GIS (geographic information system) resource management method, GIS resource management system, GIS resource management equipment and medium

Abstract

The invention discloses a GIS resource management method, system and device and a medium, and belongs to the technical field of resource management, and the method comprises the steps: obtaining a city planning map, constructing a city three-dimensional model, and collecting multi-source city feature information; constructing a multi-source data fusion matrix of a GIS graph and generating three layers of attribute tags; a spatial gravity model is constructed, and reachability analysis is carried out on urban resources; constructing a dynamic prediction model, and triggering a model iteration mechanism; and generating a resource health degree index based on a model iteration mechanism result, and triggering a resource allocation control signal. The method has the beneficial effects that through multi-source data fusion matrix construction, precise space-time association of geographic space, Internet of Things time sequence and visual feature data is realized, and the integrity and accuracy of data integration are improved; quantitative evaluation and prospective pre-judgment of resource supply and demand are realized based on reachability analysis of a space gravity model and a dynamic prediction model of a space-time diagram convolutional network.

Description

Technical Field

[0001] This invention relates to the field of resource management technology, specifically to a GIS resource management method, system, device, and medium. Background Technology

[0002] Currently, with the advancement of smart city construction, 3D motion data editing and processing technology is being used more and more widely in urban management, covering the collection and integration of geospatial data, IoT time-series data, and visual feature data. Through SLAM technology to process street view video streams and coordinate system transformation to achieve spatial alignment, 3D motion data has become the foundation for building urban 3D models and supporting resource management decisions. The demand for multi-source data fusion continues to grow with the increasing requirements of refined urban management.

[0003] Existing technologies involve the integration of traffic flow and road topology data, but do not form a multimodal fusion mechanism that links with geospatial data in real time.

[0004] However, existing GIS resource management still has significant shortcomings: First, the dynamic integration of multi-source data is insufficient, and there is a lack of real-time correlation mechanisms between geospatial, IoT time-series, and visual feature data, leading to lagging resource status assessment. Geospatial data is mostly stored in static layers with long update cycles, while IoT time-series data is highly real-time, making it easy for spatial coordinates and dynamic indicators to be out of sync when matching the two. Second, resource analysis models are static and cannot respond to dynamic scenarios such as sudden changes in traffic flow and sudden surges in resource demand. The core parameters of existing models are mostly fixed values, trained offline based on historical data, and no linkage adjustment mechanism with real-time data has been established. Faced with sudden resource demands, the models cannot adapt quickly, resulting in a disconnect between resource allocation instructions and actual needs, exacerbating the imbalance between resource supply and demand. Summary of the Invention

[0005] In view of the above-mentioned problems, the present invention is proposed.

[0006] Therefore, the technical problem addressed by this invention is: insufficient dynamic fusion of multi-source data, and a lack of real-time correlation mechanisms between geospatial, IoT time-series, and visual feature data, leading to lagging resource status assessment. Geospatial data is mostly stored in static layers with long update cycles, while IoT time-series data is highly real-time, making it prone to asynchrony between spatial coordinates and dynamic indicators when matching the two. Secondly, static resource analysis models struggle to respond to dynamic scenarios such as sudden changes in traffic flow and surges in sudden resource demand. Existing models often use fixed core parameters, are trained offline based on historical data, and lack a mechanism for linkage and adjustment with real-time data.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a GIS resource management method, comprising,

[0008] The process involves acquiring urban planning maps, constructing a 3D urban model based on these maps, and collecting multi-source urban feature information through a multi-source data acquisition system. A multi-source data fusion matrix for a single GIS map is then constructed based on the 3D urban model and the multi-source urban feature information, generating three-layer attribute labels. A spatial gravity model is built based on the multi-source data fusion matrix to analyze the accessibility of urban resources. A dynamic prediction model is constructed using machine learning, triggering an iterative model mechanism. Based on the results of this iterative model mechanism, a resource health index is generated, triggering resource allocation control signals.

[0009] As a preferred embodiment of the GIS resource management method described in this invention, the construction of a multi-source data fusion matrix for a single GIS map includes:

[0010] Multi-source urban feature information is preprocessed to form a spatial base layer.

[0011] Based on multi-source urban feature information, a rasterized time series layer is generated using a unified time reference.

[0012] Based on multi-source urban feature information, spatiotemporal pose analysis and frame sampling are performed, and the analysis results are mapped to the coordinate system of the urban 3D model to generate a dynamic texture layer.

[0013] A multi-source data fusion matrix is ​​generated based on a spatial base layer, a rasterized temporal layer, and a dynamic texture layer.

[0014] Three-layer attribute labels are generated based on a multi-source data fusion matrix.

[0015] As a preferred embodiment of the GIS resource management method described in this invention, the accessibility analysis of urban resources includes:

[0016] The entire city area is divided into grid units, and a spatial indexing system is established.

[0017] Extract the geospatial coordinates and attribute parameters of public service facilities from the multi-source data fusion matrix.

[0018] Based on the spatial gravity model, the accessibility index of urban grid units is calculated, and a global accessibility contour map is generated to identify resource-scarce and resource-redundant areas.

[0019] As a preferred embodiment of the GIS resource management method described in this invention, the calculation of the accessibility index of urban grid units based on the spatial gravity model includes:

[0020] The public service facility set is extracted from the multi-source data fusion matrix, the attraction value between each grid unit and each public service facility is obtained, and the average value of all attraction values ​​corresponding to the grid unit is processed to obtain the accessibility index of the urban grid unit.

[0021] The accessibility index is calculated by traversing all grid cells and public service facilities to generate a city-wide accessibility index matrix.

[0022] As a preferred embodiment of the GIS resource management method described in this invention, the multi-source data fusion matrix is ​​represented as follows:

[0023]

[0024] Among them, G geo This is a geospatial data matrix with dimensions N×3, where N is the number of spatial units, and the 3 dimensions include spatial coordinates and land use. iot For IoT time-series data tensors, the dimensions are N×T×K, where T is the number of time steps, K is the number of sensor types, and V is the number of time steps. vis Let F be the visual feature matrix, with dimensions N×F, where F is the dimension of the visual features extracted from keyframes of the street view video stream. Represents a spatial connectivity operator, implementing G based on spatial coordinates. geo With T iot Location association, This indicates a feature concatenation operation, which combines the spatial concatenation result with V. vis The vectors are concatenated to form a joint feature vector.

[0025] The three-layer attribute label includes an L1 spatial layer containing road topology and building outline data, an L2 temporal layer containing sensor index data, and an L3 visual layer containing street view feature data.

[0026] The construction of a dynamic prediction model through machine learning includes,

[0027] Spatiotemporal feature datasets are extracted from the multi-source data fusion matrix and then standardized preprocessed.

[0028] The spatiotemporal feature dataset includes road topology features of the geospatial data matrix, traffic flow, PM2.5 values ​​and energy consumption data features of the IoT time-series data tensor, and dynamic texture features of the visual feature matrix.

[0029] Using road topology as the graph edge connection relationship and grid cells as graph nodes, a spatiotemporal graph convolutional network is constructed. The node features include traffic flow, PM2.5 values, energy consumption data, and street scene features extracted by CNN.

[0030] The spatiotemporal graph convolutional network includes spatial convolutional layers and temporal convolutional layers, and is superimposed with an attention mechanism.

[0031] Historical fusion datasets are extracted from the database for model training. Traffic state labels and multiple types of resource demand values ​​for a future time period are used as supervision signals. The cross-entropy loss function is used to optimize the traffic congestion probability classification task, while the mean squared error loss function is used to optimize the resource demand value regression task. A joint loss function is constructed, and the model is iteratively trained to convergence using the Adam optimizer.

[0032] Load the trained dynamic prediction model, input the real-time spatiotemporal feature dataset, and output traffic congestion probability distribution maps for multiple future time slices, as well as heat maps of various types of resource demand.

[0033] This invention achieves accurate and high-granular prediction of urban traffic conditions and resource demand by constructing a spatiotemporal graph convolutional network and fusing multi-source heterogeneous data. Specifically, it utilizes graph structures to deeply mine the spatiotemporal dependencies between road topology, time-series indicators, and street scene features. Through multi-task joint learning, it simultaneously outputs the congestion probability distribution and resource demand heatmap for future multi-time slices, thereby providing spatially explicit and real-time forward-looking decision-making basis for the dynamic allocation of urban resources.

[0034] As a preferred embodiment of the GIS resource management method described in this invention, the triggering model iteration mechanism includes:

[0035] New data from the multi-source data fusion matrix is ​​acquired in real time to construct a real-time updated dataset. The real-time updated dataset includes granular sampling values ​​of IoT time-series data tensors at set time intervals, newly added keyframe features of the visual feature matrix, and dynamic corrections of geospatial data.

[0036] The residuals between the traffic congestion probability and resource demand values ​​output by the dynamic prediction model and the real-time observation values ​​for the corresponding time period are calculated periodically. Based on the residual data, spatial residual heatmaps and time-series residual curves are generated, and the mean and standard deviation of the residuals across the entire domain are calculated. The residual is equal to the real-time observation value minus the model prediction value.

[0037] The residual preset threshold is obtained from the database, and the iteration mechanism is triggered when the mean residual exceeds the preset threshold.

[0038] By acquiring multi-source fusion data in real time and building an updated dataset, the residual between predicted values ​​and actual observed values ​​can be continuously monitored, and the residual can be visualized and located using spatial heat maps and time-series curves. When the mean residual exceeds the threshold, the model iteration is automatically triggered, enabling the dynamic prediction model to have the ability to learn online and optimize adaptively, thereby improving its response speed and long-term prediction accuracy to the dynamic changes of complex urban systems.

[0039] As a preferred embodiment of the GIS resource management method described in this invention, the generation of a resource health index based on the model iteration mechanism includes:

[0040] The basic parameters of various types of resources are extracted from the multi-source data fusion matrix. These basic parameters include the facility capacity C of each type. k Real-time load rate λ k Where k represents the resource type, k = 1, 2, ..., n, and the accessibility index A of the urban grid unit m is obtained. m The resource health index is generated based on the grid cell and is represented as follows:

[0041]

[0042] Among them, w k Let λ be the weight of the k-th type of resource. m,k Let θ be the real-time load rate of the k-th type of resource within grid cell m, θ be the reachability threshold, σ be the smoothing factor, and n be the total number of resource types.

[0043] A spatial distribution map of resource health is generated based on the resource health index. The spatial resolution of the resource health spatial distribution map is consistent with the raster granularity of the multi-source data fusion matrix.

[0044] The triggering resource allocation control signal includes

[0045] Retrieve the preset set of resource health alarm thresholds {τ1,τ2} from the database.

[0046] Where τ1 is the mild alarm threshold and τ2 is the severe alarm threshold, the resource health spatial distribution map generated in real time is monitored, and when the health index H of grid cell m is identified... m When <τ1, an alarm condition is triggered. The geospatial coordinates of the grid cell are extracted through spatial coordinate encoding and associated with the resource type label in the multi-source data fusion matrix.

[0047] Based on the health index H m The relationship with alarm thresholds determines the implementation of tiered responses:

[0048] When τ2≤H m When <τ1, a mild alarm response is triggered, generating a neighborhood resource scheduling instruction to call up idle facilities of the corresponding resource type within the same administrative region.

[0049] When H m When the value is less than τ2, a severe alarm response is triggered, generating a cross-regional emergency dispatch command to call upon idle facilities of the corresponding resource type within the same administrative region.

[0050] This invention achieves refined, quantitative, and spatial monitoring of the status of various urban resources by calculating the resource health index and constructing its spatial distribution map. Specifically, it combines real-time load rate and spatial accessibility to dynamically assess the resource health level of each grid unit, and automatically triggers differentiated allocation instructions based on a hierarchical alarm mechanism, thereby improving the accuracy of identifying urban resource shortages and redundancies, the response speed, and the spatial targeting of allocation strategies.

[0051] This invention provides a system for GIS resource management.

[0052] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a system for GIS resource management, comprising: an urban data acquisition module, an urban data analysis module, and an urban data processing module;

[0053] The urban data acquisition module is used to obtain urban planning maps from the urban service platform, construct a three-dimensional urban model based on the urban planning maps, and collect multi-source urban feature information in real time through urban public data acquisition terminals.

[0054] The urban data analysis module is used to construct a multi-source data fusion matrix of a GIS map and generate three-layer attribute labels, perform accessibility analysis of urban resources based on a spatial gravity model and identify resource-scarce and resource-redundant areas, construct a dynamic prediction model through machine learning and trigger a model iteration mechanism.

[0055] The city data processing module generates a resource health index based on the model iteration mechanism, triggering a resource allocation control signal.

[0056] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the GIS resource management method.

[0057] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the GIS resource management method.

[0058] The beneficial effects of this invention are as follows: By constructing a multi-source data fusion matrix, precise spatiotemporal correlation of geospatial, IoT temporal, and visual feature data is achieved, improving the completeness and accuracy of data integration; based on the accessibility analysis of the spatial gravity model and the dynamic prediction model of the spatiotemporal graph convolutional network, quantitative assessment and forward-looking prediction of resource supply and demand are realized, providing a scientific basis for resource allocation; the combination of resource health index and differentiated allocation instructions enables dynamic optimization and precise scheduling of resources, effectively alleviating resource shortage and redundancy problems, and improving the refinement and intelligence of urban resource management. Attached Figure Description

[0059] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0060] Figure 1 This is a general flowchart of a GIS resource management method provided in one embodiment of the present invention.

[0061] Figure 2 This is an overall framework diagram of a GIS resource management system provided in one embodiment of the present invention. Detailed Implementation

[0062] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0063] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a GIS resource management method, including:

[0064] S1. Obtain city planning maps, construct a 3D city model based on the city planning maps, and collect multi-source city feature information through a multi-source data acquisition system.

[0065] S2. Based on the city's 3D model and multi-source city feature information, construct a multi-source data fusion matrix for a single GIS map and generate three-layer attribute labels.

[0066] S3. Construct a spatial gravity model based on a multi-source data fusion matrix to conduct accessibility analysis of urban resources.

[0067] S4. Build a dynamic prediction model through machine learning and trigger the model iteration mechanism.

[0068] S5. Generate a resource health index based on the model iteration mechanism results, and trigger resource allocation control signals.

[0069] This invention achieves deep integration and unified management of geospatial, IoT temporal, and visual feature data by constructing a three-dimensional urban model and a multi-source data fusion matrix; it improves the accuracy and real-time performance of urban resource accessibility analysis and demand forecasting based on a spatial gravity model and machine learning dynamic prediction; and it achieves accurate identification and graded response to resource-scarce and redundant areas through a resource health index and automatic allocation mechanism, thereby comprehensively improving the intelligence level and dynamic adaptability of urban resource management.

[0070] Example 2, an embodiment of the present invention, provides a GIS resource management method based on the previous embodiment, including:

[0071] In the embodiments of this application, the multi-source data acquisition system in S1, namely the urban public data acquisition terminal, acquires high-precision urban images through regular aerial photography by a cluster of oblique photogrammetry drones, and generates road topology, building outline vectors, and land use labels by combining airborne lidar point cloud data; it also connects to the CAD / BIM vector database of the urban planning department to achieve dynamic data verification and updating.

[0072] Vehicle flow density data is collected at a frequency of 2Hz using microwave traffic flow detectors buried in the main road.

[0073] Deploy an environmental monitoring sensor network based on the LoRaWAN protocol to upload PM2.5, temperature, humidity and other indicators every 5 minutes.

[0074] Energy consumption measurement values ​​are transmitted in real time at 15-minute intervals via the NB-IoT communication module of the smart electricity / water meter.

[0075] 4K high-definition cameras equipped with edge computing units are deployed on city light poles / traffic light poles to transmit real-time street view video streams via 5G networks.

[0076] A lightweight YOLOv7 model is run on an edge GPU device to extract dynamic features of traffic participants in real time and generate structured metadata.

[0077] The use of drones for mobile aerial photography supplements the collection of multi-angle, time-series street view images of key areas.

[0078] All collected data are embedded with UTC timestamps and spatial coordinate metadata, and benchmark alignment of multi-source data is achieved through a spatiotemporal unified encoder.

[0079] In one alternative implementation, the multi-source data acquisition system can be a three-dimensional perception network based on a city information model, relying on smart light poles to integrate multi-functional sensing terminals, detecting traffic flow status through pole-mounted millimeter-wave radar, collecting PM2.5 data using built-in gaseous pollutant sensors, and transmitting the data to edge computing nodes via a dedicated fiber optic network.

[0080] By deploying photovoltaic integrated sensor nodes on the building facade, energy consumption data and microclimate parameters are monitored simultaneously, and long-term unattended operation is achieved by using solar self-powered power and LoRa wireless transmission.

[0081] By using a formation of drones equipped with multispectral cameras, data on changes in urban land cover are collected periodically along a preset route, and the image data is transmitted and stitched together in real time via an aerial base station.

[0082] In another alternative implementation, the multi-source data acquisition system can also be a cloud-edge-device collaborative architecture oriented towards comprehensive perception, including:

[0083] Mobile data acquisition modules are installed on the roof of municipal vehicles. By combining an inertial navigation system with a 360° surround-view camera, the status of road facilities and street view video streams are dynamically captured. Real-time synchronization of vehicle-to-cloud data is achieved with the help of V2X communication technology.

[0084] By connecting to a city-level IoT platform, real-time data streams from public facilities such as smart meters, water pressure sensors, and fire hydrant status monitoring terminals are aggregated.

[0085] By utilizing satellite remote sensing data interfaces to acquire periodic macroscopic information on surface temperature distribution and vegetation cover changes, and combining this with ground-based fixed sensors, we can achieve air-space-ground data calibration and fusion.

[0086] Furthermore, in step S1, the multi-source urban feature information includes geospatial data, IoT time-series data, and visual feature data.

[0087] Geospatial data includes road topology, building outline vectors, and land use labels.

[0088] IoT time-series data includes traffic flow sensor readings, PM2.5 values ​​collected by environmental monitoring equipment, and energy consumption measurements.

[0089] Visual feature data includes real-time street view video streams captured by cameras deployed at the edge.

[0090] Furthermore, in step S2, constructing the multi-source data fusion matrix of a single GIS map includes the following steps A1-A4:

[0091] A1. Preprocess the multi-source urban feature information to form a spatial base layer.

[0092] Furthermore, preprocessing involves spatial alignment of road topology and building outline vectors through coordinate system transformation, and overlaying land use labels.

[0093] A2. Based on multi-source urban feature information, a rasterized time series layer is generated using a unified time base.

[0094] Furthermore, adopting a unified time base requires sampling to a set time interval with a unified time granularity in order to generate a rasterized time series layer through spatial interpolation.

[0095] A3. Based on multi-source urban feature information, perform spatiotemporal pose analysis and frame sampling, and map the analysis results to the coordinate system of the urban 3D model to generate a dynamic texture layer.

[0096] Furthermore, the spatiotemporal pose analysis and frame sampling specifically target street view video streams.

[0097] A4. Generate a multi-source data fusion matrix based on the spatial base layer, the rasterized temporal layer, and the dynamic texture layer.

[0098] A5. Generate three-layer attribute labels based on multi-source data fusion matrix.

[0099] Furthermore, the three-layer attribute label includes an L1 spatial layer containing road topology and building outline data, an L2 temporal layer containing sensor index data, and an L3 visual layer containing street view feature data.

[0100] In the embodiments of this application, A2 is based on a unified time reference, meaning that all sensing terminals (including traffic flow sensors, environmental monitoring equipment, smart meters, etc.) must embed a UTC timestamp conforming to the ISO 8601 standard when collecting data.

[0101] Normalization processing is performed on the original asynchronous time-series data stream at a set time interval (e.g., 5 minutes). For high-frequency data such as traffic flow, a time window aggregation algorithm (e.g., summing or averaging within a 5-minute window) is used. For continuous monitoring data such as PM2.5, linear interpolation is used to fill in missing time values, and then sampling is performed at whole time points. For cumulative values ​​such as energy consumption data, time window difference calculation is used to obtain the actual consumption within each granular interval. For data that needs to be displayed locally, the display time is dynamically converted according to the time zone of the end user while maintaining the UTC storage baseline, ensuring the separation of original data processing and display logic.

[0102] The time series anomaly detection algorithm identifies and removes abnormal timestamp data that is obviously ahead or behind, thus avoiding data pollution caused by device clock failure.

[0103] All data acquisition nodes synchronize with the reference time server at least once every 24 hours via the NTP protocol to ensure that the local clock deviates from UTC by less than 50 milliseconds.

[0104] In one alternative implementation, the unified time base can be a Unix timestamp (EpochTime), which stores and calculates timestamps in integer form, suitable for efficient time synchronization and cross-platform data processing in distributed systems.

[0105] In another alternative implementation, a unified time reference can also be based on a local time system synchronized via NTP (Network Time Protocol). All sensing terminals and servers periodically synchronize with the reference time server, keeping the local clock deviation within milliseconds, and using the synchronized local timestamp as the data time reference. This is suitable for metropolitan area sensing scenarios with extremely high real-time requirements and stable network conditions.

[0106] In the embodiments of this application, the spatiotemporal pose analysis and frame sampling in A4 utilize SLAM technology to perform inter-frame pose estimation on the street view video stream, extract keyframes, and map them to the coordinate system of the urban 3D model. Specifically, this includes: extracting ORB feature points in the video frames using the ORB-SLAM3 framework, calculating the essential matrix between consecutive frames, and decomposing it to obtain the camera pose change; simultaneously, filtering keyframes with information entropy higher than a threshold based on the common view, and mapping the camera coordinate system of the keyframes to the global UTM coordinate system of the urban 3D model using the PnP algorithm.

[0107] In one alternative implementation, spatiotemporal pose resolution and frame sampling can be a fusion localization scheme based on visual inertial odometry (VIO): through hardware synchronization of the camera and IMU, the VINS-Fusion algorithm is used to fuse visual feature points and IMU pre-integrated data, and jointly optimize to obtain a smooth and scale-consistent camera motion trajectory; key frame extraction is adaptively triggered according to changes in accelerometer and angular velocity readings.

[0108] In another alternative implementation, spatiotemporal pose resolution and frame sampling can also be a semantic SLAM method that combines a prior map: a 3D city model is preloaded as a prior map, camera motion is calculated directly based on image pixel brightness using DSO technology, and semantic elements such as road signs and streetlights are detected through a YOLOv5 instance segmentation network. The semantic elements are then matched with the model coordinates in the prior map through graph optimization.

[0109] This invention achieves precise spatiotemporal alignment and deep fusion of heterogeneous multi-source data by constructing a multi-source data fusion matrix: Based on a unified time base (UTC timestamp / NTP synchronization) and a multi-mode resampling strategy, it solves the problems of asynchronous data and clock deviation from heterogeneous terminals, ensuring millisecond-level alignment of time-series data; through visual SLAM technology and coordinate system transformation, it achieves high-precision mapping of street view video streams to urban 3D models, generating dynamic texture layers; combined with spatial interpolation and a three-layer attribute labeling system, it uniformly encodes traffic flow, environmental monitoring, energy consumption data, and geographic information into a machine-understandable fusion matrix, improving the accuracy, robustness, and multi-granularity analysis capabilities of urban dynamic perception.

[0110] Furthermore, step S3, which involves an accessibility analysis of urban resources, includes the following steps B1-B3:

[0111] B1. Divide the entire city into grid units and establish a spatial indexing system.

[0112] B2. Extract the geospatial coordinates and attribute parameters of public service facilities from the multi-source data fusion matrix.

[0113] Furthermore, the attribute parameters include facility capacity C. j Service Level L j .

[0114] B3. Calculate the accessibility index of urban grid units based on the spatial gravity model, generate a global accessibility contour map, and identify resource-scarce and resource-redundant areas.

[0115] Furthermore, the spatial gravity model is expressed as follows:

[0116]

[0117] Where m is the grid cell index, F mj Let M represent the gravitational force between grid cell m and public service facility j, where G is the gravitational constant and M is the gravitational force. j For the quality parameters of public service facility j, the facility capacity C is... j Multiply by the gravity value corresponding to the unit facility capacity stored in the database, and then apply the service level L. j Multiply the gravitational value corresponding to the service level stored in the database, and add the two products to obtain the mass parameter, P. m Let T be the population size of grid cell m. mj Let β be the shortest time distance from grid cell m to public service facility j, and let T be the shortest time distance. mj The distance attenuation coefficient.

[0118] Furthermore, step B3, which calculates the accessibility index of urban grid units based on the spatial gravity model, includes the following steps B31-B32:

[0119] B31. Extract the set of public service facilities from the multi-source data fusion matrix, obtain the attraction value between each grid unit and each public service facility, and average all the attraction values ​​corresponding to the grid unit to obtain the accessibility index of the urban grid unit.

[0120] Furthermore, the set of public service facilities is represented as {F}. j |j=1,2,...,J}.

[0121] B32. Traverse all grid cells and public service facilities to perform accessibility index calculations and generate a city-wide accessibility index matrix.

[0122] Furthermore, given grid cells m∈[1,M], public service facilities j∈[1,J], and the accessibility index matrix is... M represents the total number of urban grid units.

[0123] Furthermore, the multi-source data fusion matrix is ​​represented as follows:

[0124]

[0125] Among them, G geo This is a geospatial data matrix with dimensions N×3, where N is the number of spatial units, and the 3 dimensions include spatial coordinates and land use. iot For IoT time-series data tensors, the dimensions are N×T×K, where T is the number of time steps, K is the number of sensor types, and V is the number of time steps. vis Let F be the visual feature matrix, with dimensions N×F, where F is the dimension of the visual features extracted from keyframes of the street view video stream. Represents a spatial connectivity operator, implementing G based on spatial coordinates. geo With T iot Location association, This indicates a feature concatenation operation, which combines the spatial concatenation result with V. vis The vectors are concatenated to form a joint feature vector.

[0126] Furthermore, the construction of the dynamic prediction model through machine learning in step S4 includes the following steps C1-C4:

[0127] C1. Extract the spatiotemporal feature dataset from the multi-source data fusion matrix and perform standardization preprocessing;

[0128] The spatiotemporal feature dataset includes road topology features of the geospatial data matrix, traffic flow, PM2.5 values ​​and energy consumption data features of the IoT time series data tensor, and dynamic texture features of the visual feature matrix.

[0129] C2. Using the road topology as the graph edge connection relationship and the grid cell as the graph node, a spatiotemporal graph convolutional network is constructed. The node features include traffic flow, PM2.5 value, energy consumption data and street scene features extracted by CNN.

[0130] Spatiotemporal graph convolutional networks contain spatial and temporal convolutional layers, and are superimposed with an attention mechanism;

[0131] C3. Extract historical fusion datasets from the database for model training. Use traffic state labels and multi-type resource demand values ​​for a future time period as supervision signals. Use cross-entropy loss function to optimize the traffic congestion probability classification task, and mean squared error loss function to optimize the resource demand value regression task. Construct a joint loss function and iteratively train it to convergence using the Adam optimizer.

[0132] Furthermore, taking medical resources as an example, the IoT time-series data for medical resources includes traffic flow around hospitals (reflecting the efficiency of access to medical care), regional PM2.5 levels (related to health needs), and energy consumption data of medical facilities (such as air conditioning and equipment electricity consumption, reflecting operational status); visual feature data for medical resources includes features such as ambulance recognition and queuing crowd density extracted from hospital street view videos (assisting in judging the pressure of medical demand); geospatial data for medical resources includes hospital building outlines and road topology (related to accessibility to medical care); taking educational resources as an example, the IoT time-series data for educational resources includes traffic flow around schools (reflecting the trend of congestion when parents pick up and drop off their children), regional PM2.5 levels (related to restrictions on students' outdoor activities), and school energy consumption data (such as electricity consumption during after-school service periods, reflecting the intensity of classroom use); visual feature data for educational resources includes the density of parents gathering and the number of school buses identified extracted from school street view videos (reflecting the pressure of students returning home in a concentrated manner); geospatial data for educational resources includes school building outlines and surrounding road topology (affecting traffic efficiency).

[0133] Furthermore, taking medical resources as an example, the demand value for medical resources is a quantitative indicator derived from traffic flow, PM2.5 levels, and energy consumption data in a comprehensive medical scenario. Traffic flow: The average traffic flow within a 1km radius of the hospital during morning peak hours (7:00-9:00) and evening peak hours (17:00-19:00) is statistically analyzed. A baseline traffic flow of 2000 vehicles / hour is set. If the actual traffic flow exceeds the baseline, the excess is converted into an incremental demand value using an increment coefficient t1. The incremental demand value = max(0, actual traffic flow - baseline traffic flow) × t1. This increment reflects the increased risk of travel time to medical facilities, corresponding to increased resource demand for ambulance response and emergency room dispatch. For example, if the traffic flow is 2500 vehicles / hour, the increment is (2500-2000) × t1. PM2.5 levels: The real-time PM2.5 monitoring value of the grid unit where the hospital is located is taken, with a baseline PM2.5 value set at 150 μg / m³.3 If the actual PM2.5 exceeds the PM2.5 baseline value, the excess is converted into an incremental demand value using an increment coefficient t2. The incremental demand value is calculated as: max(0, actual PM2.5 value - baseline PM2.5 value) × t2. This increment is associated with an increase in residents' demand for respiratory disease treatment, leading to a higher demand for emergency medical services and respiratory resources. For example, if PM2.5 is 200 μg / m³... 3 When the energy consumption exceeds the benchmark, the increment is (200-150)×t2; Energy consumption data: compare the average energy consumption of medical facilities in the same period of the past 30 days, and set the benchmark excess ratio as 20%. If the current energy consumption excess ratio exceeds the benchmark value, the excess ratio is converted into the demand value increment by the increment coefficient t3. The demand value increment = max(0, current energy consumption excess ratio - benchmark excess ratio)×t3. This increment reflects the increased operating load of equipment and manpower, and the corresponding increase in the demand value for medical equipment allocation and medical staff replenishment. For example, when the energy consumption exceeds the benchmark by 30%, the increment is (30%-20%)×t3.

[0134] Furthermore, taking educational resources as an example, the demand for educational resources is determined based on traffic flow, PM2.5 levels, and energy consumption data within the educational setting. Traffic flow: The average traffic flow within a 500m radius of the school during school hours (7:30-8:30 AM and 4:30-5:30 PM) is statistically analyzed. A baseline traffic flow of 1500 vehicles / hour is set. If the actual traffic flow exceeds the baseline, the excess is converted into an incremental demand value using an increment coefficient t4. The incremental demand value is calculated as max(0, actual traffic flow - baseline traffic flow) × t4. This increment reflects the increased risk of congestion for parents picking up and dropping off children and students commuting, corresponding to an increased demand for traffic management personnel and school bus dispatch frequency. For example, if the traffic flow is 2000 vehicles / hour, the increment is (2000-1500) × t4. PM2.5 levels: The real-time PM2.5 monitoring value of the grid unit where the school is located is taken, with a baseline PM2.5 value set at 100 μg / m³. 3 If the actual PM2.5 exceeds the PM2.5 baseline value, the excess portion is converted into an incremental demand value using an increment factor t5. The incremental demand value is calculated as: max(0, actual PM2.5 value - baseline PM2.5 value) × t5. This increment reflects the increased demand for after-school care facilities and indoor activity equipment due to restrictions on students' outdoor activities. For example, if PM2.5 is 150 μg / m³... 3When the energy consumption exceeds the benchmark, the increment is (150-100)×t5; Energy consumption data: Compared with the school's average energy consumption in the same period over the past 30 days, the benchmark excess ratio is set at 15%. If the current energy consumption excess ratio exceeds the benchmark value, the excess ratio is converted into the demand value increment by the increment coefficient t6. Demand value increment = max(0, current energy consumption excess ratio - benchmark excess ratio)×t6. This increment reflects the increase in classroom usage frequency and teaching equipment operating load, corresponding to the increase in the demand value of teaching site maintenance manpower and equipment replenishment. For example, when the energy consumption exceeds the benchmark by 25%, the increment is (25%-15%)×t6.

[0135] It should be noted that the cross-entropy loss function is used for classification tasks, optimizing the traffic congestion probability output by the model; while the mean squared error loss function is used for regression tasks, optimizing the resource demand value.

[0136] C4. Load the trained dynamic prediction model, input the real-time spatiotemporal feature dataset, and output the traffic congestion probability distribution map for multiple future time slices, as well as the heat map of various types of resource demand.

[0137] Furthermore, the model iteration mechanism triggered in step S4 includes the following steps D1-D3:

[0138] D1. Real-time acquisition of new data from the multi-source data fusion matrix, construction of a real-time updated dataset. The real-time updated dataset includes granular sampling values ​​of IoT time-series data tensors at set time intervals, newly added keyframe features of the visual feature matrix, and dynamic corrections to geospatial data.

[0139] D2. Periodically calculate the residuals between the traffic congestion probability and resource demand values ​​output by the dynamic prediction model and the real-time observation values ​​for the corresponding time period. Generate spatial residual heatmaps and time-series residual curves based on the residual data, and calculate the mean and standard deviation of the residuals across the entire domain. The residual is equal to the real-time observation value minus the model prediction value.

[0140] D3. Obtain the preset threshold for residuals from the database. When the mean residual exceeds the preset threshold, trigger the iteration mechanism.

[0141] This invention constructs a dynamic prediction model through machine learning and triggers an iterative mechanism, thereby achieving accurate prediction of the multi-dimensional state of the city and model self-optimization.

[0142] By fusing road topology, traffic flow, environmental indicators, energy consumption, and street view visual features using a spatiotemporal graph convolutional network, the system deeply mines spatiotemporal dependencies and outputs traffic congestion probability distribution and heatmaps of various resource demands, significantly improving prediction accuracy and granularity. Through multi-task joint learning and attention mechanisms, the system enhances its ability to represent sudden events. Combining residual monitoring and automated iteration mechanisms, the system uses spatial residual heatmaps and temporal curves to locate deviations. When the prediction error exceeds a threshold, the system triggers a model update, ensuring that the system continuously adapts to dynamic urban changes and effectively avoids model aging issues.

[0143] In this embodiment, the triggering resource allocation control signal in S5 is triggered by a communication protocol. Specifically, it extracts the spatial coordinate code and resource type label of the alarm grid unit, encapsulates it into a structured alarm message conforming to the ISO 17850 standard, matches the unique equipment identifier of the target facility according to the resource type label, and issues instructions to the resource scheduling executor through the lightweight MQTT protocol. The receiving end transmits the instruction reception status back in real time through the ACK confirmation mechanism. If no confirmation is received, the index backoff retransmission strategy is triggered. The status changes during the resource allocation execution process are synchronized to the central database in real time through the CoAP protocol to update the resource health index calculation baseline.

[0144] In one alternative implementation, the resource allocation control signal can be triggered by a distributed collaborative triggering mechanism based on edge computing nodes: a lightweight rule engine is deployed on each edge node to monitor the resource health data of the jurisdiction in real time. When an alarm condition is detected, communication is directly carried out between edge nodes via the gRPC protocol to generate resource allocation instructions and send them to the execution terminals in the jurisdiction. At the same time, the instruction log is asynchronously reported to the municipal supervision platform, forming a response mode that combines distributed decision-making and centralized supervision.

[0145] In another alternative implementation, the resource allocation control signal can be triggered by a trusted triggering scheme based on blockchain smart contracts: the resource health alarm threshold and allocation rules are encoded into blockchain smart contracts. When the health index of a grid cell is detected to exceed the threshold, the consensus node automatically triggers the execution of the contract, generates an immutable resource allocation instruction record, and broadcasts it to all relevant facility nodes through the blockchain network.

[0146] Furthermore, the generation of the resource health index based on the model iteration mechanism in step S5 includes the following steps E1-E2:

[0147] E1. Extract basic parameters of various resource types from the multi-source data fusion matrix. The basic parameters include the facility capacity C for each type. k Real-time load rate λ k Where k represents the resource type, k = 1, 2, ..., n, and the accessibility index A of the urban grid unit m is obtained.m The resource health index is generated based on the grid cell and is represented as follows:

[0148]

[0149] Among them, w k Let λ be the weight of the k-th type of resource. m,k Let θ be the real-time load rate of the k-th type of resource within grid cell m, θ be the reachability threshold, σ be the smoothing factor, and n be the total number of resource types.

[0150] E2. Generate a spatial distribution map of resource health based on the resource health index. The spatial resolution of the resource health spatial distribution map is consistent with the raster granularity of the multi-source data fusion matrix.

[0151] Furthermore, triggering the resource allocation control signal in step S5 includes the following steps F1-F2:

[0152] F1. Retrieve the preset resource health alarm threshold set {τ1,τ2} from the database.

[0153] Where τ1 is the mild alarm threshold and τ2 is the severe alarm threshold, the resource health spatial distribution map generated in real time is monitored, and when the health index H of grid cell m is identified... m When <τ1, an alarm condition is triggered. The geospatial coordinates of the grid cell are extracted through spatial coordinate encoding and associated with the resource type label in the multi-source data fusion matrix.

[0154] F2, Based on Health Index H m The relationship with alarm thresholds determines the implementation of tiered responses:

[0155] When τ2≤H m When <τ1, a mild alarm response is triggered, generating a neighborhood resource scheduling instruction to call up idle facilities of the corresponding resource type within the same administrative region.

[0156] When H m When the value is less than τ2, a severe alarm response is triggered, generating a cross-regional emergency dispatch command to call upon idle facilities of the corresponding resource type within the same administrative region.

[0157] This invention achieves refined monitoring and adaptive regulation of urban resource status by generating a resource health index and triggering resource allocation control signals.

[0158] Based on the dynamic calculation of the gridded health index using a multi-source data fusion matrix, the system comprehensively considers resource load rate, spatial accessibility, and multiple types of weights to accurately identify resource shortage and redundant areas. Through a hierarchical alarm mechanism and standardized communication protocols, it achieves automated decision-making from neighborhood scheduling to cross-regional emergency response, ensuring reliable command transmission and real-time status synchronization. Combined with optional solutions such as edge computing and blockchain smart contracts, it further improves the system's response speed, distributed collaboration capabilities, and operational reliability.

[0159] Example 3, referring to Figure 2 This is one embodiment of the present invention, which provides a system for a GIS resource management method, including: an urban data acquisition module, an urban data analysis module, and an urban data processing module.

[0160] The city data acquisition module is used to obtain city planning maps from the city service platform, construct a 3D city model based on the city planning maps, and collect multi-source city feature information in real time through the city public data acquisition terminal.

[0161] The urban data analysis module is used to construct a multi-source data fusion matrix for a single GIS map and generate three-layer attribute labels. Based on the spatial gravity model, it performs accessibility analysis on urban resources and identifies resource-scarce and resource-redundant areas. It also constructs a dynamic prediction model through machine learning and triggers a model iteration mechanism.

[0162] The city data processing module generates a resource health index based on the model iteration mechanism, triggering resource allocation control signals.

[0163] This embodiment also provides an electronic device applicable to a GIS resource management method, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement a GIS resource management method as proposed in the above embodiment.

[0164] This embodiment also provides a storage medium on which a computer program is stored, which, when executed by a processor, implements a GIS resource management method as proposed in the above embodiments.

[0165] The storage medium proposed in this embodiment belongs to the same inventive concept as the GIS resource management method proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0166] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0167] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A GIS resource management method, characterized in that: include, Obtain urban planning maps, construct a 3D urban model based on the urban planning maps, and collect multi-source urban feature information through a multi-source data acquisition system; Based on the city's 3D model and multi-source city feature information, a multi-source data fusion matrix for a single GIS map is constructed, and three-layer attribute labels are generated. A spatial gravity model is constructed based on a multi-source data fusion matrix to conduct accessibility analysis of urban resources; Build dynamic prediction models through machine learning and trigger model iteration mechanisms; The resource health index is generated based on the model iteration mechanism, which triggers resource allocation control signals.

2. The GIS resource management method as described in claim 1, characterized in that: The construction of a multi-source data fusion matrix for a single GIS map includes... Multi-source urban feature information is preprocessed to form a spatial base layer; Based on multi-source urban feature information, a unified time base is used to generate a rasterized time series layer; Based on multi-source urban feature information, spatiotemporal pose analysis and frame sampling are performed, and the analysis results are mapped to the coordinate system of the urban 3D model to generate a dynamic texture layer. A multi-source data fusion matrix is ​​generated based on a spatial base layer, a rasterized temporal layer, and a dynamic texture layer; Three-layer attribute labels are generated based on a multi-source data fusion matrix.

3. The GIS resource management method as described in claim 2, characterized in that: The accessibility analysis of urban resources includes, The entire city area is divided into grid units, and a spatial indexing system is established; Extract the geospatial coordinates and attribute parameters of public service facilities from the multi-source data fusion matrix; Based on the spatial gravity model, the accessibility index of urban grid units is calculated, and a global accessibility contour map is generated to identify resource-scarce and resource-redundant areas.

4. The GIS resource management method as described in claim 3, characterized in that: The accessibility index for urban grid units calculated based on the spatial gravity model includes, The public service facility set is extracted from the multi-source data fusion matrix, the attraction value between each grid unit and each public service facility is obtained, and the average value of all attraction values ​​corresponding to the grid unit is processed to obtain the accessibility index of the urban grid unit. The accessibility index is calculated by traversing all grid cells and public service facilities to generate a city-wide accessibility index matrix.

5. A GIS resource management method as described in claim 4, characterized in that: The multi-source data fusion matrix is ​​represented as follows: Among them, G geo This is a geospatial data matrix with dimensions N×3, where N is the number of spatial units, and the 3 dimensions include spatial coordinates and land use. iot For IoT time-series data tensors, the dimensions are N×T×K, where T is the number of time steps, K is the number of sensor types, and V is the number of time steps. vis Let F be the visual feature matrix, with dimensions N×F, where F is the dimension of the visual features extracted from keyframes of the street view video stream. Represents a spatial connectivity operator, implementing G based on spatial coordinates. geo With T iot Location association, This indicates a feature concatenation operation, which combines the spatial concatenation result with V. vis Concatenate them into a joint feature vector; The three-layer attribute label includes an L1 spatial layer containing road topology and building outline data, an L2 temporal layer containing sensor index data, and an L3 visual layer containing street view feature data. The construction of a dynamic prediction model through machine learning includes, Spatiotemporal feature datasets are extracted from multi-source data fusion matrices and then standardized preprocessed. The spatiotemporal feature dataset includes road topology features of the geospatial data matrix, traffic flow, PM2.5 values ​​and energy consumption data features of the IoT time series data tensor, and dynamic texture features of the visual feature matrix. Using road topology as the graph edge connection relationship and grid cells as graph nodes, a spatiotemporal graph convolutional network is constructed. The node features include traffic flow, PM2.5 values, energy consumption data, and street scene features extracted by CNN. The spatiotemporal graph convolutional network includes spatial convolutional layers and temporal convolutional layers, and is superimposed with an attention mechanism; Historical fusion datasets are extracted from the database for model training. Traffic state labels and multiple types of resource demand values ​​for a future time period are used as supervision signals. The cross-entropy loss function is used to optimize the traffic congestion probability classification task, while the mean squared error loss function is used to optimize the resource demand value regression task. A joint loss function is constructed and iteratively trained to convergence using the Adam optimizer. Load the trained dynamic prediction model, input the real-time spatiotemporal feature dataset, and output traffic congestion probability distribution maps for multiple future time slices, as well as heat maps of various types of resource demand.

6. The GIS resource management method as described in claim 5, characterized in that: The triggering model iteration mechanism includes, New data in the multi-source data fusion matrix is ​​acquired in real time to construct a real-time updated dataset. The real-time updated dataset includes granular sampling values ​​of IoT time series data tensors at a set time interval, newly added keyframe features of visual feature matrices, and dynamic corrections of geospatial data. The residuals between the traffic congestion probability and resource demand values ​​output by the dynamic prediction model and the real-time observation values ​​for the corresponding time period are calculated periodically. Based on the residual data, spatial residual heat maps and time-series residual curves are generated, and the mean and standard deviation of the residuals across the entire domain are calculated. The residual is equal to the real-time observation value minus the model prediction value. The residual preset threshold is obtained from the database, and the iteration mechanism is triggered when the mean residual exceeds the preset threshold.

7. A GIS resource management method as described in claim 6, characterized in that: The resource health index generated based on the model iteration mechanism includes: The basic parameters of various types of resources are extracted from the multi-source data fusion matrix. These basic parameters include the facility capacity C of each type. k Real-time load rate λ k Where k represents the resource type, k = 1, 2, ..., n, and the accessibility index A of the urban grid unit m is obtained. m The resource health index is generated based on the grid cell and is represented as follows: Among them, w k Let λ be the weight of the k-th type of resource. m,k Let θ be the real-time load rate of the k-th type of resource within grid cell m, θ be the reachability threshold, σ be the smoothing factor, and n be the total number of resource types. A spatial distribution map of resource health is generated based on the resource health index. The spatial resolution of the spatial distribution map of resource health is consistent with the raster granularity of the multi-source data fusion matrix. The triggering resource allocation control signal includes Retrieve the preset set of resource health alarm thresholds {τ1,τ2} from the database; Where τ1 is the mild alarm threshold and τ2 is the severe alarm threshold, the resource health spatial distribution map generated in real time is monitored, and when the health index H of grid cell m is identified... m When <τ1, an alarm condition is triggered. The geospatial coordinates of the grid cell are extracted through spatial coordinate encoding and associated with the resource type label in the multi-source data fusion matrix. Based on the health index H m The relationship with alarm thresholds determines the implementation of tiered responses: When τ2≤H m When <τ1, a mild alarm response is triggered, generating a neighborhood resource scheduling instruction to call up idle facilities of the corresponding resource type within the same administrative region; When H m When the value is less than τ2, a severe alarm response is triggered, generating a cross-regional emergency dispatch command to call upon idle facilities of the corresponding resource type within the same administrative region.

8. A system for a GIS resource management method, employing the GIS resource management method as described in any one of claims 1 to 7, characterized in that, include: Urban data acquisition module, urban data analysis module, and urban data processing module; The urban data acquisition module is used to obtain urban planning maps from the urban service platform, construct a three-dimensional urban model based on the urban planning maps, and collect multi-source urban feature information in real time through urban public data acquisition terminals. The urban data analysis module is used to construct a multi-source data fusion matrix of a GIS map and generate three-layer attribute labels, perform accessibility analysis of urban resources based on a spatial gravity model and identify resource-scarce and resource-redundant areas, construct a dynamic prediction model through machine learning and trigger a model iteration mechanism. The city data processing module generates a resource health index based on the model iteration mechanism, triggering a resource allocation control signal.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the GIS resource management method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the GIS resource management method according to any one of claims 1 to 7.

Images