Urban Subway Flood Anomaly Location and Display System Integrating Multi-Source Data

Abstract

The application provides a kind of city subway water disaster anomaly positioning and display system of fusion multi-source data, with such characteristics, including data acquisition module;Data storage module;Factor analysis module is used to analyze the main factors of waterlogging;Water disaster prediction module is used to construct input data according to the data corresponding to each main factor of waterlogging in characteristic data, and input water disaster prediction model, obtain water disaster prediction place and corresponding prediction severity;Waterlogging prediction module is used to calculate each water disaster prediction place according to the corresponding characteristic data, obtain the waterlogging prediction range and the predicted water depth of each area of the water disaster prediction place as the waterlogging prediction result;Result positioning module is used to calculate each subway station affected by water disaster as disaster site according to waterlogging prediction result and subway line map data;Display module.In short, the method can accurately predict urban waterlogging and mark and display disaster-affected subway stations.

Description

Technical Field

[0001] This invention belongs to the field of flood prediction, specifically relating to a system for locating and displaying urban subway flood anomalies by integrating multi-source data. Background Technology

[0002] In recent years, an increasing number of floods and flood-related disasters, such as typhoons, landslides, and mudslides, have caused catastrophic damage to many cities worldwide. Urban flooding is becoming increasingly severe in many cities, with its impact gradually expanding from localized areas to drainage networks and river systems. The occurrence of urban flooding is influenced by a variety of factors, including runoff generation and confluence mechanisms, drainage system construction, and urban topography. These variations mean that urban flooding can occur anywhere, exhibiting a high degree of randomness. Low-lying areas, underpasses, underground parking garages, and poorly drained neighborhoods in cities are prone to flooding during heavy rains, becoming the hardest-hit areas. When heavy rainfall exceeds the capacity of drainage systems, rainwater cannot be discharged in time, leading to water accumulation in drainage networks and further triggering surface flooding. During heavy rains, water levels in urban rivers rise, potentially overflowing when they exceed their capacity, inundating surrounding areas. Due to the randomness and complexity of urban flooding, any urban area can be affected, making disaster forecasting extremely challenging, with accuracy and timeliness difficult to guarantee.

[0003] Currently, the methods used in urban flood location are as follows:

[0004] Traditional rule-based methods typically determine the likelihood of flooding based on predefined rules and thresholds; for example, if rainfall exceeds a certain threshold, it is considered a flood. While this method is simple and easy to implement, it struggles to accurately predict and respond to complex geographical and meteorological changes due to the volatile environment and insufficient data.

[0005] Statistical model-based methods use statistical models to predict the probability of flooding, utilizing both historical and real-time data. Their advantage lies in considering historical data and trends, but they may be insufficiently accurate for nonlinear relationships and complex geographical environments. The predictive accuracy of statistical models can be significantly reduced when the terrain is complex or rainfall distribution is uneven.

[0006] Remote sensing data-based methods monitor surface conditions using satellite imagery, such as detecting waterlogged areas or changes in river levels. Its advantages include good real-time performance and wide coverage. However, the acquisition of remote sensing data is limited by factors such as cloud cover and sensor resolution, leading to discontinuous or incomplete data.

[0007] Furthermore, single data sources, such as remote sensing data, often fail to provide comprehensive analysis and prediction when faced with complex flood scenarios. While remote sensing data can reflect surface conditions in real time, it lacks the integration and analysis of information from multiple aspects, including underground drainage systems, topographic changes, and historical rainfall data. This makes it difficult to accurately predict and locate the specific location and severity of floods, and fails to fully capture the multiple factors and dynamic changes that contribute to flood formation, thus affecting the accuracy of early warnings and the effectiveness of prevention and control measures.

[0008] Therefore, existing flood location and prediction methods still suffer from low prediction accuracy. Summary of the Invention

[0009] This invention is made to solve the above-mentioned problems, and its purpose is to provide a system for locating and displaying urban subway flood anomalies by integrating multi-source data.

[0010] This invention provides a system for locating and displaying urban subway flood anomalies by integrating multi-source data. The system comprises: a data acquisition module for real-time acquisition of rainfall, meteorological, and underground drainage data; a data storage module for storing existing topographic data, as well as rainfall, meteorological, and underground drainage data, as feature data; a factor analysis module for analyzing the feature data to identify multiple major factors contributing to urban flooding; a flood prediction module storing a flood prediction model for constructing input data based on the data corresponding to each major factor contributing to urban flooding in the feature data, and inputting the input data into the flood prediction model to obtain the predicted flood location and corresponding predicted severity; an urban flooding prediction module for calculating the predicted flood range and predicted water depth for each predicted flood location based on the corresponding feature data, as the predicted flood prediction result; a result location module storing subway route map data for calculating the affected subway stations based on the flood prediction result and the subway route map data; and a display module for displaying the affected stations.

[0011] The urban subway flood anomaly location and display system provided by the present invention, which integrates multi-source data, may also have the following features: the factor analysis module includes: a correlation analysis unit, used to perform correlation analysis of the corresponding factors in the feature data on the flooding event, and obtain multiple related factors; and a multicollinearity test unit, used to perform multicollinearity test on all related factors, and screen out multiple related factors as the main factors of flooding.

[0012] The urban subway flood anomaly location and display system provided by this invention, which integrates multi-source data, may also have the following features: the correlation analysis of waterlogging events is performed by statistical methods or tree-based methods. The statistical methods include linear correlation calculation and chi-square test. The tree-based methods calculate the importance scores of each factor based on a trained random forest model or XGBoost model, and select the factors with large importance scores as the relevant factors.

[0013] The urban subway flood anomaly location and display system that integrates multi-source data provided by the present invention may also have the following features: the flood prediction module includes: an input data generation unit, which is used to preprocess, standardize and transform the data in sequence to generate an input matrix as input data; and a first prediction unit, which stores a flood prediction model and is used to input the input data into the flood prediction model to obtain the flood prediction location and the corresponding prediction severity.

[0014] The urban subway flood anomaly location and display system provided by this invention, which integrates multi-source data, may also have the following features: the flood prediction model includes: a convolutional layer, used to perform multiple convolution operations on the input data to extract local features; a pooling layer, used to aggregate local features to obtain global features; a fully connected layer, used to perform a fully connected operation on local features and global features to generate a feature vector; and an output layer, used to process the feature vector to obtain the flood prediction location and the corresponding prediction severity.

[0015] The urban subway flood anomaly location and display system provided by this invention, which integrates multi-source data, may also have the following features: the waterlogging prediction module includes: a runoff simulation unit, used to calculate surface runoff simulation data based on meteorological and topographic data; a runoff simulation unit, which stores an underground drainage model constructed based on underground drainage data, used to calculate groundwater flow and drainage simulation data based on the surface runoff simulation data and the underground drainage model; and a waterlogging and water accumulation calculation unit, used to calculate the groundwater flow and drainage simulation data and the surface runoff simulation data to obtain waterlogging prediction results.

[0016] The urban subway flood anomaly location and display system provided by the present invention, which integrates multi-source data, may also have the following features: the result location module includes: a data matching unit, which stores subway line map data and is used to perform coordinate matching between the flood prediction results and the subway line map data to obtain a matching result; a location unit, which is used to calculate the distance between each flood prediction range in the matching result and the nearest subway station using a nearest neighbor algorithm; and a disaster screening unit, which stores a distance threshold and is used to identify subway stations with a distance less than the distance threshold as disaster-stricken stations.

[0017] The urban subway flood anomaly location and display system that integrates multi-source data provided by the present invention may also have the following features: an early warning module, used to generate early warning information based on the disaster-affected stations and the results of urban flooding prediction, and send it to a designated recipient.

[0018] The urban subway flood anomaly location and display system that integrates multi-source data provided by the present invention may also have the following features: a disposal module, wherein the data storage module is also used to store the affected stations, flood prediction results, flood response measures and processing results, and the disposal module is used to perform in-depth analysis on the data stored in the data storage module to generate an analysis report, which is used to optimize emergency plans and understand historical flood situations.

[0019] The role and effect of invention

[0020] According to the urban subway flood anomaly location and display system integrating multi-source data of the present invention, firstly, the factor analysis module extracts the main factors closely related to urban flooding from topographic data, rainfall data, meteorological data, and underground drainage data for flood prediction; secondly, the flood prediction model analyzes and processes the data corresponding to each main factor of urban flooding to obtain accurate flood prediction locations; thirdly, the urban flooding prediction module simulates the underground drainage system using a one-dimensional model and the surface runoff using a two-dimensional model for each flood prediction location, generating accurate urban flooding prediction results through numerical simulation; fourthly, the result location module matches and filters the urban flooding prediction results with subway station information to identify subway stations prone to flooding as disaster-stricken stations, which are then displayed through the display module. Therefore, the urban subway flood anomaly location and display system integrating multi-source data of the present invention can accurately predict urban flooding and mark and display disaster-stricken subway stations. Attached Figure Description

[0021] Figure 1 This is a block diagram of the subway flood prediction system in an embodiment of the present invention;

[0022] Figure 2 This is a block diagram of the flood prediction model in an embodiment of the present invention;

[0023] Figure 3 This is a schematic diagram of the process for locating and displaying abnormal urban subway flooding in an embodiment of the present invention. Detailed Implementation

[0024] To make the technical means, creative features, objectives and effects of the present invention easy to understand, the following embodiments, in conjunction with the accompanying drawings, specifically illustrate the urban subway flood anomaly location and display system that integrates multi-source data.

[0025] This embodiment provides a system for locating and displaying urban subway flood anomalies by integrating multi-source data, hereinafter referred to as the subway flood prediction system, which is used to predict subway floods.

[0026] Figure 1 This is a block diagram of the subway flood prediction system in an embodiment of the present invention.

[0027] like Figure 1 As shown, the subway flood prediction system 100 includes a data acquisition module 11, a data storage module 12, a factor analysis module 13, a flood prediction module 14, an urban flooding prediction module 15, a result location module 16, a display module 17, an early warning module 18, a response module 19, and a control module 20 that controls the operation of the above modules.

[0028] The data acquisition module 11 is used to collect real-time rainfall data, meteorological data, and underground drainage data.

[0029] In this embodiment, by installing rainfall monitoring stations, groundwater level sensors, and using field water accumulation detection technology, the data acquisition module 11 collects rainfall and drainage network hydraulic data in real time. It also integrates the above data through a GIS system using various technical means such as field water accumulation detection data analysis, on-site investigation, and online public opinion, and establishes surface-ground runoff dataset and inundation dataset as stormwater data.

[0030] In this embodiment, the data acquisition module 11 obtains information such as rainfall, wind speed, temperature, and humidity as meteorological data by accessing the real-time data interface of the National Meteorological Administration, and updates the meteorological data in real time according to the preset automated data update mechanism.

[0031] In this embodiment, the data acquisition module 11 collects the design and operation data of the subway drainage system, including the subway design drainage volume, pump station location, pipeline capacity, flow rate, etc., as underground drainage data.

[0032] The data storage module 12 is used to store existing topographic data, as well as rainfall and flood data, meteorological data, and underground drainage data, as feature data. The data storage module 12 is also used to store disaster-affected sites, urban flooding prediction results, flood response measures, and processing results.

[0033] In this embodiment, the terrain data includes subway data and urban data. Subway data consists of detailed terrain data collected along subway lines using LiDAR technology, including elevation, slope, and geological features. Urban data comprises land cover information extracted using radar imagery, drone aerial photography, etc., including building distribution and vegetation cover.

[0034] The factor analysis module 13 is used to analyze the feature data and obtain multiple main factors of urban flooding. In this embodiment, after generating the main factors of urban flooding, the factor analysis module 13 can continue to use these main factors until the type of collected data changes or the flood prediction model needs further optimization.

[0035] The factor analysis module 13 includes a correlation analysis unit 131 and a collinearity test unit 132.

[0036] The correlation analysis unit 131 is used to perform correlation analysis on the various factors in the feature data for the waterlogging event, and obtain multiple related factors.

[0037] Among them, the correlation analysis of waterlogging events is carried out using statistical methods or tree-based model methods.

[0038] Statistical methods include linear correlation calculation and chi-square test. In the linear correlation calculation, the correlation coefficient between each continuous feature and waterlogging is calculated, and the feature with the larger absolute value is selected as the relevant factor. For the chi-square test, the continuous variable is discretized, and the feature with the smaller p-value is selected as the relevant factor.

[0039] The tree-based approach calculates the importance score for each factor based on a trained random forest or XGBoost model, and selects the factor with the highest importance score as the relevant factor.

[0040] In this embodiment, the correlation analysis of urban flooding events includes correlation analysis of urban flooding elements, temporal continuity analysis of rainfall and flood, and spatial continuity analysis of rainfall and flood.

[0041] Correlation analysis of waterlogging factors involves spatial and temporal registration of features, integrating historical waterlogging events and factors into a single data framework, and calculating correlation coefficients. Alternatively, scatter plots or regression diagrams can be used to visually demonstrate the correlation between waterlogging factors and waterlogging events, thereby identifying factors that have a significant impact on waterlogging as relevant factors.

[0042] Rainfall and flood temporal continuity analysis collects rainfall and flood data at different time steps, such as hours or days, and uses time series analysis methods to calculate the temporal continuity of rainfall and flood events, obtaining the temporal patterns of rainfall and flood occurrences and their corresponding factors as correlation factors. Time series analysis methods include the autocorrelation function (ACF) and the partial autocorrelation function (PACF), used to detect the periodicity of rainfall and flood events, such as seasonality and diurnal periodicity.

[0043] Rainfall spatial continuity analysis collects rainfall and flood data from different spatial locations and uses spatial interpolation methods such as Kriging interpolation to analyze the spatial distribution of rainfall and flood events, identify the risk of urban flooding in specific areas and its corresponding factors as relevant factors.

[0044] Multicollinearity test unit 132 is used to perform multicollinearity test on all relevant factors and screen out multiple relevant factors as the main factors of waterlogging.

[0045] This embodiment uses the variance inflation factor (VIF) to detect multicollinearity. Each relevant factor is treated as a variable, and the VIF value is calculated. A higher VIF value indicates stronger collinearity between that variable and other variables. In this embodiment, a VIF > 10 indicates severe multicollinearity. When severe multicollinearity is present, indistinguishable variables are removed or variable transformations such as principal component analysis are performed to eliminate the collinearity problem. The remaining variables then represent the main factors contributing to urban flooding. This embodiment uses a collinearity test unit 132 to check for multicollinearity among relevant factors, ensuring the independence of each factor and avoiding model overfitting.

[0046] The flood prediction module 14 stores a flood prediction model, which is used to construct input data based on the data corresponding to the main factors of waterlogging in the feature data, and input the input data into the flood prediction model to obtain the flood prediction location and the corresponding prediction severity.

[0047] The flood prediction module 14 includes an input data generation unit 141 and a first prediction unit 142.

[0048] The input data generation unit 141 is used to preprocess, standardize and transform the data in sequence to generate an input matrix as input data.

[0049] In this embodiment, the input data is a multi-channel input matrix composed of stacked feature maps generated from various data sources. For example, terrain data, rainfall data, and meteorological data are aligned on the same spatial and temporal scales. For spatial data such as terrain and pipeline information, gridding is used to match them with temporal data such as rainfall and meteorological data. The multi-source data are combined into a complete input matrix, where each feature vector represents a combined state of time and spatial location.

[0050] The first prediction unit 142 stores a flood prediction model, which is used to input input data into the flood prediction model to obtain the predicted flood location and the corresponding predicted severity. In this embodiment, the flood prediction model is constructed based on a convolutional neural network (CNN).

[0051] Figure 2 This is a block diagram of the flood prediction model in an embodiment of the present invention.

[0052] like Figure 2 As shown, the flood prediction model 200 includes a convolutional layer 21, a pooling layer 22, a fully connected layer 23, and an output layer 24.

[0053] Convolutional layer 21 is used to perform multiple convolution operations on the input data to extract local features. In this embodiment, the convolution operations of convolutional layer 21 help to capture spatial patterns and local features in the input data, such as: terrain changes, identifying steep slopes and depressions in the terrain; rainfall distribution, analyzing the distribution of rainfall in different areas; and water flow paths, capturing the flow paths of water on the terrain.

[0054] Pooling layer 22 is used to aggregate local features to obtain global features. In this embodiment, the main steps of pooling layer 22 include: feature aggregation, which aggregates local features into more representative global features; computation optimization, which reduces the computational load of subsequent layers and improves the training and prediction efficiency of the model; and prevention of overfitting, which reduces the complexity of the model through dimensionality reduction operations to prevent overfitting.

[0055] Fully connected layer 23 is used to perform a fully connected operation on local and global features to generate feature vectors.

[0056] Output layer 24 is used to process the feature vectors to obtain the predicted flood locations and corresponding predicted severity.

[0057] In this embodiment, the output layer 24 generates a probability distribution using the Softmax function or outputs a continuous risk score using a linear activation function. The output layer 24 generates flood prediction levels for each area in the city based on the feature vectors, and, according to a preset level threshold, identifies areas exceeding the level threshold as flood prediction locations, and assigns the corresponding flood prediction level as the predicted severity.

[0058] The waterlogging prediction module 15 is used to calculate the waterlogging prediction range and the predicted waterlogging depth of each area for each flood prediction location based on the corresponding feature data, as the waterlogging prediction result.

[0059] The waterlogging prediction module 15 includes a runoff generation simulation unit 151, a runoff confluence simulation unit 152, and a waterlogging and water accumulation calculation unit 153.

[0060] The runoff simulation unit 151 is used to calculate surface runoff simulation data based on meteorological and topographic data. In this embodiment, the runoff simulation unit 151 calculates the runoff per unit area based on rainfall and surface characteristics, including the SCS-CN method (Curve Number method), which calculates runoff based on rainfall according to land use and soil type, and the Green-Ampt method, which simulates surface infiltration and runoff during rainfall based on soil permeability.

[0061] The runoff simulation unit 152 stores a groundwater drainage model constructed based on groundwater drainage data. This model is used to calculate groundwater flow and drainage simulation data based on surface runoff simulation data and the groundwater drainage model. In this embodiment, the runoff simulation unit 152 calculates the flow and convergence of runoff after rainfall in the urban drainage system. This includes pipe network simulation, i.e., calculating runoff flow in pipes using the Manning equation or dynamic wave equation based on the topology of drainage pipes, and surface flow equations such as the shallow water equation, simulating the diffusion and flow of surface water under different terrain conditions.

[0062] The waterlogging and water accumulation calculation unit 153 is used to calculate the waterlogging prediction results based on groundwater flow and drainage simulation data and surface runoff simulation data. In this embodiment, the waterlogging prediction results include the water accumulation depth of each area.

[0063] The result positioning module 16 stores subway line map data, which is used to calculate each subway station affected by the flood as the disaster site based on the flood prediction results and the subway line map data.

[0064] The result location module 16 includes a data matching unit 161, a location unit 162, and a disaster screening unit 163.

[0065] The data matching unit 161 stores subway route map data and is used to perform coordinate matching between the flooding prediction results and the subway route map data to obtain a matching result. In this embodiment, the data matching unit 161 matches the flooding prediction results with the subway route map data so that the two have the same coordinate system, such as WGS84 or UTM. Then, GIS software is used to overlay the flooding range, water depth, etc. from the flooding prediction results onto the subway route map to obtain the matching result.

[0066] The positioning unit 162 is used to calculate the distance between each flood prediction range in the matching results and the nearest subway station using the nearest neighbor algorithm.

[0067] The disaster screening unit 163 stores a distance threshold, which is used to identify subway stations with a distance less than the distance threshold as disaster sites.

[0068] Display module 17 is used to display the disaster-affected sites.

[0069] In this embodiment, the display module 17 marks the locations of floods on the electronic map of the subway management system and updates it dynamically to display the specific locations of abnormal events. Furthermore, to enhance user experience and readability, the electronic map employs various visualization effects. For example, markers of different colors and shapes are used to indicate the severity and type of flooding; blue indicates mild flooding, and red indicates severe flooding. It also supports heatmap functionality, using color gradients to display the distribution density and impact range of floods, allowing managers to understand the flood situation more intuitively.

[0070] Users can click on markers to view detailed information displayed in the information window, such as water depth, affected area, and possible causes. The information window is not merely a static display of information; it also offers interactive functionality. Users can access more relevant information via links or buttons within the window, such as viewing historical data, downloading detailed reports, or issuing emergency commands directly through the system. This design enhances the system's usability and user experience, enabling managers to make decisions and handle emergencies more efficiently.

[0071] The early warning module 18 is used to generate early warning information based on the disaster-affected sites and the results of urban flooding prediction, and send it to the designated recipients.

[0072] In this embodiment, the early warning information includes the time, location, severity, estimated impact range, and suggested response measures for the flood. The early warning module 18 sends information via SMS, email, and app push notifications. Designated recipients include subway management personnel, emergency rescue teams, and relevant government departments.

[0073] The response module 19 is used to perform in-depth analysis of the data stored in the data storage module and generate analysis reports. These reports are used to optimize emergency response plans and to understand historical flood data.

[0074] In this embodiment, the handling module 19 analyzes the frequency, seasonality, geographical distribution, and effectiveness of different response measures of floods, and generates analysis reports in the form of charts, reports, and maps.

[0075] The control module 20 stores the control program that controls the operation of each module.

[0076] The following description, in conjunction with the accompanying drawings, illustrates the process of using the Metro Flood Prediction System 100 to locate and display anomalies in urban metro flooding.

[0077] Figure 3 This is a schematic diagram of the process for locating and displaying abnormal urban subway flooding in an embodiment of the present invention.

[0078] like Figure 3As shown, the process of using the subway flood prediction system 100 to locate and display urban subway flood anomalies includes the following steps:

[0079] Step S1: Data acquisition module 11 is used to collect rainwater and flood data, meteorological data and underground drainage data in real time.

[0080] Step S2: The flood prediction module 14 constructs input data based on the data corresponding to the main factors of waterlogging in the feature data, and inputs the input data into the flood prediction model to obtain the flood prediction location and the corresponding prediction severity.

[0081] Step S3: The waterlogging prediction module 15 is used to calculate the waterlogging prediction range and the predicted water depth of each area for each flood prediction location based on the corresponding feature data, and the waterlogging prediction result is obtained.

[0082] Step S4: The result positioning module 16 calculates the subway stations affected by the flood based on the flood prediction results and the subway line map data, and identifies the affected stations.

[0083] Step S5: Display the affected sites using display module 17.

[0084] The role and effect of the embodiments

[0085] According to the urban subway flood anomaly location and display system involving multi-source data in this embodiment, firstly, the factor analysis module extracts the main factors closely related to urban flooding from topographic data, rainfall data, meteorological data, and underground drainage data for flood prediction; secondly, the flood prediction model analyzes and processes the data corresponding to each main factor of urban flooding to obtain accurate flood prediction locations; thirdly, the urban flooding prediction module simulates the underground drainage system using a one-dimensional model and surface runoff using a two-dimensional model for each flood prediction location, generating accurate urban flooding prediction results through numerical simulation; fourthly, the result location module matches and filters the urban flooding prediction results with subway station information to identify subway stations prone to flooding as disaster-stricken stations, which are then displayed through the display module. In summary, this method can accurately predict urban flooding and mark and display disaster-stricken subway stations.

[0086] Those skilled in the art should understand that this invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to this invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A city subway water disaster anomaly positioning and display system that fuses multi-source data, characterized in that, The system comprises: a data acquisition module for acquiring rain flood data, meteorological data and underground drainage data in real time; a data storage module for storing existing terrain data, the rain flood data, the meteorological data and the underground drainage data as feature data; a factor analysis module for analyzing the feature data to obtain a plurality of main factors of waterlogging; a water disaster prediction module storing a water disaster prediction model, for constructing input data according to the data corresponding to each of the main factors of waterlogging in the feature data, and inputting the input data into the water disaster prediction model to obtain water disaster prediction locations and corresponding prediction severity; a waterlogging prediction module for calculating each of the water disaster prediction locations according to the corresponding feature data to obtain a waterlogging prediction range and a predicted water depth of each region of the water disaster prediction location as waterlogging prediction results; a result positioning module storing subway line map data, for calculating each subway station affected by water disasters as an affected station according to the waterlogging prediction results and the subway line map data; a display module for displaying the affected station. The waterlogging prediction module comprises: a runoff simulation unit for calculating surface runoff simulation data according to the meteorological data and the terrain data; a confluence simulation unit storing an underground drainage model constructed according to the underground drainage data, for calculating underground water flow and drainage simulation data according to the surface runoff simulation data and the underground drainage model; a waterlogging and waterlogging calculation unit for calculating the underground water flow and drainage simulation data and the surface runoff simulation data to obtain the waterlogging prediction results.

2. The urban subway flood abnormal positioning and display system of fused multi-source data according to claim 1, The factor analysis module comprises: a correlation analysis unit for performing waterlogging event correlation analysis on each of the corresponding factors in the feature data to obtain a plurality of related factors; a multicollinearity test unit for performing multicollinearity test on all the related factors to select a plurality of the related factors as the main factors of waterlogging. The water disaster prediction module comprises: an input data generation unit for sequentially pre-processing, standardizing and data transforming the data to generate an input matrix as the input data; wherein a first prediction unit storing the water disaster prediction model, for inputting the input data into the water disaster prediction model to obtain the water disaster prediction locations and the corresponding prediction severity. The water disaster prediction model comprises: ​ 4. The urban subway water disaster abnormal positioning and display system of fused multi-source data according to claim 1, ​ ​ ​ ​ 5. The urban subway water disaster abnormal positioning and display system of fused multi-source data according to claim 1, ​ ​ The convolutional layer is configured to perform a plurality of convolution operations on the input data to extract local features. The pooling layer is configured to aggregate the local features to obtain global features. The fully connected layer is configured to perform a fully connected operation on the local features and the global features to generate a feature vector. The output layer is configured to process the feature vector to obtain the water disaster prediction location and the corresponding prediction severity.

6. The urban subway flood abnormal positioning and display system of fused multi-source data according to claim 1, The result positioning module comprises: The data matching unit stores subway line map data and is configured to perform coordinate matching on the waterlogging prediction result and the subway line map data to obtain a matching result. The positioning unit is configured to calculate distances between each waterlogging prediction range in the matching result and the nearest subway station by using a nearest neighbor algorithm. The disaster-affected station screening unit stores a distance threshold and is configured to screen the subway stations with distances less than the distance threshold as the disaster-affected stations. The early warning module is configured to generate early warning information according to the disaster-affected stations and the waterlogging prediction result and send the early warning information to designated receiving objects.

7. The urban subway water disaster abnormal positioning and display system of fused multi-source data according to claim 1, characterized in that, The disposal module comprises: The data storage module is further configured to store the disaster-affected stations, the waterlogging prediction result, water disaster countermeasures, and processing results.

8. The urban subway water disaster abnormal positioning and display system of fused multi-source data according to claim 1, characterized in that, The disposal module is configured to perform deep analysis on the data stored in the data storage module to generate an analysis report. The analysis report is configured to optimize emergency plans and understand historical water disasters. ​ ​ ​

Images