A natural resource three-dimensional space-time database construction method and system

Abstract

The application relates to the technical field of database construction, in particular to a natural resource three-dimensional space-time database construction method and system, which comprises the following steps: constructing a three-dimensional space-time database of a geographic space region, and dividing the three-dimensional space-time database into a plurality of grids; acquiring resource attribute data of each grid in real time; acquiring resource characteristic heterogeneity of each grid at a current sampling time according to a spatial structure distortion degree of a region where each grid is located and a similarity degree of the resource attribute data; acquiring resource instability at the current sampling time according to a state change characteristic of the resource characteristic heterogeneity in a historical period and a discrete degree of all resource characteristic heterogeneities at the current sampling time, so as to acquire a weight factor in a multi-objective genetic algorithm, and then generate an optimal monitoring parameter layout scheme. The weight factor in the multi-objective genetic algorithm is corrected in a self-adaptive mode, and the ability of the database to describe nonlinear spatial distribution characteristics is improved.

Description

A method and system for constructing a three-dimensional spatiotemporal database of natural resources Technical Field

[0001] This application relates to the field of database construction technology, specifically to a method and system for constructing a three-dimensional spatiotemporal database of natural resources. Background Technology

[0002] In the context of new urbanization and sustainable development, the past two-dimensional management model is no longer sufficient to meet the requirements. As a result, a three-dimensional spatiotemporal database of natural resources has emerged. This database integrates various information data, including surface modeling, underground space, surface buildings, and geological information, and collects data in real time. Through various means such as oblique photography, laser scanning, and Internet of Things sensing, the database achieves three-dimensional modeling from vertical to horizontal, promoting the transformation of cities from two-dimensional to three-dimensional, improving resource utilization, and enhancing urban resilience.

[0003] Because three-dimensional spatial analysis and calculation are inherently more complex than two-dimensional planar analysis, three-dimensional analysis requires extensive calculations on massive numbers of voxels and temporal-spatial relationships. Due to the significant differences among various elements of urban natural resources in spatiotemporal attributes, existing three-dimensional databases struggle to accurately quantify the spatiotemporal evolution differences of natural resources in three-dimensional space, and cannot effectively integrate spatial structure and resource attributes. This results in unreasonable allocation of monitoring parameters, inaccurate coverage, and consequently, insufficient ability to characterize the nonlinear spatial distribution features of complex terrain, geological structures, etc., within the database. Summary of the Invention

[0004] To address the aforementioned technical problems, the purpose of this application is to provide a method and system for constructing a three-dimensional spatiotemporal database of natural resources. The specific technical solution adopted is as follows:

[0005] In a first aspect, embodiments of this application provide a method for constructing a three-dimensional spatiotemporal database of natural resources, the method comprising the following steps:

[0006] Construct a three-dimensional spatiotemporal database of a geospatial region and uniformly divide its corresponding coordinate system into several grids; acquire various resource attribute data of each grid within a preset historical time period in real time;

[0007] The spatial coordinates of all grids are transformed from high-dimensional space to low-dimensional manifold space. Based on the difference in distance between each grid and its nearest neighbor in high-dimensional space and low-dimensional manifold space, as well as the similarity of various resource attribute data between each grid and its neighboring grids, the resource feature heterogeneity of each grid at the current sampling time is obtained, which is used to characterize the possibility that each grid is located in a complex geographical region with distorted geographical structure.

[0008] Based on the state changes of resource characteristic heterogeneity of all grids between all adjacent sampling times before the current sampling time, a state transition frequency is constructed. Combined with the distribution dispersion of resource characteristic heterogeneity of all grids at the current sampling time, the resource instability at the current sampling time is obtained, which is used to characterize the instability of natural resource evolution in a geographic spatial region.

[0009] The weight factor at the current sampling time is obtained based on the resource instability, and the fitness function of the multi-objective genetic algorithm is constructed to generate the optimal monitoring parameter layout scheme for the three-dimensional spatiotemporal database at the current sampling time.

[0010] Preferably, the specific process of converting from a high-dimensional space to a low-dimensional manifold space is as follows:

[0011] The vector composed of all class resource attribute data of each grid at each sampling time is denoted as the resource attribute vector of each grid at each sampling time;

[0012] Construct the Euclidean distance matrix of the entire geospatial region at the current sampling time; wherein the size of the Euclidean distance matrix is ​​N×N, where N is the total number of grids, and the element in the nth row and jth column of the Euclidean distance matrix is ​​the Euclidean distance between the resource attribute vectors of the nth and jth grids at the current sampling time;

[0013] The Euclidean distance matrix at the current sampling time and the spatial coordinates of all grid center points are used as inputs to the Isomap algorithm, which outputs the low-dimensional manifold space at the current sampling time and the low-dimensional coordinates of all grids.

[0014] Preferably, the method for obtaining the resource feature heterogeneity of each grid at the current sampling time is as follows:

[0015] The spatial difference factor of each grid at the current sampling time is obtained based on the difference in distance between each grid and its nearest neighbor in the high-dimensional space and low-dimensional manifold space at the current sampling time.

[0016] The algorithm takes the resource attribute vectors of all grids at the current sampling time as input to the spatial autocorrelation analysis algorithm and outputs the local Moran index of each grid at the current sampling time.

[0017] The resource feature heterogeneity of each grid at the current sampling time is positively correlated with the spatial difference factor and negatively correlated with the local Moran index.

[0018] Preferably, the method for obtaining the spatial difference factor of each grid at the current sampling time is as follows:

[0019] Based on the Euclidean distance, obtain a preset number of nearest neighbors for each grid in the high-dimensional space at the current sampling time; based on the geodesic distance, obtain a preset number of nearest neighbors for each grid in the low-dimensional manifold space at the current sampling time.

[0020] Calculate the mean Euclidean distance between each grid cell and the spatial coordinates of all its nearest neighbors in the high-dimensional space at the current sampling time; calculate the mean Euclidean distance between each grid cell and the low-dimensional coordinates of all its nearest neighbors in the low-dimensional manifold space at the current sampling time.

[0021] The absolute difference between the two means is denoted as the spatial difference factor of each grid at the current sampling time.

[0022] Preferably, the method for constructing the state transition frequency is as follows:

[0023] The resource feature heterogeneity of each grid in the preset historical period before the current sampling time is sorted in chronological order to obtain the resource feature heterogeneity sequence of each grid at the current sampling time.

[0024] Arrange the resource feature heterogeneity sequences of all grids at the current sampling time from top to bottom to obtain the resource feature heterogeneity matrix of the entire geographic spatial region at the current sampling time;

[0025] The obtained resource feature heterogeneity matrix is ​​used as the input of the hidden Markov model, and the hidden state at each sampling time in the preset historical period is obtained by iterative convergence through the Baum-Welch algorithm.

[0026] The formula for calculating the state transition frequency at the current sampling time is: In the formula, The state transition frequency at the current sampling time; This represents the total number of sampling moments within a preset historical time period; , These represent the hidden states at the t-th and t+1-th sampling times in the preset historical time period, respectively; This is a state transition indicator function, where the function outputs 0 when the two input data are equal, and 1 otherwise.

[0027] Preferably, the method for obtaining the resource instability at the current sampling time is as follows:

[0028] Cluster the resource feature heterogeneity of all grids at the current sampling time;

[0029] Calculate the variance of the total number of grid cells within all clusters at the current sampling time;

[0030] The resource instability at the current sampling moment is positively correlated with both the variance and the state transition frequency.

[0031] Preferably, the weighting factor at the current sampling time is positively correlated with the resource instability.

[0032] Preferably, the specific formula for constructing the fitness function of the multi-objective genetic algorithm is as follows: In the formula, Let be the fitness function of the multi-objective genetic algorithm at the current sampling time; The weighting factor at the current sampling time; and These are the preset cost weight and the preset uniform weight, respectively, both of which have a value range of [0.5, 1]. This represents the average data update frequency of all monitoring devices in the entire geographic space at the current sampling time. This is the ratio between the sum of the maintenance costs of all items in the 3D spatiotemporal database at the current sampling time and the preset cost. This represents the variance of the maintenance cost of all items in the 3D spatiotemporal database at the current sampling time.

[0033] Preferably, the method for obtaining the optimal monitoring parameter layout scheme is as follows: all grids in the entire geographic space area are used as input to a multi-objective genetic algorithm, the population size is set to a number of individuals, each individual represents a monitoring parameter layout scheme, and Pareto sorting is used as the selection mechanism to finally output the optimal monitoring parameter layout scheme.

[0034] Secondly, embodiments of this application also provide a system for constructing a three-dimensional spatiotemporal database of natural resources, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described methods for constructing a three-dimensional spatiotemporal database of natural resources.

[0035] This application has at least the following beneficial effects:

[0036] To address the shortcomings of existing 3D spatiotemporal databases in meeting the requirements of complex spatial computation and refined spatial management, this application employs a unified spatial grid unit. It utilizes spatial autocorrelation analysis and the Isomap algorithm to characterize the spatial clustering of resource attribute data for each grid, as well as the nonlinear distortion characteristics in local spatial structures. This constructs the resource characteristic heterogeneity of each grid, thereby quantifying the probability that each grid is located in a complex geographical region with distorted geographic structures. By analyzing the complexity of the resource characteristic heterogeneity of each grid at the current moment, and the frequency characteristics of state changes in the geographic spatial region during evolution, the resource instability at the current sampling moment is constructed. This quantifies the instability of natural resource evolution, avoiding the analytical limitations of single-time and single-space dimensions. Furthermore, the resource stability is combined with adaptive correction of the weight factors in the multi-objective genetic algorithm, enabling adaptive optimization of the monitoring equipment update frequency. Ultimately, this yields the optimal monitoring parameter layout scheme, thereby improving the database's ability to characterize nonlinear spatial distribution features such as complex terrain and geological structures. Attached Figure Description

[0037] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] Figure 1 is a flowchart of the steps of a method for constructing a three-dimensional spatiotemporal database of natural resources according to an embodiment of this application;

[0039] Figure 2 is a flowchart of obtaining the resource instability at the current sampling moment according to an embodiment of this application. Detailed Implementation

[0040] To further illustrate the technical means and effects adopted by this application to achieve the intended inventive purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method and system for constructing a three-dimensional spatiotemporal database of natural resources proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0042] The following description, in conjunction with the accompanying drawings, details the specific scheme of the method and system for constructing a three-dimensional spatiotemporal database of natural resources provided in this application.

[0043] Please refer to Figure 1, which shows a flowchart of the steps of a method for constructing a three-dimensional spatiotemporal database of natural resources according to an embodiment of this application. The method includes the following steps:

[0044] Step 1: Construct a three-dimensional spatiotemporal database of the geospatial region and divide its corresponding coordinate system into several grids evenly; acquire various resource attribute data of each grid within a preset historical time period in real time.

[0045] By integrating airborne and vehicle-mounted lidar, oblique photogrammetry, ground-penetrating radar, borehole exploration, and IoT sensor networks, multi-source heterogeneous data (including but not limited to resource attribute data and point cloud data) covering a geographic area within a preset historical time period are collected in real time, encompassing the surface, above-ground buildings, underground spaces, and geological bodies. In this embodiment, the sampling time interval for each type of data is one day, and the preset historical time period refers to six months of historical data. All data from each sampling moment are uniformly converted to the National Geodetic Coordinate System 2000 and UTC time reference to construct a spatiotemporally integrated three-dimensional spatiotemporal database.

[0046] Furthermore, a more refined description and analysis of the real-world geographic region is needed. Specifically, in the coordinate system corresponding to the constructed 3D spatiotemporal database, the X, Y, and Z axes are divided every v meters, dividing the geographic space corresponding to the entire coordinate system into multiple uniformly sized cubic grids. In this embodiment, v is set to 1, but the implementer can choose a value based on the actual distribution of geographic resources. By calling the natural resource thematic database interface (such as WCS, WFS, and other map services) and based on the spatial location of each grid, the resource attribute data of each grid is obtained in real time. The resource attribute data includes, but is not limited to, land use type, vegetation cover, lithology code, and water content attribute. It should be noted that for land use type and lithology code in the resource attribute data, unique thermal coding is used to map them to numerical values; for vegetation cover and water content, measured values ​​are used directly; for grids such as buildings where resource attribute data such as vegetation cover and water content cannot be directly measured, their corresponding attribute values ​​are uniformly assigned to 0.

[0047] To conduct subsequent spatial structure analysis, the vector composed of all class resource attribute data in numerical form within each grid at each sampling time is denoted as the resource attribute vector of each grid at each sampling time.

[0048] Thus, the resource attribute vectors of each grid at each sampling time in the entire geospatial region have been obtained.

[0049] Step 2: Convert the spatial coordinates of all grids from high-dimensional space to low-dimensional manifold space. Based on the difference in distance between each grid and its nearest neighbor in high-dimensional space and low-dimensional manifold space, as well as the similarity of various resource attribute data between each grid and its neighboring grids, obtain the resource feature heterogeneity of each grid at the current sampling time. This is used to characterize the possibility that each grid is located in a complex geographical region with distorted geographical structure.

[0050] Since the distribution of natural resources in three-dimensional space is often affected by nonlinear factors such as topographic relief, geological structure and hydrological processes, it exhibits manifold morphological characteristics. For spatial analysis methods based on traditional linear dimensionality reduction algorithms, it is difficult to accurately characterize geometric structures with manifold morphological characteristics such as bending, folding and non-uniform distribution, which easily leads to insufficient quantitative representation of spatial heterogeneity.

[0051] Therefore, the Euclidean distance between the resource attribute vectors of any two grids in the entire geospatial region at the current sampling time is calculated, constructing an Euclidean distance matrix for the entire geospatial region at the current sampling time. The Euclidean distance matrix has a size of N×N, where N is the total number of grids. The element in the nth row and jth column of the Euclidean distance matrix represents the Euclidean distance between the resource attribute vectors of the nth and jth grids at the current sampling time. The resulting Euclidean distance matrix reflects the resource attribute similarity among all grids. Then, the Euclidean distance matrix for the entire geospatial region at the current sampling time and the spatial coordinates of all grid center points are used as input to the Isomap algorithm. This algorithm transforms the data of each grid at the current sampling time from a high-dimensional space into a low-dimensional manifold space, outputting the low-dimensional manifold space of the entire geospatial region at the current sampling time and the low-dimensional coordinates of all grids.

[0052] A preset number k is set, in this embodiment k=50. The nearest neighbor points are selected to represent the local topological connections between grids in different spaces. The k nearest neighbors of each grid in the high-dimensional space at the current sampling time are obtained according to the Euclidean distance, and the k nearest neighbors of each grid in the low-dimensional manifold space at the current sampling time are obtained according to the geodesic distance. The Euclidean distance between each grid in the high-dimensional space and its k nearest neighbors at the current sampling time is calculated and the average is calculated. The Euclidean distance between each grid in the low-dimensional manifold space and its k nearest neighbors at the current sampling time is calculated and the average is calculated. The absolute difference between the two averages is used as the spatial difference factor of each grid at the current sampling time. The obtained spatial difference factor reflects the degree of nonlinear distortion of the spatial structure of the local spatial region where each grid is located at the current sampling time. The larger the value, the greater the possibility that each grid is located near a topographical abrupt change or geological interface at the current sampling time.

[0053] In existing technologies, spatial autocorrelation analysis algorithms are mainly used to find and measure spatial clustering distribution. The local Moran index output by this algorithm can characterize the degree of clustering of high or low values ​​in each region. Therefore, to further quantify the spatial clustering pattern of natural resource attributes of each grid, using the resource attribute vectors of all grids in the geographic spatial region at the current sampling time as input, a spatial autocorrelation analysis algorithm is adopted, setting the neighborhood distance threshold to 200 meters to match the minimum spatial influence range of urban management, and outputting the local Moran index of each grid at the current sampling time. This value represents the similarity of each grid with its neighboring grids in terms of resource attributes at the current sampling time. The larger the value, the greater the similarity between the resource attributes of each grid and its surrounding grids at the current sampling time, and the more prominent the spatial clustering. The spatial autocorrelation analysis algorithm is a well-known technology, and the specific process will not be described in detail.

[0054] To eliminate the influence of dimensions, the spatial difference factor and local Moran's index of all grids at the current sampling time are calculated. The minimum-maximum normalization method is then used to normalize the spatial difference factor and local Moran's index of all grids, yielding normalized values ​​for each grid at the current sampling time. The minimum-maximum normalization method is a well-known technique, and its specific process will not be elaborated further.

[0055] As a preferred implementation, the resource characteristic heterogeneity of each grid at the current sampling time is obtained based on the differences in distances between each grid and its nearest neighbor in the high-dimensional space and low-dimensional manifold space, as well as the similarity of various resource attribute data between each grid and its neighboring grids at the current sampling time. This heterogeneity is used to characterize the probability that each grid is located in a complex geographical region with distorted geographical structure at the current sampling time. The method for obtaining the resource characteristic heterogeneity of each grid at the current sampling time is as follows: obtaining the spatial difference factor of each grid at the current sampling time; obtaining the local Moran index of each grid at the current sampling time; the resource characteristic heterogeneity of each grid at the current sampling time is positively correlated with the spatial difference factor and negatively correlated with the local Moran index. The positive correlation means that the dependent variable increases (decreases) as the independent variable increases (decreases), and the negative correlation means that the dependent variable decreases (increases) as the independent variable increases (decreases).

[0056] Preferably, in this embodiment, the resource feature heterogeneity of the i-th grid at the current sampling time is denoted as... Its specific expression is: In the formula, Let be the resource feature heterogeneity of the i-th grid at the current sampling time; It is the normalized value of the local Moran index of the i-th grid at the current sampling time; It is the normalized value of the spatial difference factor of the i-th grid at the current sampling time; This is a preset constant used to prevent the denominator from being 0. Its value range is [0.0001, 0.01]. In this embodiment, it is 0.001.

[0057] Among them, the results The larger the value, the more complex the structure and the more severe the nonlinear distortion in the local three-dimensional space region of the i-th grid at the current sampling time, and the greater the difference in resource attributes between the grid and its surrounding grids. In this case, the grid is more likely to be located near abrupt terrain change zones or in areas with faults in the geological structure, and thus the grid is more likely to be located in a complex geographical region.

[0058] Step 3: Based on the state changes of resource characteristic heterogeneity of all grids between all adjacent sampling times before the current sampling time, construct the state transition frequency, and combine it with the distribution dispersion of resource characteristic heterogeneity of all grids at the current sampling time to obtain the resource instability at the current sampling time, which is used to characterize the instability of natural resource evolution in the geospatial region.

[0059] In the process of constructing a three-dimensional spatiotemporal database of natural resources, the management and analysis of the database is crucial. The feature characteristics of massive units cannot reflect the overall regional collaborative evolution process and system stability in the spatiotemporal dimension. Understanding the spatial heterogeneity of individual grid units cannot provide a basis for planning decisions and development trend judgments at the overall scale. Therefore, managers cannot gain a global perspective on the dynamic situation of natural resources within the region.

[0060] Based on the above, the resource feature heterogeneity of all grids at the current sampling time is used as the input to the DBSCAN clustering algorithm. The absolute difference in resource feature heterogeneity between two grids is used to measure the distance between them. A minimum sample size of 10 is set. All grids at the current sampling time are divided into several clusters, and all clusters are output. The variance of the total number of grids within each cluster at the current sampling time is calculated. This variance measures the complexity of the natural resource heterogeneity of the entire geographic space at the current sampling time. The larger the value, the greater the difference in heterogeneity between different grids within the entire geographic space and the more irregular the spatial pattern. The DBSCAN clustering algorithm is a well-known technique, and its specific process will not be elaborated further.

[0061] To eliminate the influence of dimensional differences on subsequent calculation results, the variance of all sampling times within a preset historical period is calculated. The minimum-maximum normalization method is then used to normalize all variances, yielding the normalized value of the variance of the total number of grid cells within all clusters at each sampling time. The minimum-maximum normalization method is a well-known technique, and its specific process will not be elaborated further.

[0062] However, the spatial morphology at the current sampling moment is insufficient to analyze the overall evolution of the system. It is also necessary to quantify the dynamics and stability of the overall heterogeneity of the geographic spatial region over time. The resource feature heterogeneity of each grid within a preset historical period prior to the current sampling moment is sorted chronologically to obtain the resource feature heterogeneity sequence for each grid at the current sampling moment. These sequences are then arranged from top to bottom to obtain the resource feature heterogeneity matrix for the entire geographic spatial region at the current sampling moment (the number of rows in the matrix equals the total number of grids, and the number of columns equals the total number of elements in the resource feature heterogeneity sequence). Using this resource feature heterogeneity matrix as input, a Hidden Markov Model (HMM) algorithm is employed, setting the number of hidden states to three, corresponding to the three overall heterogeneity levels of "low," "medium," and "high." The Baum-Welch algorithm is used iteratively to converge the hidden state sequence. and state transition indicator functions; where, , , These represent the hidden states of the entire geographic spatial region at sampling times 1, 2, and T, respectively; the specific expression of the state transition indicator function is: In the formula, This is the state transition indicator function; , These represent the hidden states of the entire geographic spatial region at sampling times t and t+1, respectively. Both the HMM algorithm and the Baum-Welch algorithm are well-known techniques, and their specific processes will not be elaborated upon here.

[0063] Furthermore, based on the state changes of resource feature heterogeneity of all grids between all adjacent sampling times before the current sampling time, the state transition frequency at the current sampling time is obtained, which is used to characterize the degree of change of the hidden state of the entire geospatial region under all historical adjacent sampling times.

[0064] In this embodiment, the state transition frequency at the current sampling time is denoted as... Its specific expression is: In the formula, The state transition frequency at the current sampling time; This represents the total number of sampling moments within a preset historical time period; , These represent the hidden states of the entire geographic spatial region at sampling times t and t+1, respectively. This is the state transition indicator function.

[0065] A Hidden Markov Model (HMM) is a probabilistic graphical model used to characterize the process by which a sequence of hidden states generates an observed sequence. By deriving the hidden state sequence and transition frequencies, it represents the intrinsic evolutionary patterns of an entire geographic spatial region. It reveals the intrinsic dynamic process of heterogeneity evolution in the entire geospatial region. The larger the value, the more frequently the overall heterogeneity level of the geospatial region switches between "low, medium, and high" states at each sampling time, indicating that the evolution of the geospatial region is more unstable and more active; the smaller the value, the more stable the evolution is, as all grids in the geospatial region remain at a certain level of heterogeneity for a long time.

[0066] In a preferred embodiment, the resource instability at the current sampling time is obtained based on the state transition frequency and the dispersion of resource characteristic heterogeneity distribution across all grids at the current sampling time. This instability characterizes the degree of instability in the evolution of natural resources in the geographic spatial region at the current sampling time. The method for obtaining the resource instability at the current sampling time is as follows: calculate the variance of the total number of grids within all clusters at the current sampling time; the resource instability at the current sampling time is positively correlated with both the variance and the state transition frequency. The flowchart for obtaining the resource instability at the current sampling time is shown in Figure 2.

[0067] Preferably, in this embodiment, the resource instability at the current sampling time is recorded as... The specific calculation formula is as follows: In the formula, H represents the resource instability at the current sampling time; H is the normalized value of the variance of the total number of grids in all clusters at the current sampling time. The state transition frequency at the current sampling time; This is a preset constant.

[0068] The larger the obtained B value, the more complex the current evolution pattern of the entire geographic space region and the more unstable the overall evolution state. In this case, the more unstable the evolution of natural resources in the geographic space region at the current sampling time, the greater the weight should be set when allocating monitoring parameters at the current sampling time.

[0069] Step 4: Obtain the weight factor at the current sampling time based on resource instability, and construct the fitness function of the multi-objective genetic algorithm to generate the optimal monitoring parameter layout scheme for the three-dimensional spatiotemporal database at the current sampling time.

[0070] The resource instability obtained from the above steps can reflect the irregularity of resource distribution in the entire region and the instability of the overall regional evolution. However, since the resource instability at the current sampling time cannot directly guide the specific monitoring resource optimization and allocation work, it is difficult to effectively link system-level risk warning with unit-level decision-making actions, thus failing to achieve adaptive management.

[0071] In existing technologies, multi-objective optimization algorithms (MOGA) use fixed weights at different times when solving for resource space monitoring parameter configurations. However, these fixed weights do not consider the real-time changes of the overall system. To address this issue, this application designs a dynamic region weight allocation strategy based on resource instability. This weight parameter is embedded into the fitness function of the multi-objective genetic algorithm, ultimately enabling priority adjustment of monitoring parameters for regions with higher instability. The specific strategy is as follows:

[0072] The optimization objectives of the embedded multi-objective genetic algorithm are set as follows: maximizing the monitoring data update frequency, minimizing the total equipment deployment and maintenance cost, and maximizing the spatial uniformity of the overall monitoring network. Cost and uniformity are treated as global constraints, with their corresponding weights remaining constant to maintain the feasibility and balance of the overall scheme. The monitoring data update frequency refers to the update frequency of various data items in the three-dimensional spatiotemporal database of the geographic spatial region.

[0073] Since the resource instability at the current sampling time represents the degree of instability of the geographic space region at that time, the weight of the monitoring data update frequency for this geographic space region is dynamically and adaptively adjusted based on the resource instability at the current sampling time. This guides the MOGA algorithm to adaptively adjust the monitoring data update frequency for the current geographic space region. When the state of the geographic space region is more unstable, i.e., the resource instability at the current sampling time is greater, the update frequency of the monitoring data needs to be increased in order to better monitor the natural resources of the geographic space region. Therefore, the weight corresponding to the monitoring data update frequency at the current sampling time should be set larger, so that the allocation of monitoring parameters pays more attention to improving the data update frequency.

[0074] Based on the above analysis, calculate the weighting factor at the current sampling time. The specific formula is as follows: In the formula, The weighting factor at the current sampling time; The initial weight is preset, and its value range is [0.5, 1]. In this embodiment, it is set to 0.6. The resource instability at the current sampling moment; This is the maximum value of resource instability among all sampling times in the preset historical period.

[0075] when The closer (That is, the closer the geographic space region at the current sampling time is to the historical most unstable state), the more accurate the calculated result will be. The larger the value, the greater the collective weight of units within the high-risk level in the fitness function of the optimization algorithm, thus strongly driving the MOGA algorithm to prioritize the data update frequency of geospatial regions.

[0076] Furthermore, the fitness function G of the multi-objective genetic algorithm at the current sampling time is constructed. The specific fitness function formula is as follows: In the formula, Let be the fitness function of the multi-objective genetic algorithm at the current sampling time; The weighting factor at the current sampling time; and These are the preset cost weight and the preset uniform weight, respectively, and their values ​​range from [0.5, 1]. In this embodiment, the specific value is 0.8. The average data update frequency of all monitoring devices in the entire geographic space at the current sampling time is obtained by: statistically analyzing the data update frequency of all monitoring devices in the entire geographic space at the current sampling time under any monitoring parameter layout scheme, and recording the average of all data update frequencies as the average data update frequency of the entire geographic space at the current sampling time. The monitoring devices include, but are not limited to, UAV oblique photography systems, ground laser scanners, geological exploration equipment, and Internet of Things sensor nodes. The total cost is the ratio between the sum of all maintenance costs of the 3D spatiotemporal database at the current sampling time and the preset cost. Specifically, the predicted total cost is dynamically calculated based on the monitoring parameter layout plan, covering the entire lifecycle costs of equipment, storage, and labor. The preset cost is set based on the project budget, historical data, and management requirements, serving as a cost constraint benchmark. This represents the variance of the operational costs of all items in the 3D spatiotemporal database at the current sampling time. The calculation process for the average data update frequency and operational costs of the entire geospatial region at each sampling time is a well-known technique and will not be elaborated upon further.

[0077] Using all grids in the entire geographic space as input to the multi-objective genetic algorithm (MOGA), the population size is set to 200 individuals, with each individual representing a monitoring parameter layout scheme. The maximum number of iterations is set to 200, Pareto sorting is used as the selection mechanism, and the crossover probability is set to 0.7 and the mutation probability to 0.02. The algorithm converges after reaching the maximum number of iterations and finally outputs the optimal monitoring parameter layout scheme.

[0078] In the fitness function of the multi-objective genetic algorithm (MOGA), the weight factors at different sampling times can be used as a basis. This is used to control the proportion of geographic spatial regions with different levels of stability in the optimization objective. Based on the obtained optimal monitoring parameter layout scheme, the layout of various monitoring parameters in the three-dimensional spatiotemporal database at the current sampling time is optimized.

[0079] The above methods can adaptively adjust and optimize the layout and allocation of various monitoring parameters in the geographic space region based on the calculation results of resource instability at each sampling time, giving priority to strengthening the attention to update frequency parameters in highly unstable areas, thereby providing accurate and efficient data sampling services for the construction of a three-dimensional spatiotemporal database. This enables the data in the database to be analyzed and dynamically simulated more accurately in three-dimensional spatiotemporal space, forming a closed-loop system from data acquisition to intelligent decision-making, thus obtaining a high-quality three-dimensional spatiotemporal database.

[0080] Based on the same inventive concept as the above method, this application embodiment also provides a natural resource three-dimensional spatiotemporal database construction system, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the method described in the above-mentioned natural resource three-dimensional spatiotemporal database construction method.

[0081] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0082] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0083] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.

Claims

1. A method for constructing a three-dimensional spatiotemporal database of natural resources, characterized in that, The method includes the following steps: constructing a three-dimensional spatiotemporal database of a geographic region and uniformly dividing its corresponding coordinate system into several grids; acquiring various resource attribute data of each grid within a preset historical time period in real time; converting the spatial coordinates of all grids from a high-dimensional space to a low-dimensional manifold space; based on the difference in distance between each grid and its nearest neighbor in the high-dimensional space and the low-dimensional manifold space, and the similarity of various resource attribute data between each grid and its neighboring grids, obtaining the resource feature heterogeneity of each grid at the current sampling time, which is used to characterize the possibility that each grid is located in a complex geographic region with distorted geographic structure; and based on all adjacent samples taken before the current sampling time... The state changes of resource characteristic heterogeneity of all grids between sampling times are used to construct the state transition frequency. Combined with the dispersion of resource characteristic heterogeneity distribution of all grids at the current sampling time, the resource instability at the current sampling time is obtained to characterize the instability of natural resource evolution in a geographic spatial region. Based on the resource instability, a weighting factor is obtained for the current sampling time, and a fitness function for a multi-objective genetic algorithm is constructed to generate an optimal monitoring parameter layout scheme for the three-dimensional spatiotemporal database at the current sampling time. The weighting factor at the current sampling time is positively correlated with the resource instability. The specific formula for constructing the fitness function of the multi-objective genetic algorithm is as follows: In the formula, Let be the fitness function of the multi-objective genetic algorithm at the current sampling time; The weighting factor at the current sampling time; and These are the preset cost weight and the preset uniform weight, respectively, both of which have a value range of [0.5, 1]. This represents the average data update frequency of all monitoring devices in the entire geographic space at the current sampling time. This is the ratio between the sum of the maintenance costs of all items in the 3D spatiotemporal database at the current sampling time and the preset cost. The variance of the maintenance cost of all items in the 3D spatiotemporal database at the current sampling time is given. The method for obtaining the optimal monitoring parameter layout scheme is as follows: all grids in the entire geographic space area are used as input to a multi-objective genetic algorithm, the population size is set to a number of individuals, each individual represents a monitoring parameter layout scheme, and Pareto sorting is used as the selection mechanism to finally output the optimal monitoring parameter layout scheme.

2. The method for constructing a three-dimensional spatiotemporal database of natural resources as described in claim 1, characterized in that, The specific process of converting from high-dimensional space to low-dimensional manifold space is as follows: The vector composed of all resource attribute data of each grid at each sampling time is denoted as the resource attribute vector of each grid at each sampling time; an Euclidean distance matrix of the entire geographic space region at the current sampling time is constructed; wherein, the size of the Euclidean distance matrix is ​​N×N, where N is the total number of grids, and the element in the nth row and jth column of the Euclidean distance matrix is ​​the Euclidean distance between the resource attribute vectors of the nth and jth grids at the current sampling time; the Euclidean distance matrix at the current sampling time and the spatial coordinates of all grid center points are used as input to the Isomap algorithm, and the low-dimensional manifold space at the current sampling time and the low-dimensional coordinates of all grids are output.

3. The method for constructing a three-dimensional spatiotemporal database of natural resources as described in claim 2, characterized in that, The method for obtaining the resource feature heterogeneity of each grid at the current sampling time is as follows: The spatial difference factor of each grid at the current sampling time is obtained based on the difference in distance between each grid and its nearest neighbor in the high-dimensional space and low-dimensional manifold space; the resource attribute vectors of all grids at the current sampling time are used as input to a spatial autocorrelation analysis algorithm, and the local Moran index of each grid at the current sampling time is output; the resource feature heterogeneity of each grid at the current sampling time is positively correlated with the spatial difference factor and negatively correlated with the local Moran index.

4. The method for constructing a three-dimensional spatiotemporal database of natural resources as described in claim 3, characterized in that, The method for obtaining the spatial difference factor of each grid at the current sampling time is as follows: Obtain a preset number of nearest neighbors of each grid in the high-dimensional space at the current sampling time according to Euclidean distance; obtain a preset number of nearest neighbors of each grid in the low-dimensional manifold space at the current sampling time according to geodesic distance; calculate the mean of the Euclidean distances between each grid in the high-dimensional space and the spatial coordinates of all its nearest neighbors at the current sampling time; calculate the mean of the Euclidean distances between each grid in the low-dimensional manifold space and the low-dimensional coordinates of all its nearest neighbors at the current sampling time; and record the absolute difference between the two means as the spatial difference factor of each grid at the current sampling time.

5. The method for constructing a three-dimensional spatiotemporal database of natural resources as described in claim 1, characterized in that, The method for constructing the state transition frequency is as follows: sort the resource feature heterogeneity of each grid in the preset historical period before the current sampling time in chronological order to obtain the resource feature heterogeneity sequence of each grid at the current sampling time; arrange the resource feature heterogeneity sequences of all grids at the current sampling time from top to bottom to obtain the resource feature heterogeneity matrix of the entire geographic spatial area at the current sampling time. The obtained resource feature heterogeneity matrix is ​​used as input to the hidden Markov model, and the hidden states at each sampling time in the preset historical period are obtained through iterative convergence using the Baum-Welch algorithm; the formula for calculating the state transition frequency at the current sampling time is: In the formula, The state transition frequency at the current sampling time; This represents the total number of sampling moments within a preset historical time period; 、 These represent the hidden states at the t-th and t+1-th sampling times in the preset historical time period, respectively; This is a state transition indicator function, where the function outputs 0 when the two input data are equal, and 1 otherwise.

6. The method for constructing a three-dimensional spatiotemporal database of natural resources as described in claim 1, characterized in that, The method for obtaining the resource instability at the current sampling time is as follows: clustering the resource feature heterogeneity of all grids at the current sampling time; calculating the variance of the total number of grids in all clusters at the current sampling time; the resource instability at the current sampling time is positively correlated with the variance and the state transition frequency.

7. A system for constructing a three-dimensional spatiotemporal database of natural resources, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for constructing a three-dimensional spatiotemporal database of natural resources as described in any one of claims 1-6.

Images