Urban rainstorm disaster dynamic risk assessment method and system based on multi-source data fusion

Abstract

The invention discloses an urban rainstorm disaster dynamic risk assessment method and system based on multi-source data fusion, and relates to the technical field of industrial data processing and big data, and the method comprises the steps: deploying an edge data network at a distributed source end, and determining urban dynamic data; constructing a simplified feature space, adding a time axial direction, and updating the simplified feature space; a multi-dimensional index system is introduced, a dynamic evaluation module is constructed, rainstorm disaster risk evaluation is executed, and a risk evaluation result is determined; executing interface display of the dynamic risk; the multi-dimensional index system comprises disaster-inducing factor danger, disaster-pregnancy environment exposure, disaster-bearing body vulnerability and disaster prevention and reduction capability. The technical problems that traditional urban rainstorm disaster risk assessment mainly depends on a single data source or a simple assessment model, multi-aspect factors are difficult to comprehensively consider and a multi-source data fusion effective mechanism is lacked are solved, accurate and comprehensive assessment of urban rainstorm disaster risks is achieved, and the urban rainstorm disaster risk assessment efficiency is improved. And scientific decision support is provided for urban emergency management.

Description

Technical Field

[0001] The present invention relates to the field of industrial data processing and big data technology, and in particular to a method and system for dynamic risk assessment of urban rainstorm disasters by integrating multi-source data. Background Art

[0002] Currently, traditional urban rainstorm disaster risk assessments primarily rely on single data sources or simple assessment models. For example, disaster risk assessments rely solely on rainfall data provided by meteorological departments, combined with simple empirical formulas. While these methods have been effective in the past, their limitations have become increasingly apparent as cities expand and their environments become more complex. Urban development has led to more complex population and infrastructure distribution, and a single data source cannot fully reflect the overall risk profile of urban rainstorm disasters. Furthermore, simple assessment models fail to comprehensively consider multiple factors, including hazard-causing factors, the hazard-prone environment, the hazard-bearing body, and disaster prevention and mitigation capabilities. This results in inaccurate and time-sensitive assessment results, making them unable to provide scientific and effective decision-making support for urban emergency management departments and unable to meet the actual needs of cities in responding to rainstorm disasters. Summary of the Invention

[0003] This application solves the technical problems that traditional urban rainstorm disaster risk assessment mainly relies on a single data source or a simple assessment model, is difficult to comprehensively consider multiple factors, and lacks an effective mechanism for multi-source data fusion. This application deploys an edge data network at the distributed source end, reconstructs and encapsulates multi-source data and identifies it with spatiotemporal codes to obtain urban dynamic data, constructs a simplified feature space and updates it, introduces a multi-dimensional indicator system combined with entropy weight combination to construct a dynamic assessment module for risk assessment, and feeds back the results to realize the interface display of dynamic risks, thereby accurately assessing the dynamic risks of urban rainstorm disasters, providing scientific decision-making support for urban emergency management departments, and making urban rainstorm disaster risk assessment more accurate and effective.

[0004] In response to the above technical problems, this application proposes a technical solution for a dynamic risk assessment method and system for urban rainstorm disasters based on multi-source data fusion.

[0005] In the first aspect, the present application provides a dynamic risk assessment method for urban rainstorm disasters based on multi-source data fusion, wherein the method includes: deploying an edge data network at a distributed source end, reconstructing, encapsulating and space-time coding the received multi-source data, and determining urban dynamic data, wherein the edge data network is opened at the sensor side of the communication thread; for the urban space, constructing a simplified feature space and adding a time axis, updating the simplified feature space according to the urban dynamic data; introducing a multidimensional indicator system, constructing a dynamic assessment module and deploying it in an embedded manner in the urban management system, and performing rainstorm disaster risk assessment under a multi-dimensional indicator cascade by interacting with the simplified feature space to determine the risk assessment results; feeding back the risk assessment results to the simplified feature space, and executing an interface display of dynamic risks; wherein the multidimensional indicator system includes the hazard of disaster-causing factors-exposure to disaster-prone environments-vulnerability of disaster-bearing bodies-disaster prevention and mitigation capabilities.

[0006] On the second aspect, the present application provides a dynamic risk assessment system for urban rainstorm disasters based on multi-source data fusion, wherein the system includes: an urban data determination module, which is used to deploy an edge data network at a distributed source end, reconstruct and encapsulate the received multi-source data, and identify the space-time code to determine the urban dynamic data, wherein the edge data network is opened at the sensor side of the communication thread; a feature space simplification module, which is used to construct a simplified feature space for the urban space and add a time axis, and update the simplified feature space according to the urban dynamic data; an assessment result determination module, which is used to introduce a multi-dimensional indicator system, construct a dynamic assessment module and embed it in the urban management system, and execute a rainstorm disaster risk assessment under a multi-dimensional indicator cascade by interacting with the simplified feature space to determine the risk assessment result; an interface display execution module, which is used to feed back the risk assessment result to the simplified feature space and execute an interface display of dynamic risk; an indicator system determination module, wherein the multi-dimensional indicator system includes the hazard of disaster-causing factors-exposure to disaster-prone environments-vulnerability of disaster-bearing bodies-disaster prevention and mitigation capabilities.

[0007] This application proposes one or more technical solutions, which have at least the following technical effects: This application collects multi-source data by deploying an edge data network at a distributed source end, and forms urban dynamic data through reconstruction, encapsulation and space-time code identification. Construct a simplified feature space and rely on the update of urban dynamic data, introduce a multi-dimensional indicator system, and use entropy weight combination empowerment to construct a dynamic assessment module to assess the risk of urban rainstorm disasters. Feedback the risk assessment results to the simplified feature space to realize the dynamic risk interface display. If multi-partition data processing is involved, the data interaction of the edge nodes of each partition will also be optimized to ensure the accuracy and real-time nature of the risk assessment, provide a scientific basis for urban rainstorm disaster prevention and control, achieve an accurate and comprehensive assessment of urban rainstorm disaster risks, and provide scientific decision-making support for urban emergency management.

[0008] The above content outlines the method and system for dynamic risk assessment of urban rainstorm disasters by multi-source data fusion in this application. This application will describe the steps of the technical solution in detail in the following specific implementation methods to facilitate technical personnel to have a clear and complete understanding of this application. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0010] Figure 1 It is a flow chart of the method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion provided in an embodiment of the present application.

[0011] Figure 2 It is a structural diagram of the urban rainstorm disaster dynamic risk assessment system based on multi-source data fusion provided in an embodiment of the present application.

[0012] Explanation of the accompanying symbols: city data determination module 1, feature space simplification module 2, evaluation result determination module 3, interface display execution module 4, indicator system determination module 5. DETAILED DESCRIPTION

[0013] This application acquires urban dynamic data by deploying an edge data network at a distributed source end, constructs a simplified feature space, and updates it. A multi-dimensional indicator system is introduced, and a dynamic assessment module is constructed using entropy weight combination empowerment to assess urban rainstorm disaster risks. The risk assessment results are fed back to the simplified feature space for interface display. If multi-partition data exists, data interaction between edge nodes is optimized, ultimately accurately presenting the dynamic risk of urban rainstorm disasters, achieving an accurate and comprehensive assessment of urban rainstorm disaster risks and providing scientific decision-making support for urban emergency management.

[0014] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only some of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0015] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0016] Example 1, as Figure 1 As shown, a dynamic risk assessment method for urban rainstorm disasters based on multi-source data fusion is provided, wherein the method includes: Step A100: deploy an edge data network at the distributed source end, reconstruct and encapsulate the received multi-source data and identify it with time and space codes to determine urban dynamic data, wherein the edge data network is opened at the sensor side of the communication thread.

[0017] In this embodiment, the distributed source is a system that divides a city into multiple zones based on its structure. The distributed multi-source sensors deployed in these zones are guided by the structural characteristics of the zones. The edge data network consists of an edge node deployed in each zone. Urban dynamic data is urban perception data generated by sensors monitoring urban elements.

[0018] Specifically, when deploying the edge data network at the distributed source end, first, divide the city into multiple partitions according to the city structure, deploy distributed multi-source sensors as distributed sources according to the characteristics of the partition structure, deploy edge nodes in each partition to form an edge data network and establish interactive communication. The specific steps are described in detail in A110-A130; then, the multi-source sensors monitor urban elements to determine the perception data, which is imported into the partition edge nodes through interactive communication, and the data features are extracted and classified and packaged into urban dynamic data, and then timestamps and location codes are introduced to mark them. The specific steps are described in detail in A140-A160.

[0019] Step A200: constructing a simplified feature space for the city space and adding a time axis, and updating the simplified feature space according to the city dynamic data.

[0020] In the embodiment of the present application, the simplified feature space is formed by performing a miniature projection of the urban space. After the spatial distribution is determined, the simplified feature space is filled based on the urban features under the constraints of relative spatial positions.

[0021] Optionally, for urban space, when constructing a simplified feature space and adding a time axis, the urban space is first projected in miniature to determine the spatial distribution, and then based on the urban characteristics, the distribution is filled under the relative spatial position constraints to determine the simplified feature space, and then the time axis is added. The specific steps are described in detail in A210-A230; when updating the simplified feature space based on urban dynamic data, the urban dynamic data is first received, and the data plane is generated on the time axis by identifying the timestamp therein, and then the location code is identified in the data plane, and the urban dynamic data is distributedly deployed to determine the updated simplified feature space. The specific steps are described in detail in A240-A250.

[0022] Step A300: Introduce a multidimensional indicator system, build a dynamic assessment module and embed it in the urban management system. By interacting with the simplified feature space, perform a rainstorm disaster risk assessment under the multidimensional indicator cascade and determine the risk assessment result.

[0023] In this embodiment, the city management system serves as the terminal integration and decision-making platform for the entire assessment method. It establishes interactive communication with the edge data network, receives city dynamic data processed by edge nodes, and embeds a dynamic assessment module to implement risk assessment. The risk assessment results are quantitative conclusions drawn by the dynamic assessment module based on a multidimensional indicator system and simplified feature space data, using cascade analysis and entropy weighting.

[0024] In one embodiment of the present application, first, a multidimensional indicator system of disaster-causing factor hazard - disaster-prone environment exposure - disaster-bearing body vulnerability - disaster prevention and mitigation capabilities is introduced to construct a cascade analysis layer with the same number of layers, and then the entropy weight combination empowerment principle is used to supervise the training of the cascade analysis layer, and the dynamic evaluation module is determined by empowering within and between layers. The specific steps are described in detail in A310-A320.

[0025] The constructed dynamic assessment module is then embedded and deployed within the urban management system, interacting in real time with the simplified feature space established earlier. This simplified feature space is identified through spatiotemporal coding to form a three-dimensional data cube of time, space, and indicators (e.g., each time plane contains real-time monitoring data for a 500×500 grid). The dynamic assessment module accesses urban dynamic data for specific grids in real time through an API (e.g., at 2:00 PM on August 15, 2024, grid 07 in District X had a rainfall of 42 mm / h, a water level of 35 cm, a population density of 12,000 people / km², and a drainage system return period of 3 years). This data is then fed into a four-level cascade analysis layer: the first layer calculates a hazard factor score; the second layer combines terrain data to calculate an exposure score; the third layer integrates population and building data to calculate a vulnerability score; and the fourth layer uses emergency resource data to calculate a disaster prevention and mitigation capability score. Finally, a weighted aggregation formula (risk index = 0.3 × hazard factor + 0.25 × exposure + 0.25 × vulnerability + 0.2 × disaster prevention and mitigation capability) is used to output a grid-level risk index, which then determines the risk assessment results.

[0026] Through the above series of steps, a comprehensive, dynamic and accurate assessment of urban rainstorm disaster risks has been achieved, providing scientific decision-making support for urban emergency management departments and effectively improving the city's ability to respond to rainstorm disasters.

[0027] Step A400: Feedback the risk assessment result to the simplified feature space, and perform interface display of dynamic risk.

[0028] Specifically, when the risk assessment results are fed back to the simplified feature space and the dynamic risk interface display is executed, the simplified feature space is first marked with data planes based on the risk assessment results including the risk coefficient, so that different risk levels can be intuitively reflected in the space. Then, corresponding risk warning instructions are generated based on the risk coefficient, so as to carry out early warning management of urban rainstorm disasters and realize the visualization and practical application of risk assessment results. The specific steps are detailed in A410-A420.

[0029] Step A500: wherein the multi-dimensional indicator system includes the risk of disaster-causing factors - exposure to disaster-prone environments - vulnerability of disaster-bearing bodies - disaster prevention and mitigation capabilities.

[0030] Specifically, in the urban rainstorm disaster risk assessment, the construction of a multidimensional indicator system takes the danger of disaster-causing factors - exposure to disaster-prone environments - vulnerability of disaster-bearing bodies - disaster prevention and mitigation capabilities as the core framework, and realizes the systematic integration of complex risk factors through layered decomposition and quantitative analysis.

[0031] The hazard factor risk dimension focuses on the direct causes of rainstorm disasters and includes key indicators such as rainfall intensity, duration, and the frequency of short-duration extreme rainfall. For example, based on meteorological monitoring data, hourly rainfall of 50 mm or more is defined as a high-risk rainfall intensity. Combined with historical data from the past 10 years, the frequency of rainstorms (daily rainfall 50 mm or more) in various regions is calculated. For example, an industrial zone has an average annual frequency of 12 rainstorms, significantly higher than the 8 in residential areas. Its hazard factor risk score can be preliminarily determined to be 0.8 (on a scale of 0-1, with higher values ​​indicating higher risk). Technical personnel deploy rain sensors to collect minute-by-minute rainfall data in real time and dynamically update the instantaneous risk parameters of the hazard factor.

[0032] The dimension of environmental exposure to disasters focuses on the sensitivity of the urban environment to heavy rainstorms, encompassing indicators such as terrain elevation, water system distribution, and drainage network density. Taking terrain data as an example, urban areas where low-lying areas with an altitude of ≤5m account for more than 30% have a significantly increased risk of rainwater retention and an exposure score of 0.7. If the river network coverage rate in the area is less than 10% and the drainage network density is less than 5km / km² (i.e., the length of the drainage network is less than 5 kilometers per square kilometer), the problem of poor rainwater drainage is further exacerbated, and the exposure score can be revised to 0.85. Using GIS spatial analysis technology, technicians in this field overlay elevation data, water system distribution, and real-time rainfall data to quantify the environmental exposure risks of different grid cells.

[0033] The vulnerability dimension of hazard-prone objects focuses on the vulnerability of affected areas, including population, economy, and infrastructure, and includes indicators such as population density, GDP density, and the proportion of older buildings. For example, a commercial district with a population density exceeding 15,000 people / km² receives a base vulnerability score of 0.6. If the proportion of buildings older than 30 years in the area reaches 40%, and underground space (such as garages and shopping malls) accounts for more than 20%, the vulnerability score increases to 0.75. Technical personnel can dynamically assess the exposure risk and loss potential of hazard-prone objects to heavy rain by combining urban population distribution heat maps, building census data, and real-time pedestrian monitoring data.

[0034] The disaster prevention and mitigation capacity dimension examines a city's hardware facilities and management effectiveness in responding to rainstorm disasters. This includes indicators such as drainage system design standards (return period), density of emergency material storage points, and warning response time. For example, a region with a drainage system designed with a return period of five years (i.e., capable of handling a five-year rainstorm) receives a disaster prevention capacity score of 0.8. This score increases to 0.9 if there are at least two emergency material storage points per square kilometer and the delay in warning information is less than 10 minutes. Technical personnel can assess the effectiveness of a city's disaster prevention and mitigation measures in real time by accessing real-time drainage system operational data (such as pump station flow and pipe network water levels) and data from the emergency management platform.

[0035] On the basis of the construction of the index system, the entropy weight combination method is used to determine the weights of each dimension and sub-indicator and build a dynamic evaluation module. The specific steps are described in detail in A320.

[0036] By breaking down urban rainstorm disaster risks into four core dimensions and refining quantitative indicators, and combining subjective and objective weighting methods to build a dynamic assessment module, the problems of single indicators and fixed weights in traditional assessments are solved, and the technical effects of multi-dimensional coupling analysis, dynamic quantitative assessment and precise risk identification of urban rainstorm disaster risks are achieved, providing a scientific risk decision-making basis for urban emergency management.

[0037] Furthermore, step A100 in the method provided in the embodiment of the present application includes: A110: For urban space, multiple urban zones are divided and determined based on the urban structure.

[0038] A120: Based on the partition structure characteristics, distributed multi-source sensor deployment is performed as the distributed source end.

[0039] A130: Deploy an edge node in each city district to form the edge data network and establish interactive communication between multi-source sensors, edge data network, and city management system.

[0040] In this embodiment of the application, the partition structure characteristics refer to the comprehensive characteristics that reflect the attributes related to the rainstorm disaster risk in the area based on the division of urban areas. Edge nodes are distributed computing units deployed in each urban area and are the core components of the edge data network.

[0041] Specifically, first, based on structural features such as urban functional zoning and geographical features (such as the distribution of commercial districts, river courses, and terrain elevation), the urban space is divided into several basic units with similar risk attributes. For example, a medium-sized city can be divided into 50-80 urban zones with a grid density of 1-3 square kilometers. This ensures that the types of hazard-prone objects (such as residential areas, industrial areas, and transportation hubs) and the distribution of hazard-causing factors (such as flood-prone road sections and drainage network nodes) within each zone are relatively homogeneous, facilitating the subsequent differentiated deployment of sensors.

[0042] Next, guided by the structural characteristics of each zone, mainly including urban functional zoning (such as commercial areas, residential areas, industrial areas, transportation hubs, etc.), geographical and geomorphological features (such as terrain elevation, water system distribution, road network density), disaster-prone body types (such as population density, building age, underground space ratio) and disaster-causing factor distribution (such as the location of waterlogged road sections, drainage network node layout, historical waterlogging point distribution), distributed multi-source sensors are deployed as the end point of data collection.

[0043] In densely populated commercial areas, meteorological sensors (real-time monitoring of rainfall and wind speed, with a sampling frequency of 1 time / minute) and video surveillance equipment (resolution ≥1080P, supporting intelligent identification of water depth) are deployed. In low-lying residential areas, IoT water level sensors (accuracy of ±0.5cm, supporting 24-hour real-time monitoring) and soil moisture sensors are deployed more frequently. Traffic flow sensors and structural settlement monitoring equipment are added along major arterial roads and bridge nodes. Due to space limitations, sensors deployed in other areas based on structural characteristics are not described here. This targeted deployment ensures that sensor coverage density matches the regional risk level. For example, in high-risk areas, the deployment density can reach 50 sensors per square kilometer, ensuring real-time capture and coverage of risk factors such as waterlogging and traffic congestion caused by heavy rain.

[0044] Finally, edge nodes are deployed within each city district. The specific steps are detailed in A131-A133. An edge data network is constructed via wired (fiber) or wireless (5G) communication networks. Edge nodes typically feature a quad-core processor (2.4GHz), 8GB of memory, and 128GB of solid-state storage, allowing simultaneous access to the various sensor devices mentioned above. This creates a three-tiered interactive communication architecture: multi-source sensors, edge data network, and city management system. Raw sensor data (e.g., a water level sensor output of 32.7cm in a specific district at a specific moment, or a weather station output of 45mm of hourly rainfall) is first transmitted to the corresponding edge node via the MQTT protocol. After preliminary noise removal and format standardization, it is uploaded to the city management system.

[0045] Through refined zoning based on urban structure, targeted sensor deployment and distributed networking of edge computing nodes, the problems of uneven data collection coverage, poor compatibility of heterogeneous data and high transmission delay in traditional assessment methods have been solved. The technical effects of grid-based precise collection of urban rainstorm disaster-related data, lightweight edge processing and time-space synchronous transmission have been achieved, laying a solid data foundation for subsequent multi-source data fusion and dynamic risk assessment.

[0046] Furthermore, step A100 in the method provided in the embodiment of the present application includes: A131: For the first city partition, determine multiple communication threads between the multi-source sensors and the city management system, where each sensor corresponds to a communication thread.

[0047] A132: On the sensor side, the multiple communication threads are aggregated to determine a thread aggregation node.

[0048] A133: Open an external interface at the thread aggregation node and expand a data processing component, wherein the data processing component performs data reconstruction encapsulation and space-time code identification.

[0049] In this embodiment, multiple communication threads refer to independent data transmission channels established between multi-source sensors within a city zone and the city management system. A thread aggregation node is a software-defined logical processing unit built on the sensor-side communication path and integrated into existing communication equipment. A data processing component is a lightweight software module extended on the thread aggregation node.

[0050] Alternatively, the deployment of edge nodes abandons the traditional edge-side data center construction model and focuses on optimizing the communication process. By implementing data aggregation and lightweight processing at the thread level on the sensor side, an efficient data transmission link is built.

[0051] Taking the first urban district as an example, the system first identifies the communication connections between the multi-source sensors (such as meteorological sensors, water level sensors, and video surveillance equipment) within the district and the city management system. Each sensor, depending on the protocol (such as MQTT, Modbus, or TCP), corresponds to an independent communication thread, forming parallel data transmission channels. For example, if 20 sensors in a district are operating simultaneously, 20 independent communication threads will be generated, each carrying raw data in a different format (such as ASCII data from a rain sensor or RTSP streaming data from a video device).

[0052] These parallel threads are aggregated at the physical access point of the sensor-side communication path (such as a smart gateway or aggregation router within a zone), forming a thread aggregation node. This node is not an additional hardware device, but rather a software-defined node that leverages the underlying hardware resources of existing communication devices (such as smart gateways and aggregation routers) and partitions them into independent logical processing units using embedded software or lightweight container technologies (such as Docker).

[0053] This processing unit implements real-time monitoring and management of multi-source sensor communication threads through customized software. It parses protocol data from each thread (such as MQTT and Modbus) to identify and filter invalid information such as sensor self-test signals and duplicate data. It also dynamically adjusts thread priorities and bandwidth allocation (for example, prioritizing the water level sensor thread during heavy rain) to ensure the real-time transmission of high-value data. This allows for real-time monitoring and traffic management of multi-threaded data. For example, if a thread detects a sensor self-test signal or duplicate invalid data, the thread aggregation node automatically filters this information, retaining only valid monitoring data (such as key indicators like rainfall and water level), thus providing preliminary data screening.

[0054] Next, an external interface is opened on the thread aggregation node to expand the lightweight data processing component, which integrates data reconstruction encapsulation and space-time code identification functions. The specific steps are detailed in the following A140-A160.

[0055] By performing multi-threaded aggregation and lightweight processing on the sensing side of the communication thread, the invalid redundancy problem caused by directly collecting raw data in the traditional method is avoided. Without additional hardware investment, real-time screening, format unification and time-space labeling of data can be achieved, achieving the technical effects of optimizing the data transmission process, reducing network load and improving data validity, and providing efficient communication and processing support for urban management systems to obtain high-quality urban dynamic data in real time.

[0056] Furthermore, step A100 in the method provided in the embodiment of the present application includes: A140: The multi-source sensor performs urban element monitoring and determines urban perception data.

[0057] A150: According to the interactive communication relationship, the city perception data is imported into the partitioned edge node of the edge data network, and data feature element extraction and classification packaging are performed to determine the city dynamic data.

[0058] A160: Introduce timestamps and location codes to mark the city dynamic data.

[0059] In this embodiment of the present application, urban elements include disaster-causing factors, disaster-prone environments, disaster-bearing bodies, and disaster prevention and mitigation elements. Urban perception data is raw data acquired by multi-source sensors through physical monitoring or signal acquisition. Urban dynamic data is standardized, structured data generated by edge nodes after processing urban perception data.

[0060] Specifically, first, multi-source sensors distributed in various urban areas (such as meteorological sensors, IoT water level sensors, video surveillance equipment, etc.) continuously perform urban element monitoring: meteorological sensors collect data such as rainfall and rainfall intensity at a frequency of 1 time per minute, water level sensors monitor the depth of accumulated water in real time with an accuracy of ±0.5cm, and video equipment uses computer vision technology to extract characteristic values ​​such as the proportion of flooded areas on roads, forming urban perception data containing original signals, protocol codes, and equipment identification.

[0061] The collected sensor data is then transmitted in real time to the edge nodes in the corresponding zones via an interactive communication link between multi-source sensors, the edge data network, and the city management system. These edge nodes, acting as distributed computing units, first perform protocol parsing and feature extraction on the data. For Modbus water level data (e.g., the original hexadecimal string 0x01 0x03 0x00 0x01 0x00 0x02 0x75 CB), the core monitoring value of 32.7 cm is extracted, and redundant fields such as the checksum and device address are removed. For video stream data, a deep learning model (e.g., YOLOv5) is used to extract the characteristic value of the accumulated water depth (e.g., 45 cm) and compress the video data to 1 / 10 its original size. This feature extraction increases the proportion of valid information in the data compared to the original data.

[0062] Next, the edge node categorizes and encapsulates these features according to a unified data structure: meteorological data, water level data, and video feature values ​​are categorized into data types such as disaster-causing factors and disaster-prone environments, generating structured data units consisting of data type, monitored value, acquisition time, and spatial location. For example, data from a water level sensor is encapsulated as: {"type":"water_level","value":"32.7cm","device_id":"A01","timestamp":"20241001143000","location":"Area A - Grid 03"}. During this process, the edge node automatically reads the sensor's preset location code (e.g., zone number - grid number) and system real-time time, generating a second-accurate timestamp (e.g., 20241001143000) and location code (e.g., Area A - Grid 03 corresponds to 30.5°N, 114.3°E), thus completing the dual temporal and spatial tagging of urban dynamic data. After processing, the amount of single data is compressed, the data transmission efficiency is improved, and each data has unique space-time coordinates, which facilitates subsequent GIS spatial analysis and time series modeling.

[0063] Through the collaborative monitoring of multi-source sensors, feature extraction and standardized packaging of edge nodes, and precise labeling of spatiotemporal codes, the technical effect of converting multi-source raw perception data into standardized urban dynamic data with spatiotemporal attributes is achieved, providing high-quality basic data support for simplifying feature space construction and dynamic risk assessment.

[0064] Furthermore, step A200 in the method provided in the embodiment of the present application includes: A210: Project the urban space in miniature to determine its spatial distribution.

[0065] A220: Based on the city characteristics, the spatial distribution is filled under the relative spatial position constraints to determine the simplified feature space.

[0066] A230: Add a time axis in the simplified feature space.

[0067] In the embodiment of the present application, urban characteristics refer to comprehensive indicators that reflect urban spatial structure, functional attributes, and factors related to rainstorm disaster risks.

[0068] Specifically, first, a grid-like projection of urban space is performed using a geographic information system (GIS), dividing continuous urban areas into discrete spatial units. For example, a medium-sized city can be divided into 5,000 x 5,000 basic grids, using a 100m x 100m grid unit. Each grid corresponds to a unique spatial coordinate (e.g., longitude and latitude are mapped to an X-zone-Y-grid grid). This results in a spatial distribution matrix containing approximately 25 million grid cells, covering the urban built-up area. This gridding process transforms complex urban space into calculable discrete spatial units, providing a basic framework for subsequent feature filling.

[0069] Next, based on core characteristics such as urban functional zoning, terrain elevation, and population density, each grid cell was populated with attributes. Within spatial location constraints, urban characteristics were mapped to the grid through data interpolation and association analysis. For terrain elevation, elevation data (with an accuracy of ±0.1m) was populated into each grid cell using a digital elevation model (DEM). Population density was quantified into units of 1,000 people / km² (e.g., population density of x people / km² for a commercial district grid and y people / km² for an industrial district grid) by combining neighborhood committee demographic data and real-time heat maps. For drainage system distribution, infrastructure data such as pipe network density (km / km²) and pump station locations were associated with the corresponding grid cells. After this populating process, each grid cell formed a vector containing 10-15 basic features (e.g., [elevation, population density, pipe network density, mean building age, etc.]), constructing a spatial feature matrix with uniform dimensions—the simplified feature space.

[0070] Finally, a time axis is added to the spatial feature matrix, coupling the one-dimensional time series with the two-dimensional spatial grid to form a three-dimensional time-space-feature data cube. The time axis uses minutes as the smallest unit (e.g., 1 minute per time slice), and each time point corresponds to a two-dimensional data plane (i.e., the set of feature vectors for all grids at that moment). For example, the time plane at 2:00 PM on August 15, 2024, contains real-time monitoring data (such as rainfall, water level, and traffic flow) for 5000 × 5000 grids. Together with historical time planes (e.g., the previous hour and the previous day), this constitutes a dynamic time series. In this way, spatiotemporal data related to urban rainstorm disasters is organized into a structured model that can be efficiently retrieved and analyzed. Approximately 1440 time planes are generated for a single city per day, significantly improving data storage efficiency compared to unstructured storage.

[0071] The urban space is discretized through grid miniature projection, the spatial unit attributes are filled based on the core features, and then a three-dimensional data cube is constructed through time axis expansion. This achieves the technical effect of converting the spatiotemporal data related to urban rainstorm disasters into a standardized and structured feature space, providing underlying data architecture support for the efficient operation of subsequent dynamic risk assessment models.

[0072] Furthermore, step A200 in the method provided in the embodiment of the present application includes: A240: Receive the city dynamic data, and generate a data plane in the time axis by identifying the timestamp.

[0073] A250: In the data plane, the city dynamic data is distributedly deployed by identifying the location code, and the updated simplified feature space is determined.

[0074] In the embodiment of the present application, a data plane is a two-dimensional data layer corresponding to a specific time point in a simplified feature space.

[0075] Specifically, the city management system first receives real-time city dynamic data from the edge data network. Each data entry includes a timestamp accurate to the second (in the format YYYYMMDDHHMMSS, such as 20241001143000, where YYYY represents the four-digit year, the first MM represents the two-digit month, DD represents the two-digit day, HH represents the two-digit hour, the second MM represents the two-digit minute, and SS represents the two-digit second) and a grid location code (such as Area A - Grid 03). The system parses the timestamp and locates the corresponding time slice along the preset time axis (usually with 1 minute as the minimum time unit), generating a two-dimensional data plane corresponding to that point in time. For example, the time plane for 14:30 on October 1, 2024, will be created as an independent data layer to carry the real-time monitoring data of all grid cells at that moment.

[0076] After generating the data plane, the system further identifies the location codes in the city's dynamic data and maps them to grid coordinates in the simplified feature space. The location codes use a two-layer encoding system: zone number and grid number (e.g., zone X and grid Y correspond to longitude and latitude coordinates in a GIS system). Using predefined spatial indexing algorithms (e.g., quadtree indexing), the corresponding grid cells (typically with a grid resolution of 100m x 100m) can be located: First, the city is divided into first-level zones, such as X and Y, based on administrative or geographic characteristics. Each zone corresponds to the root node of a quadtree. A single zone is then recursively divided into four equal parts, generating four child nodes each time, until the sub-zone side length approaches the grid precision (e.g., 100 meters). Each leaf node is assigned a unique grid number. During positioning, after inputting longitude and latitude coordinates, the algorithm begins at the root node (zone) and determines, layer by layer, which sub-quadrant the coordinate belongs to: the upper left, upper right, lower left, or lower right of the current zone. By rapidly traversing the tree structure to a leaf node, the corresponding zone number—the grid number—is obtained. This process leverages the hierarchical spatial partitioning properties of the quadtree to reduce coordinate matching complexity to logarithmic levels, achieving precise mapping from longitude and latitude to grid cells within 5 milliseconds, ensuring the association of dynamic urban data with the simplified feature space.

[0077] For example, consider water level monitoring data: Grid 03 in Area A corresponds to the geographic coordinates 30.5°N, 114.3°E. The system automatically updates the real-time water level value (32.7 cm) for that grid cell to the corresponding location on the data plane, overlaying or integrating historical data for that grid cell (e.g., the average water level value for the previous 10 minutes). If multiple sensor data are received at the same time plane (e.g., rainfall from a meteorological sensor, congestion index from a traffic sensor), the system updates the grid feature vectors in multiple dimensions based on data type (e.g., hazard factor, hazard-affected object), forming a real-time feature set containing 10-15 indicators (e.g., [rainfall 45 mm / h, water level 32.7 cm, population density 12,000 people / km²]).

[0078] This update process, through the dual identification of timestamps and location codes, achieves a data deployment mechanism with orderly timeline arrangement and precise spatial grid positioning. The updated data model can be directly used by the dynamic assessment module. For example, in a risk assessment at 2:30 PM, the system can retrieve all grid features at that time in real time and calculate the risk coefficient for each area based on a multi-dimensional indicator system.

[0079] Through timestamp-driven dynamic surface generation and location code-guided grid data deployment, the problems of time dislocation and inefficient spatial matching in traditional spatiotemporal data integration are solved. The real-time monitored urban dynamic data is accurately embedded in the simplified feature space to form a three-dimensional spatiotemporal data model updated at the minute level. This provides efficient and accurate underlying data support for the real-time dynamic assessment of rainstorm disaster risks.

[0080] Furthermore, step A300 in the method provided in the embodiment of the present application includes: A310: Construct a cascade analysis layer using the multidimensional indicator system, wherein the number of layers in the cascade analysis layer is consistent with the number of dimensions in the multidimensional indicator system.

[0081] A320: Introduce the entropy weight combination weighting principle, perform supervised training based on the cascade analysis layer, and determine a dynamic evaluation module, which includes first-level intra-level weighting and second-level inter-level weighting.

[0082] In one embodiment, four cascade analysis layers are first constructed based on the four dimensions of the indicator system. Each layer corresponds to a core dimension and integrates key indicators for that dimension. For example, the hazard risk layer includes indicators such as hourly rainfall (threshold ≥50 mm / h is marked as high risk), short-duration rainstorm frequency (an average of ≥15 times per year over the past 10 years is defined as high risk), and rainfall duration (≥6 consecutive hours triggers a waterlogging warning). The hazard exposure layer incorporates indicators such as terrain elevation (areas with an elevation of ≤5 m or ≥30% increase exposure risk), water system density (river coverage <10% increases stagnation), and drainage network density (<5 km / km² is considered an inefficient drainage area). Each layer of indicators is normalized and converted into dimensionless values ​​in the range of 0-1, forming a hierarchical input feature matrix.

[0083] Next, based on the cascade analysis layer, the entropy weight combination empowerment principle is introduced to carry out supervised training. The intra-layer empowerment focuses on the importance ranking of single-dimensional indicators: Step a: First, the objective weight is calculated using the entropy weight method. This method, based on the principle of information entropy, uses the degree of dispersion in historical data to measure the importance of an indicator. For example, for the short-duration rainstorm frequency indicator, historical data for this indicator for each city district over the past 10 years are collected. For example, the data for a particular region is [12, 15, 8, 10, 18...]. This data is then normalized to eliminate dimensionality. Next, the information entropy is calculated based on this standardized data. Lower information entropy indicates greater data volatility and more information. For example, an information entropy of 0.7 for the short-duration rainstorm frequency indicator indicates relatively low data fluctuation. Finally, the objective weight of this indicator is calculated to be 0.35, meaning that data variability accounts for 35% of its importance within the hazard factor layer.

[0084] Step b: Simultaneously incorporate the Analytic Hierarchy Process (AHP) to obtain subjective expert weights. The AHP relies on expert experience and quantifies the relative importance of indicators by constructing a judgment matrix. Invite 5-10 experts in fields such as urban planning and meteorological disasters to compare indicators within the same dimension using a 1-9 scale. For example, if an expert believes that rainfall intensity has twice the impact of rainfall duration on a hazard, then enter 2 in the corresponding position in the judgment matrix; otherwise, enter 1 / 2. After constructing the matrix, perform a consistency check to ensure the rationality of the judgment. If the check passes, solve the matrix using the eigenvector method or square root method to obtain the subjective weight of each indicator.

[0085] Step c: Determine the comprehensive weights of the indicators within the stratum through linear combination. This linear combination proportionally combines the objective weights obtained by the entropy weight method with the subjective weights obtained by the AHP. Those skilled in the art can set the weight ratios based on actual needs, such as 60% objective weights and 40% subjective weights. This result balances the objectivity of data-driven analysis with the subjectivity of expert experience, and can more comprehensively reflect the importance of indicators within the stratum.

[0086] Step d: Inter-layer weighting and dynamic weight matrix construction. Inter-layer weighting focuses on the interrelated impacts between different dimensions, such as disaster-causing factors, disaster-prone environments, and vulnerability of disaster-bearing bodies. Using historical disaster loss data, such as direct economic losses and the number of affected people from 120 flooding incidents in the past five years, as supervisory signals, a loss function is defined to measure the gap between the model's predicted risk level and actual losses. A gradient descent algorithm is used to iteratively update the inter-layer weight parameters by calculating the gradient of the loss function with respect to the weight parameters. For example, when the deviation between the model's predicted risk level and the actual loss exceeds 15%, the contribution coefficient of the disaster prevention and mitigation capacity layer to the final risk index is automatically adjusted from an initial 0.2 to 0.25. After multiple rounds of iteration, a dynamic weight matrix is ​​formed, comprising intra-layer indicator weights and inter-layer adjustment coefficients, enabling precise calibration of the risk assessment model, thereby determining the dynamic assessment module.

[0087] During training, standardized urban dynamic data (e.g., real-time rainfall of 45 mm / h, water level of 32 cm, population density of 12,000 people / km², and drainage system return period of 3 years for a given grid cell) is sequentially fed into a four-level cascade analysis layer. The first layer calculates the hazard factor score, the second layer combines terrain data to calculate the exposure score, the third layer integrates population and building data to calculate the vulnerability score, and the fourth layer uses emergency resource data to calculate the disaster prevention and mitigation capability score. Finally, a grid-level risk index is output using an inter-layer weighting formula (risk index = 0.3 × hazard factor + 0.25 × exposure + 0.25 × vulnerability + 0.2 × disaster prevention and mitigation capability). Through repeated training with historical samples, the model's risk level prediction accuracy has steadily improved.

[0088] Through the above steps, the technical effects of multi-dimensional coupling analysis, dynamic weight calculation and accurate level assessment of urban rainstorm disaster risks were achieved, providing a scientific real-time risk decision-making basis for emergency management.

[0089] Furthermore, step A330 in the method provided in the embodiment of the present application includes: A331: Introduce a normalization processing node, and embed the normalization processing node into the dynamic evaluation module.

[0090] A332: wherein the normalization processing node provides indicator normalization conditions for indicator cross-linking analysis.

[0091] In this embodiment, the normalization processing node is a component embedded in the dynamic assessment module, used to address the issues of indicator specificity and data standard discrepancies in multi-source data. Indicator cross-linking analysis is a process within the dynamic assessment module that comprehensively assesses the urban rainstorm disaster risk by comprehensively considering the interrelationships and impacts between multiple indicators based on normalized indicator data.

[0092] Optionally, first, classify the indicators into categories based on their characteristics and formulate normalization rules: Extremely large indicators (the larger the value, the higher the risk, such as rainfall and population density): use the extreme value standardization formula For example, the historical extreme hourly rainfall in a city is 80 mm / h, and the current value is 45 mm / h. The normalized score is (45-20) / (80-20)=0.417 (assuming the minimum value of 20 mm / h is the no-risk threshold).

[0093] Extremely small indicators (smaller values ​​indicate higher risks, such as terrain elevation and drainage system response time): using the reverse extreme value formula For example, if the lowest altitude in a certain area is 2m (high risk) and the highest altitude is 50m (low risk), and the current elevation is 5m, the normalized score is (50-5) / (50-2)=0.9375, where a higher value indicates a lower risk.

[0094] Interval-type indicators (there is an optimal interval; deviations from this interval increase the risk, such as drainage system design recurrence period and emergency material storage point density): use piecewise function normalization. Taking the drainage system recurrence period as an example, the optimal interval is 5-10 years. If it is less than 5 years or more than 10 years, the risk increases. The formula is: If the recurrence period of a region is 3 years ( =1), the score is 1−(5−3) / (5−1)=0.5, indicating a medium risk.

[0095] The process for embedding the normalization processing node into the dynamic assessment module is as follows: Urban dynamic data (e.g., hourly rainfall of 45 mm / h, terrain elevation of 5 m, and a drainage system return period of 3 years) first enters the normalization processing node. The system automatically identifies the indicator type (using predefined indicator attribute tags, such as rainfall:max elevation:min drainage:interval, representing rainfall, terrain elevation, and drainage system indicators, respectively) and invokes the corresponding normalization algorithm to generate standardized values ​​(e.g., rainfall 0.417, elevation 0.9375, and return period 0.5). The processed data is then fed into the cascade analysis layer to ensure that all dimensional indicators are analyzed in the same dimensionality. For example, for the hazard-prone vulnerability dimension, population density (extremely large, normalized to 0.8) and the proportion of old buildings (extremely large, normalized to 0.7) are weighted within the tier (assuming weights of 0.6 and 0.4, respectively), resulting in a score of 0.8 × 0.6 + 0.7 × 0.4 = 0.76 for this dimension, thus avoiding assessment bias caused by dimensional differences.

[0096] By embedding a normalization processing node in the dynamic assessment module, different normalization algorithms are designed for indicators with different characteristics such as extremely large and interval types, and heterogeneous data are converted into standardized values ​​of unified dimensions. This solves the compatibility issues caused by different data standards in the comprehensive assessment of multiple indicators, and achieves the technical effect of providing homogeneous input data for the cascade analysis layer and improving the accuracy and robustness of the risk assessment model.

[0097] Furthermore, step A400 in the method provided in the embodiment of the present application includes: A410: Perform data plane identification on the simplified feature space according to the risk assessment result, wherein the risk assessment result includes a risk coefficient.

[0098] A420: Based on the risk coefficient, a risk warning instruction is generated to conduct early warning management of urban rainstorm disasters.

[0099] In one embodiment, the risk assessment results (including risk coefficients ranging from 0 to 1, such as a grid cell with a risk coefficient of 0.75) output by the dynamic assessment module are first transmitted in real time to a simplified feature space. The system then hierarchically labels the corresponding data planes in the three-dimensional data cube based on preset risk level thresholds (e.g., below 0.4 for low risk, 0.4-0.7 for medium risk, and above 0.7 for high risk). Specifically, the grid cells in each time plane are color-coded based on the risk coefficient (e.g., red for high risk, orange for medium risk, and yellow for low risk), forming a global risk distribution heat map. For example, if the risk coefficient of a grid cell in an industrial zone is 0.82 in the time plane at 2:30 PM on August 15, 2024, the grid cell will be marked red on the GIS interface, with a flashing animation superimposed to highlight it, allowing emergency personnel to quickly locate high-risk areas.

[0100] Then, based on the data plane identification, the system automatically generates graded warning instructions based on the risk factor. The warning rules utilize a threshold trigger mechanism: when the risk factor is ≥0.7, an orange warning is generated, prompting the relevant areas to operate drainage pumping stations at full capacity and close underground entrances, among other measures. When the risk factor is ≥0.8, a red warning is upgraded, triggering evacuation plans and traffic control orders. These warning instructions contain precise spatiotemporal information (e.g., if the X-area-Y grid indicates a red risk level, immediate evacuation of residents in low-lying areas is recommended). These instructions are simultaneously pushed to drainage departments, traffic police command centers, community management platforms, and other terminals via the city management system's emergency interface.

[0101] By mapping the risk coefficient to the grid cells of the simplified feature space, real-time visual identification of the risk distribution is achieved, and accurate graded warning instructions are automatically generated based on threshold rules. This solves the problem that traditional risk assessment results are difficult to present intuitively and the warning response is delayed. The technical effect of converting abstract risk assessment results into operational visual interfaces and emergency instructions is achieved, providing an integrated solution for real-time monitoring and efficient handling of urban rainstorm disasters.

[0102] In summary, the method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion provided by the embodiments of the present application has the following technical effects: This application builds a data transmission link between urban edge nodes and the urban management system, uses thread aggregation technology to aggregate multi-source sensor data, and obtains standardized urban dynamic data through operations such as data processing component reconstruction and packaging, and spatiotemporal code identification. It then stores and updates data in a simplified feature space. After normalization of processing nodes and risk calculation by the dynamic assessment module, combined with entropy weight combination empowerment and cascade analysis mechanisms, early warning management based on risk factors is carried out and real-time feedback is provided, ensuring the efficiency and accuracy of the dynamic assessment of urban rainstorm disaster risks. This achieves an accurate and comprehensive assessment of urban rainstorm disaster risks and provides scientific decision-making support for urban emergency management.

[0103] Example 2, as Figure 2 As shown, based on the same inventive concept as in the aforementioned embodiment 1, this embodiment of the present application provides a dynamic risk assessment system for urban rainstorm disasters fused with multi-source data, the system comprising: City data determination module 1, the city data determination module 1 is used to deploy an edge data network at a distributed source end, reconstruct and encapsulate the received multi-source data and identify it with a time-space code, and determine the city dynamic data, wherein the edge data network is opened at the sensor side of the communication thread.

[0104] The feature space simplification module 2 is used to construct a simplified feature space for the urban space and add a time axis, and update the simplified feature space according to the urban dynamic data.

[0105] The evaluation result determination module 3 is used to introduce a multidimensional indicator system, construct a dynamic evaluation module and embed it in the urban management system. By interacting with the simplified feature space, it performs a rainstorm disaster risk assessment under a multidimensional indicator cascade to determine the risk assessment result.

[0106] The interface display execution module 4 is used to feed back the risk assessment result to the simplified feature space and execute the interface display of dynamic risk.

[0107] Index system determination module 5, wherein the multi-dimensional index system includes the risk of disaster-causing factors - exposure to disaster-prone environments - vulnerability of disaster-bearing bodies - disaster prevention and mitigation capabilities.

[0108] Furthermore, the city data determination module 1 is configured to perform the following steps: For urban space, multiple urban zones are divided and determined based on the urban structure; distributed multi-source sensors are deployed as the distributed source ends based on the characteristics of the zone structure; an edge node is deployed in each urban zone to form the edge data network, and interactive communication between the multi-source sensor, edge data network and urban management system is established.

[0109] Furthermore, the city data determination module 1 is configured to perform the following steps: For the first city partition, multiple communication threads between multi-source sensors and the city management system are determined, wherein each sensor corresponds to a communication thread; on the sensor side, the multiple communication threads are aggregated to determine a thread aggregation node; an external interface is opened at the thread aggregation node, and a data processing component is expanded, wherein the data processing component performs data reconstruction encapsulation and space-time code identification.

[0110] Furthermore, the city data determination module 1 is configured to perform the following steps: The multi-source sensor performs urban element monitoring and determines urban perception data; based on the interactive communication relationship, the urban perception data is imported into the partitioned edge node of the edge data network, and data feature element extraction and classification packaging are performed to determine urban dynamic data; timestamps and location codes are introduced to mark the urban dynamic data.

[0111] Furthermore, the feature space simplification module 2 is used to perform the following steps: The urban space is miniature-projected to determine the spatial distribution; based on the urban characteristics, the spatial distribution is filled under the relative spatial position constraints to determine the simplified feature space; and a time axis is added to the simplified feature space.

[0112] Furthermore, the feature space simplification module 2 is used to perform the following steps: The city dynamic data is received, and a data plane is generated in the time axis by identifying the timestamp; in the data plane, the city dynamic data is distributedly deployed by identifying the location code, and the updated simplified feature space is determined.

[0113] Furthermore, the evaluation result determination module 3 is configured to perform the following steps: A cascade analysis layer is constructed using the multidimensional indicator system, wherein the number of layers in the cascade analysis layer is consistent with the number of dimensions of the multidimensional indicator system; the entropy weight combination weighting principle is introduced, and supervised training based on the cascade analysis layer is performed to determine a dynamic evaluation module, which includes a first-level intra-level weighting and a second-level inter-level weighting.

[0114] Furthermore, the evaluation result determination module 3 is configured to perform the following steps: A normalization processing node is introduced and embedded into the dynamic evaluation module; wherein the normalization processing node provides an indicator normalization condition for indicator cross-linking analysis.

[0115] Furthermore, the interface display execution module 4 is used to execute the following steps: According to the risk assessment result, the simplified feature space is marked with data planes, wherein the risk assessment result includes a risk coefficient; according to the risk coefficient, a risk warning instruction is generated to carry out early warning management of urban rainstorm disasters.

[0116] The multi-source data fusion urban rainstorm disaster dynamic risk assessment system provided by the embodiment of the present invention can execute the multi-source data fusion urban rainstorm disaster dynamic risk assessment method provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0117] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules may be used and run on the user terminal and / or server, and the various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention.

[0118] The above specific embodiments do not constitute a limitation to the scope of protection of this application. It should be understood by those skilled in the art that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of this application should be included in the scope of protection of this application. In some cases, the actions or steps recorded in this application can be performed in an order different from that in the embodiments and can still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A dynamic risk assessment method for urban rainstorm disasters based on multi-source data fusion, characterized by: The method comprises: Deploy an edge data network at the distributed source end to reconstruct, encapsulate and identify the received multi-source data with time and space codes to determine the city dynamic data, wherein the edge data network is opened at the sensor side of the communication thread; For urban space, a simplified feature space is constructed and a time axis is added, and the simplified feature space is updated according to the urban dynamic data; A multi-dimensional indicator system is introduced, a dynamic assessment module is constructed and embedded in the urban management system. By interacting with the simplified feature space, a rainstorm disaster risk assessment is performed under the multi-dimensional indicator cascade to determine the risk assessment results. Feeding back the risk assessment results to the simplified feature space and performing interface display of dynamic risks; The multidimensional indicator system includes the risk of disaster-causing factors, exposure to disaster-prone environments, vulnerability of disaster-bearing bodies, and disaster prevention and mitigation capabilities.

2. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 1, characterized in that: Deploy edge data networks at distributed sources, including: For urban space, multiple urban zones are divided and determined based on the urban structure; Guided by the characteristics of the partition structure, distributed multi-source sensor deployment is performed as the distributed source end; An edge node is deployed in each city partition to form the edge data network, and interactive communication between multi-source sensors, edge data network and city management system is established.

3. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 2, characterized in that: An edge node is deployed in each city zone. The edge node deployment methods include: For the first city zone, determining multiple communication threads between the multi-source sensors and the city management system, wherein each sensor corresponds to one communication thread; On the sensor side, the multiple communication threads are aggregated to determine a thread aggregation node; An external interface is opened on the thread aggregation node, and a data processing component is extended, wherein the data processing component performs data reconstruction encapsulation and space-time code identification.

4. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 2, characterized in that: Reconstruct and encapsulate the received multi-source data and identify it with time and space codes, including: The multi-source sensors perform urban element monitoring and determine urban perception data; According to the interactive communication relationship, the city perception data is imported into the partitioned edge nodes of the edge data network, and data feature elements are extracted and classified and packaged to determine the city dynamic data; Timestamps and location codes are introduced to mark the city dynamic data.

5. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 4, characterized in that: Construct a simplified feature space and add a time axis, including: Projecting the urban space in miniature to determine the spatial distribution; Based on the city characteristics, the spatial distribution is filled under the relative spatial position constraint to determine the simplified feature space; A time axis is added to the simplified feature space.

6. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 5, characterized in that: Updating the simplified feature space according to the city dynamic data includes: Receiving the city dynamic data, and generating a data plane in the time axis by identifying the timestamp; In the data plane, the city dynamic data is distributedly deployed by identifying the location code, and the updated simplified feature space is determined.

7. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 1, characterized in that: Introducing a multi-dimensional indicator system and building a dynamic evaluation module, including: Constructing a cascade analysis layer using the multidimensional indicator system, wherein the number of layers in the cascade analysis layer is consistent with the number of dimensions in the multidimensional indicator system; The entropy weight combination weighting principle is introduced, and supervised training based on the cascade analysis layer is performed to determine a dynamic evaluation module, which includes a first-level intra-level weighting and a second-level inter-level weighting.

8. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 7, characterized in that: Introducing a normalization processing node, and embedding the normalization processing node into the dynamic evaluation module; The normalization processing node provides indicator normalization conditions for indicator cross-linking analysis.

9. The method for dynamic risk assessment of urban rainstorm disasters based on multi-source data fusion according to claim 1, characterized in that: Feeding back the risk assessment result to the simplified feature space and performing a dynamic risk interface display includes: Performing data plane identification on the simplified feature space according to the risk assessment result, wherein the risk assessment result includes a risk coefficient; Based on the risk coefficient, a risk warning instruction is generated to carry out early warning management of urban rainstorm disasters.

10. A multi-source data fusion urban rainstorm disaster dynamic risk assessment system, characterized by: A method for dynamic risk assessment of urban rainstorm disasters using multi-source data fusion according to any one of claims 1 to 9, the system comprising: A city data determination module is used to deploy an edge data network at a distributed source end, reconstruct and encapsulate the received multi-source data, and identify it with a time-space code to determine the city dynamic data, wherein the edge data network is opened at the sensor side of the communication thread; A feature space simplification module is used to construct a simplified feature space for the urban space and add a time axis, and update the simplified feature space according to the urban dynamic data; An assessment result determination module is used to introduce a multi-dimensional indicator system, construct a dynamic assessment module and embed it into the urban management system. By interacting with the simplified feature space, it performs a rainstorm disaster risk assessment under a multi-dimensional indicator cascade and determines the risk assessment result. An interface display execution module, configured to feed back the risk assessment result to the simplified feature space and execute an interface display of dynamic risk; An indicator system determination module, wherein the multidimensional indicator system includes the hazard of disaster-causing factors - exposure to disaster-prone environments - vulnerability of disaster-bearing bodies - disaster prevention and mitigation capabilities.

Images