0197] During the specific implementation process, one or more processors 110 execute computer executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the artificial intelligence-based geospatial data management method of the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 can be used to control the sending and receiving actions of the communication unit 140.

[0198] The specific implementation process of the processor 110 can be found in the various method embodiments executed by the above-mentioned artificial intelligence-based geospatial data management system 100. The implementation principles and technical effects are similar and will not be repeated here in this embodiment.

[0199] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer-executable instructions are set. When a processor runs the computer-executable instructions, the above-mentioned artificial intelligence-based geospatial data management method is implemented.

[0200] It should be noted that, in order to simplify the description of the present invention and thus facilitate understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the invention, various features may sometimes be grouped together into one embodiment, figure, or description thereof. Similarly, it should be noted that, in order to simplify the description of the present invention and thus facilitate understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the invention, various features may sometimes be grouped together into one embodiment, figure, or description thereof.

Claims

1. A geospatial data management method based on artificial intelligence, characterized in that: The method comprises: Acquire a geospatial data set of a target area, wherein the geospatial data set includes time-series-changing geographical entity observation records and corresponding spatial location identifiers; Extracting features from the geographic spatial data set to generate temporal correlation features and spatial topological features of the geographic entity observation records; Calling a pre-trained geospatial analysis model to perform dynamic feature fusion processing on the temporal correlation features and the spatial topology features to generate a fused feature set; generating a geospatial optimization strategy based on the fused feature set, wherein the geospatial optimization strategy includes a geographic entity state prediction result and a resource allocation priority configuration; A geographic information system update operation of the target area is triggered according to the geospatial optimization strategy.

2. The method according to claim 1, characterized in that The feature extraction of the geographic spatial data set to generate temporal correlation features and spatial topological features of the geographic entity observation records includes: Performing a data integrity check operation on the geographic entity observation records in the geospatial data set, detecting the timestamps and spatial location identifiers of missing data, and generating supplementary data based on an interpolation algorithm of adjacent observation records to form a preprocessed geographic entity data set with continuous spatiotemporal coverage; Performing outlier detection processing on the preprocessed geographic entity data set, calculating the numerical distribution range of each geographic entity observation record, removing data points that exceed the preset distribution range, and performing sliding window mean filling processing on the vacant positions after removal to generate a standardized geographic entity data set; Performing a time series segmentation operation on the standardized geographic entity data set, dividing the geographic entity observation records into a plurality of time window subsequences according to a preset period length, each time window subsequence containing a fixed number of consecutive observation records, and performing overlapping cutting processing on the observation records at the time window boundary to avoid data truncation; Perform trend decomposition processing on each time window subsequence, and use seasonality and trend decomposition algorithm to separate the long-term trend component, seasonal cycle component and residual noise component of the geographic entity observation record, and extract the slope change characteristics of the long-term trend component and the amplitude fluctuation characteristics of the seasonal cycle component as time series correlation features; Performing spatial clustering analysis based on the spatial location identifier to construct a topological network containing spatial adjacency relationships of geographic entities, and extracting spatial topological features of the topological network; The temporal correlation features are aligned with the spatial topological features, and the temporal correlation features of all time window subsequences in the spatial subregion are aggregated to generate feature dimensions consistent with the number of spatial subregions, so that the mapping relationship between the time window subsequences and the spatial subregions is consistent, and the aligned features are combined into a multidimensional feature matrix.

3. The method according to claim 2, characterized in that The performing of a time series segmentation operation on the standardized geographic entity data set to divide the geographic entity observation records into a plurality of time window subsequences according to a preset period length includes: Obtaining a timestamp sequence of all geographic entity observation records in the standardized geographic entity data set, sorting the observation records in the order of the timestamps, and generating an ordered time series data stream; Determine the time window length and sliding step parameters according to a preset division rule, and divide the ordered time series data stream into multiple initial time window subsequences in a sliding window manner; Detecting the number of observation records in each initial time window subsequence, and if the number is lower than a preset minimum record threshold, performing a window expansion operation to dynamically adjust the window length to include a sufficient number of observation records; The expanded time window subsequences are time aligned, and the dynamic time warping algorithm is used to align the time offsets of observation records in different time windows to eliminate the time sequence misalignment caused by inconsistent data acquisition frequencies. The aligned time window subsequences are normalized, the normalized time window subsequences are associated with the corresponding spatial location identifiers and stored, and a bidirectional mapping relationship between the time window index and the spatial region identifier is established.

4. The method according to claim 2, characterized in that The performing of spatial clustering analysis based on the spatial location identifier to construct a topological network including spatial adjacency relationships of geographic entities and extracting spatial topological features of the topological network includes: Parse the latitude and longitude coordinate information of the spatial location identifier of each geographic entity and construct a geographic entity coordinate set; Calculating the spatial distance matrix between each pair of geographic entity coordinates, dynamically adjusting the adjacency threshold based on the distribution density of the geographic entity coordinates, and establishing a corresponding connection edge in the topological network if the spatial distance between two geographic entities is less than the dynamic adjacency threshold; A weight assignment operation is performed on the connection edge, wherein the weight value is inversely proportional to the spatial distance, and the smaller the spatial distance, the larger the weight value of the connection edge; Traverse all geographic entity nodes, construct an undirected topological network containing weighted connection edges, and calculate the degree centrality index of each node to characterize its spatial connection density; executing a community detection algorithm on the undirected topological network to identify geographical entity subgroups with high cohesion, and extracting spatial coverage and internal connection strength indicators of each subgroup; A spatial topology feature vector is generated based on the degree centrality index and subgroup features, where the feature vector includes node connection density, subgroup coverage radius, and cross-group connection strength.

5. The method according to claim 1, wherein The calling of the pre-trained geospatial analysis model to perform dynamic feature fusion processing on the temporal correlation features and the spatial topology features to generate a fused feature set includes: Input the temporal correlation features into the temporal feature encoder of the geospatial analysis model, extract local temporal patterns at different time scales through a temporal convolution layer, and use a multi-head attention mechanism to capture global dependencies across time windows to generate a temporal encoding feature vector; Inputting the spatial topological features into the spatial feature encoder of the geospatial analysis model, aggregating the topological features of adjacent nodes through a graph convolution layer, and compressing the feature dimensions using a spatial pooling layer to generate a spatial encoding feature vector; Performing a feature alignment operation on the temporal coding feature vector and the spatial coding feature vector, adjusting the time dimension of the temporal coding feature vector and the spatial dimension of the spatial coding feature vector to the same order of magnitude by spatial sub-region aggregation or time interpolation, and filling in null values ​​at feature missing positions; Constructing a cross-modal attention mechanism, calculating a correlation matrix between the temporal encoding feature vector and the spatial encoding feature vector, and generating a dynamic weight coefficient based on the correlation matrix; Perform weighted fusion of the temporal coding feature vector and the spatial coding feature vector according to the dynamic weight coefficient to generate a preliminary fusion feature matrix; A feature enhancement operation is performed on the preliminary fused feature matrix, the interaction between features is enhanced through a nonlinear transformation layer, and redundant noise is removed using a denoising autoencoder to generate an optimized fused feature set.

6. The method according to claim 5, characterized in that The constructing of a cross-modal attention mechanism, calculating a correlation matrix between the temporal coding feature vector and the spatial coding feature vector, and generating a dynamic weight coefficient based on the correlation matrix, includes: Map the temporal encoding feature vector and the spatial encoding feature vector to a high-dimensional latent space to generate a temporal query vector and a spatial key vector respectively; Calculate the dot product similarity between the temporal query vector and the spatial key vector to generate the initial attention score matrix; Normalize the initial attention score matrix, use the Softmax function to normalize the score into a probability distribution, and generate the attention weight matrix; Perform weighted aggregation on the spatial encoding feature vector according to the attention weight matrix to generate a spatial attention feature vector; The spatial attention feature vector is concatenated with the temporal encoding feature vector, and the cross-modal information is fused through a fully connected layer to generate a joint feature representation. The joint feature representation is layer-normalized to eliminate the scale differences between features, and the original feature information is retained using residual connections to generate the final dynamic weight coefficients.

7. The method according to claim 1, characterized in that Generating a geospatial optimization strategy based on the fused feature set includes: Inputting the fused feature set into the geographic entity state prediction model, calculating the probability of the state change of the geographic entity in the future time window through the multi-layer perceptron network, and generating a probability distribution prediction result; Identifying risky geographic entities based on the probability distribution prediction results, calculating resource demand urgency scores for the risky geographic entities, and generating a priority list by sorting based on the resource demand urgency scores; Constructing a multi-objective optimization model with maximizing resource allocation efficiency and minimizing geographic entity protection costs as objective functions, introducing constraints to limit the total amount of resources, allocation time window, and resource allocation order based on the priority list; An evolutionary algorithm is used to solve the multi-objective optimization model, generate a Pareto optimal solution set, and select the optimal solution that meets the preset strategy through an interactive decision interface; Integrating the resource allocation plan corresponding to the optimal solution with the geographic entity state prediction results to generate a geospatial optimization strategy including specific execution steps and a timetable; The feasibility of the geospatial optimization strategy is verified, and the changes in the state of geographic entities after the execution of the geospatial optimization strategy are simulated. If the simulation result exceeds the preset risk threshold, the objective function weight or constraint relaxation range of the multi-objective optimization model is dynamically adjusted according to the risk type, and the geospatial optimization strategy is regenerated.

8. The method according to claim 7, characterized in that The method of using an evolutionary algorithm to solve the multi-objective optimization model, generating a Pareto optimal solution set, and selecting the optimal solution that meets the preset strategy through an interactive decision interface includes: Setting population initialization parameters, crossover probability parameters and mutation probability parameters of the evolutionary algorithm; An initial population is generated based on the resource allocation priority configuration and the constraints of the multi-objective optimization model, wherein each individual represents a resource allocation scheme. A constraint satisfaction algorithm is used to verify the legitimacy of each individual, eliminating individuals that violate the total resource constraints, allocation time window constraints, and resource allocation sequence constraints, and supplementing and generating new individuals that meet the constraints. Performing fitness evaluation on each individual in the initial population, and calculating a multi-objective fitness value based on the objective function of maximizing resource allocation efficiency and the objective function of minimizing geographic entity protection cost; Performing genetic operations on the initial population, including a selection operation based on the fitness value, a crossover operation based on the crossover probability parameter, and a mutation operation based on the mutation probability parameter, to generate an offspring population; Merging the initial population with the offspring population, performing a non-dominated sorting process to determine the Pareto front rank of individuals, and calculating the crowding distance to screen diverse individuals, thereby generating a new generation population; Iteratively execute the above steps until the evolution termination condition is reached, and output the Pareto optimal solution set; The Pareto optimal solution set is visualized in the interactive decision interface, a selection instruction input by a user is received, and a final optimal solution is determined based on the preset strategy matching degree.

9. The method according to claim 1, characterized in that The triggering of a geographic information system update operation of a target area according to the geospatial optimization strategy includes: Analyze the specific execution steps and timetable of the geospatial optimization strategy, and extract resource allocation plan parameters that match the geo-entity state prediction results and the spatial location identifier of the target geo-entity; Generate an incremental update instruction for the geographic information system according to the resource allocation plan parameters, wherein the incremental update instruction includes a spatial coordinate sequence of the resource allocation path, a timestamp constraint condition, and a corresponding geographic entity attribute modification value; Performing spatiotemporal conflict detection on the incremental update instruction to verify whether the resource allocation path has overlapping conflicts with the spatial coordinate sequence of existing tasks in the geographic information system under the timestamp constraint condition, and dynamically adjusting the priority order of the resource allocation path based on the conflict detection result; Splitting the adjusted incremental update instruction into multiple atomic transaction operation units according to the schedule, binding each atomic transaction operation unit to a corresponding geographic entity spatial location identifier and resource allocation parameter verification rule, and adding a version lock identifier to each atomic transaction operation unit so that concurrent operations on the same geographic entity are executed in priority order; triggering the asynchronous execution queue of the atomic transaction operation unit based on the timestamp constraint of the schedule, monitoring the execution status of each atomic transaction operation unit in real time and capturing resource allocation parameter verification failure events; According to the resource allocation parameter verification failure event, the resource allocation parameter verification event is traced back to the corresponding geographic entity spatial location identifier, and a local incremental update instruction is regenerated and inserted into a preset fault-tolerant time window of the asynchronous execution queue; Rendering a resource allocation heat map corresponding to the executed atomic transaction operation unit in the visualization layer of the geographic information system, superimposing the change trajectory of the geographic entity state prediction result, and generating a dynamically updated spatial decision view; The execution log of the atomic transaction operation unit is reversely mapped with the spatial location identifier of the geographic entity, and an incremental version snapshot is generated and synchronized to the geographic spatial data copy of the edge node.

10. A geospatial data management system based on artificial intelligence, characterized in that: It includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to run the programs, instructions or codes in the memory to implement the artificial intelligence-based geospatial data management method described in any one of claims 1 to 9.

Images