Emergency event visualized collaboration method and terminal device

Abstract

The application provides an emergency event visualization collaboration method and a terminal device, the method comprising: obtaining historical monitoring data sets of emergency events in a target area and preprocessing to generate a standardized event feature set; training a time series prediction model according to dynamic environment parameters therein to obtain an event evolution prediction model; constructing a geographic area matching model based on static geographic parameters therein to generate a visualization layer template associated with the target area. Real-time collection of current environment data streams of the target area, classification through the event evolution prediction model to obtain event classification results, and calling corresponding visualization layer templates from the geographic area matching model according to the event classification results. Fusion of diffusion paths, intensity levels and the like in the event classification results with the called visualization layer templates to generate a dynamic visualization instruction set, and rendering and outputting an emergency event evolution trend graph through an interactive interface. The application can improve the timeliness and decision reliability of emergency response.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, in particular to an emergency event visualized collaboration method and terminal equipment. BACKGROUND

[0002] In the field of emergency event processing, the existing technology generally has the bottleneck problems of data island, disconnection between model prediction and geographical constraints, and lagging visualization response. The traditional method usually processes dynamic environment monitoring data and static geographic information separately, which makes it difficult for the prediction model to capture the physical constraints of geographical space on event diffusion. In addition, the mainstream time series prediction model (such as single LSTM) cannot simultaneously analyze the spatio-temporal coupling characteristics, and the prediction result lacks the dynamic correlation of event intensity with spatial spread, which leads to insufficient basis for emergency resource scheduling.

[0003] In the aspect of visualization, the existing GIS system mostly uses static layer rendering, which cannot dynamically adjust the display elements according to the real-time event evolution. At the same time, the traditional rendering engine has problems such as high delay and layer superposition conflict when processing multi-source heterogeneous data, which makes it difficult to support real-time interactive analysis of the emergency event evolution process. The more prominent problem is the disconnection between prediction and disposal: the model output result needs to be manually converted into geographical interpretable information, which leads to too long response delay. These problems seriously restrict the timeliness and decision reliability of emergency response. SUMMARY

[0004] The present application provides an emergency event visualized collaboration method and terminal equipment.

[0005] According to one aspect of the present application, an emergency event visualized collaboration method is provided, which comprises:

[0006] acquiring a historical monitoring data set of emergency events in a target area, and preprocessing the historical monitoring data set to generate a standardized event feature set; the standardized event feature set includes dynamic environment parameters and static geographic parameters associated with multiple emergency event types;

[0007] training a time series prediction model according to the dynamic environment parameters in the standardized event feature set to obtain an event evolution prediction model; the output of the event evolution prediction model includes the diffusion path and intensity level of the emergency event within a preset time window;

[0008] constructing a geographical area matching model based on the static geographic parameters in the standardized event feature set to generate a visual layer template associated with the target area; the visual layer template includes terrain distribution, infrastructure coordinates, and emergency resource deployment information;

[0009] Collecting the current environmental data stream of the target area in real time, classifying the current environmental data stream using the event evolution prediction model to obtain an event classification result, and calling a corresponding visualization layer template from the geographic area matching model based on the event classification result;

[0010] The diffusion path, intensity level and the called visualization layer template in the event classification result are integrated to generate a dynamic visualization instruction set, and the emergency event evolution situation map is rendered and output through the interactive interface.

[0011] According to another aspect of the present invention, there is provided a terminal device, comprising:

[0012] at least one processor;

[0013] and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the above method.

[0014] The present invention has at least the following beneficial effects:

[0015] The emergency event visualization method provided by the present invention establishes an intelligent decision-making closed loop of prediction, matching, and rendering through the collaborative processing of dynamic environmental parameters and static geographic data. In this way, dynamic environmental parameters can capture real-time evolutionary patterns such as meteorological changes and surface displacements, while static geographic parameters can accurately reflect spatial constraints such as terrain structure and resource layout. This fusion analysis of multi-dimensional parameters can overcome the limitations of traditional visualization systems that separate time and space. Furthermore, by combining the diffusion path probability distribution and intensity level time-series predictions output by the time series prediction model with customized layer templates generated by the geographic matching model, it is possible to accurately overlay disaster evolution trends with the actual geographic scene in the visualization interface. By dynamically fusing the spatiotemporal evolution characteristics of the prediction model with the static constraint characteristics of the geographic model, the generated thermal diffusion trajectory can adapt to changes in terrain undulations, and the intensity level rendering can reflect resource deployment density in real time, thus forming a complete decision-making chain of "prediction and warning - path simulation - resource allocation" in emergency command. In addition, the multi-source data fusion mechanism based on standardized feature sets enables the visualization system to maintain stable situation deduction capabilities when data collection is incomplete or there is noise interference, significantly enhancing the robustness of emergency response decisions in complex environments.

[0016] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings illustrate exemplary embodiments and constitute a part of the specification. Together with the description of the specification, they serve to explain exemplary implementation of the embodiments. The illustrated embodiments are for illustrative purposes only and do not limit the scope of the claims. Throughout the drawings, the same reference numerals designate similar, but not necessarily identical, elements.

[0018] Figure 1 A schematic diagram of an application scenario of the emergency event visualization collaboration method according to an embodiment of the present invention is shown.

[0019] Figure 2 A flowchart of a visual collaboration method for emergency events according to an embodiment of the present invention is shown.

[0020] Figure 3 A schematic diagram showing the composition of a terminal device according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0021] The following description of exemplary embodiments of the present invention is made in conjunction with the accompanying drawings, and various details of the embodiments of the present invention are included to facilitate understanding. These details should be considered as merely exemplary. Therefore, it should be appreciated by those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope of the present invention. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0022] Figure 1 A schematic diagram of an application scenario provided according to an embodiment of the present invention is shown. The application scenario includes one or more emergency monitoring devices 101, a terminal device 120, and one or more communication networks 110 that couple the one or more emergency monitoring devices 101 to the terminal device 120. The emergency monitoring device 101 may be, for example, a device that monitors data related to an emergency event, such as a sensor device in a meteorological observation station, a sensor device in an integrated air-space-ground sensor network, and the like, without limitation.

[0023] exist Figure 1 In the configuration shown, the terminal device 120 can be a server or a computer, which includes one or more components that implement the functions performed by the terminal device 120. These components may include software components, hardware components, or a combination thereof that can be executed by one or more processors. The user operating the emergency monitoring device 101 can in turn use one or more applications to interact with the terminal device 120 to utilize the services provided by these components. It should be understood that various different system configurations are possible, which may be different from the application scenario. Therefore, Figure 1 is one example of a system for implementing the various methods described herein and is not intended to be limiting.

[0024] The network 110 may be any type of network known to those skilled in the art that can support data communications using any of a variety of available protocols, including but not limited to TCP / IP, SNA, IPX, etc. By way of example only, the one or more networks 110 may be a local area network (LAN), an Ethernet-based network, a token ring, a wide area network (WAN), the Internet, a virtual network, a virtual private network (VPN), an intranet, an extranet, a blockchain network, a public switched telephone network (PSTN), an infrared network, a wireless network (e.g., Bluetooth, WIFI), and / or any combination of these and / or other networks.

[0025] The terminal device 120 may include one or more general-purpose computers, dedicated server computers (e.g., PC (personal computer) servers, UNIX servers, mid-range servers), blade servers, mainframe computers, server clusters, or any other suitable arrangement and / or combination. The computing unit in the terminal device 120 may run one or more operating systems including any of the above-mentioned operating systems and any commercially available server operating systems.

[0026] The application scenarios provided by the embodiments of the present invention may also include one or more databases 130. In certain embodiments, these databases may be used to store data and other information. For example, one or more of the databases 130 may be used to store historical data. The databases 130 may reside in various locations.

[0027] Please refer to Figure 2 The emergency event visualization collaboration method provided by the embodiment of the present invention may include the following steps:

[0028] Step S100: Acquire a historical monitoring dataset of emergency events in a target area, and preprocess the historical monitoring dataset to generate a standardized event feature set; the standardized event feature set includes dynamic environmental parameters and static geographic parameters associated with multiple emergency event types.

[0029] The historical monitoring dataset refers to a set of original data related to emergency events recorded in a target area within a past time period, and its data sources include, for example, an environmental sensor network, a geographic information system, a historical event report database, and a manual inspection log. The dynamic environmental parameter refers to an environmental factor that changes over time and has a direct impact on the evolution process of an emergency event, such as real-time wind speed, rainfall intensity, temperature gradient change, air pollutant concentration fluctuation, and crowd movement heat distribution. The static geographic parameter refers to a geographic spatial attribute that remains relatively stable over a long time scale, such as terrain elevation data, geological structure characteristics, permanent building distribution density, traffic network topology, and fixed emergency shelter coordinates. The preprocessing process includes cleaning, aligning, normalizing, and feature extraction operations on the historical monitoring dataset, such as: filling in missing values of sensors through interpolation algorithms, eliminating abnormal noise by using a sliding window filter, aligning time stamps of multi-source heterogeneous data based on event type labels, mapping different dimension parameters to a unified numerical interval by using the maximum and minimum value normalization method, and finally selecting dynamic environmental parameters and static geographic parameters that are strongly related to the evolution of emergency events through feature engineering to form a standardized event feature set. The generation of this set makes the data collected by different time, different monitoring equipment have consistency and comparability, and lays a data foundation for subsequent model training and visualization analysis. For example, for forest fire emergency events, the dynamic environmental parameters can extract the wind speed sequence of the fire point diffusion direction and the combustible humidity change curve, and the static geographic parameters include the mountain slope distribution map, the fireproof isolation belt position, and the fire water reservoir coordinates.

[0030] Step S200: training a time series prediction model according to the dynamic environmental parameters in the standardized event feature set to obtain an event evolution prediction model; the output of the event evolution prediction model includes the diffusion path and intensity level of the emergency event within a preset time window.

[0031] Exemplarily, the time series prediction model employs a long-short-term memory neural network architecture. Its input layer receives a multidimensional vector of dynamic environmental parameters arranged in time steps from a standardized event feature set. The hidden layer captures temporal dependencies and nonlinear interactions between parameters through a gating mechanism. The output layer maps the predicted event evolution within a preset time window. The preset time window refers to the future prediction duration set based on the emergency response level, such as 6 hours, 24 hours, or 72 hours. Its length is related to the event type and regional risk level. For example, the diffusion path is the spatial trajectory of the emergency event, specifically expressed as the direction and angle extending outward from the event center, the diffusion speed, and the obstructed boundary conditions. The intensity level can quantify the severity of the event through a discrete grading index, such as classifying fire intensity into levels 1-5, corresponding to different burning rates and heat radiation ranges. During training, the model optimizes a loss function using a backpropagation algorithm. The loss function is composed of the Euclidean distance error of the diffusion path and the weighted cross-entropy of the intensity level. After multiple rounds of iterations, the model can accurately capture the mapping rules between dynamic environmental parameters and event evolution trends. For example, in flood prediction scenarios, the model can predict the flood peak arrival area and flooding depth level in the next 12 hours based on real-time rainfall, river water level and soil permeability data.

[0032] Step S300: Based on the static geographic parameters in the standardized event feature set, a geographic area matching model is constructed to generate a visualization layer template associated with the target area; the visualization layer template includes terrain distribution, infrastructure coordinates and emergency resource deployment information.

[0033] For example, the geographic region matching model can employ a graph convolutional neural network architecture, where nodes represent spatially continuous geographic units within the target region, and edge weights are calculated based on inter-unit adjacency and attribute similarity. Terrain data is obtained by rasterizing a digital elevation model (DEM), with each raster unit containing elevation, slope, and aspect information. Infrastructure coordinates include the latitude and longitude locations of key buildings such as hospitals, schools, and transportation hubs, supplemented by building type, capacity, and seismic rating. Emergency resource deployment information includes fire station locations, material storage capacity, drone takeoff and landing point distribution, and communication base station coverage. The model uses a hierarchical clustering algorithm to aggregate static geographic parameters into spatially correlated geographic feature vectors. A feature fusion layer is then used to generate multi-scale visualization layer templates. For example, for an earthquake disaster scenario, a base layer could display fault zone distribution and building seismic ratings, while an overlay layer would indicate rescue team locations and medical supply distribution routes. Visualization layer templates can utilize layered rendering technology, allowing for dynamic adjustment of display levels of detail to ensure clear presentation of key geographic features at varying zoom levels.

[0034] Step S400: Collect the current environmental data stream of the target area in real time, classify the current environmental data stream through the event evolution prediction model to obtain an event classification result, and call the corresponding visualization layer template from the geographic area matching model according to the event classification result.

[0035] For example, current environmental data streams can be continuously collected by IoT devices deployed in the target area. These data include real-time temperature, humidity, and wind speed and direction data transmitted by meteorological stations, river level changes reported by hydrological monitoring stations, and crustal deformation monitoring data released by seismic networks. The event evolution prediction model samples the data stream using a sliding window, inputting the multidimensional time series data within the window into a trained neural network. The output layer uses a softmax function to calculate the probability distribution of each emergency event type, selecting the event type with a probability exceeding a threshold as the classification result. For example, if the model detects continuous heavy rainfall, river water levels exceeding the warning line, and excessive soil water saturation, it will classify the result as "Level II flood disaster." The geographic area matching model uses the event type code in the classification result to retrieve the visualization template with the highest spatial match from a pre-generated layer template library. Specifically, the template for flood disasters highlights low-lying areas, drainage network distribution, and sandbag storage points; while chemical leaks require a dedicated template that highlights the location of hazardous materials warehouses, wind roses, and decontamination station coordinates. This process achieves rapid retrieval by establishing a mapping relationship table between event types and layer elements, ensuring timely response.

[0036] Step S500: The diffusion path and intensity level in the event classification result are integrated with the called visualization layer template to generate a dynamic visualization instruction set, and an emergency event evolution situation diagram is rendered and output through an interactive interface.

[0037] For example, the dynamic visualization instruction set can consist of spatial overlay instructions, symbol rendering instructions, and animation generation instructions. The spatial overlay instructions align the geometric polygon data of the diffusion path with the geographic features in the visualization layer template. For example, overlaying a fire spread prediction area onto a forest vegetation distribution map. The symbol rendering instructions assign color gradients and icon styles to different areas based on intensity levels, such as using dark red to represent the impact area of ​​a Category 5 hurricane and adding rotating wind field symbols to map markers. The animation generation instructions render the spatial morphological changes of the event evolution frame by frame, based on time windows. The interactive interface uses WebGL technology to render the 3D scene. Users can use gestures to rotate the view, toggle layer transparency, or click on hotspots to view detailed attribute data. The resulting emergency event evolution situation map dynamically displays changes in the disaster impact area at multiple future time points, optimize resource scheduling routes, and plan evacuation routes. For example, a heat map displays areas at risk of epidemic spread within 24 hours, along with predicted bed occupancy rates at makeshift hospitals and queue times at nucleic acid testing sites. This visualization provides intuitive spatial decision support for command and control, significantly improving the efficiency of cross-departmental collaborative response.

[0038] As an embodiment, the step S100 of preprocessing the historical monitoring data set to generate a standardized event feature set may specifically include:

[0039] Step S110: interpolating and filling missing data in the historical monitoring data set, and filtering noise data according to a preset abnormality threshold to generate a cleaned monitoring data subset.

[0040] For example, missing data interpolation refers to the process of numerically estimating and filling missing parameter values ​​in historical monitoring datasets due to sensor failure or transmission interruptions. Specific implementations include linear interpolation based on temporal proximity, kriging interpolation based on spatial similarity, and multivariate regression interpolation based on feature correlation. For example, in a flood monitoring scenario, if a hydrological station is missing three consecutive hours of rainfall data due to equipment failure, the data can be interpolated using the linear trend of data from adjacent stations during the same period and the data before and after the station. The preset anomaly threshold is the parameter's reasonable range boundary, set based on historical statistical distribution or domain expert experience, and is used to identify and filter out noise data that exceeds the normal fluctuation range. Noise data filtering uses a sliding window standard deviation method, which calculates the parameter mean and standard deviation within a fixed time window. Data points that deviate from the mean by more than three standard deviations are identified as outliers and removed. For example, an abnormal record of a sudden increase in surface displacement to 500 mm / s in earthquake monitoring data is marked as noise data and removed if it exceeds the historical maximum displacement threshold of 400 mm / s for the area. The cleaned monitoring data subset eliminates data inconsistencies and interference factors through the above operations, forming a normalized data set that can be used for feature extraction. For example, in the case of forest fires, the cleaned data subset contains a continuous and complete sequence of temperature, humidity, wind speed and fire point coordinates, without abnormal mutation noise.

[0041] Step S120: extracting target characteristic parameters that are strongly correlated with the type of emergency event from the cleaned monitoring data subset; the target characteristic parameters include meteorological change rate and surface displacement in dynamic environmental parameters, and altitude and geological structure stability index in static geographic parameters.

[0042] For example, target characteristic parameters that are strongly correlated with the type of emergency event refer to physical quantities or derived indicators that can significantly distinguish different event categories and play a decisive role in the evolution of the event. The meteorological change rate is defined as the ratio of the amplitude of the change in meteorological parameters (such as wind speed, rainfall intensity, and pressure gradient) within a specific time period to the time interval. For example, in a typhoon event, the central wind speed increases from 20 m / s to 45 m / s within 24 hours, and the change rate is (45-20) / 24≈1.04 m / s². The surface displacement is obtained through satellite synthetic aperture radar interferometry technology and represents the absolute value of the horizontal or vertical displacement of the surface. For example, in an earthquake event, the vertical displacement of the surface in the epicenter area reaches 2.3 meters. Static geographic parameters, such as altitude, are extracted through digital elevation models to reflect the blocking effect of terrain on event diffusion. For example, mountains above 800 meters above sea level create a natural barrier to flood spread. Geological structural stability indicators, derived from a weighted calculation of rock shear strength, fault activity frequency, and soil permeability, are used to assess the risk of landslides. For example, a red alert is triggered when the stability index of a slope area falls below 0.5. The extraction of target characteristic parameters requires a combination of the event's physical mechanisms and the statistical characteristics of the data. For example, in debris flow warnings, the hourly rainfall change rate, slope inclination angle, and loose deposit thickness are selected as core characteristic parameters.

[0043] Step S130: Normalize the target feature parameters according to the physical constraints corresponding to each emergency event type to generate the standardized event feature set; wherein the normalization includes mapping parameters of different dimensions to a unified numerical interval and matching the corresponding nonlinear scaling function according to the event type.

[0044] Normalization aims to eliminate scale differences between characteristic parameters of different dimensions, making them comparable and suitable for model input requirements. For example, a unified numerical interval mapping can be achieved through the use of a maximum-minimum normalization method, linearly transforming the original parameters to the range [0, 1]. For example, altitude from 0-5000 meters is mapped to 0-1, and surface displacement from 0-10 meters is mapped to 0-1. Based on the physical constraints of different emergency event types, nonlinear scaling functions are used to enhance the resolution of key features. For example, for earthquakes, a logarithmic function is used to process surface displacement to suppress the impact of extreme values. For floods, a sigmoid function is used to nonlinearly transform the rate of change of rainfall intensity to highlight threshold effects. For fires, an exponential function is used to scale wind speed parameters to amplify their sensitivity to fire spread. For example, when the original value of the geological structural stability index for earthquakes is 0.3, it is mapped to the normalized interval of 0.75 after logarithmic transformation. Meanwhile, the hourly rainfall rate of 50 mm / h in floods is mapped to 0.88 after sigmoid transformation. The standardized event feature set is processed as described above to form a multidimensional feature matrix in a unified format, where each row corresponds to an event sample and each column represents a standardized feature parameter. For example, a row vector in the matrix can be expressed as [0.72, 0.65, 0.91, 0.43], which respectively correspond to the normalized meteorological change rate, surface displacement, altitude, and geological structure stability index.

[0045] As an embodiment, the step S120 of extracting target characteristic parameters that are strongly correlated with the emergency event type includes the following steps:

[0046] Step S121: Calculate the information gain of each characteristic parameter in the cleaned monitoring data subset according to the event type labels marked in the historical emergency event case library.

[0047] The information gain degree is used to quantify the contribution of a single feature parameter to the classification of emergency event types. The calculation can be based on the information entropy theory by comparing the reduction of classification uncertainty with and without the feature. The historical emergency event case library contains complete monitoring data records of labeled event types, such as 1000 earthquake events, 800 flood events, and 500 fire event data. The information entropy calculation method is: Let the entropy H(C) of the event type set C be -Σp(c)log2p(c), where p(c) is the probability of event type c in the case library; for a feature parameter F, its conditional entropy H(C|F)=Σp(f)Σp(c|f)log2p(c|f), and the information gain degree IG(F)=H(C)-H(C|F). For example, when calculating the information gain degree of the meteorological change rate, first discretize the case library according to the parameter into several intervals, count the distribution frequency of different event types in each interval, and then calculate the conditional entropy and the information gain degree. If the proportion of earthquake events in the high meteorological change rate interval is significantly higher than that of other types, the information gain degree of the feature is higher, indicating that it plays an important role in distinguishing earthquakes from other events.

[0048] Step S122: Select the meteorological change rate and the ground displacement amount with an information gain degree greater than a first threshold value from the dynamic environmental parameters as the first candidate feature set.

[0049] Exemplarily, the first threshold value is a screening criterion set according to the model performance requirement and the feature dimension limit, for example, the optimal value can be determined by cross-validation. For example, the first threshold value is set to 0.15, when the information gain degree of the meteorological change rate is 0.22 and the information gain degree of the ground displacement amount is 0.18, both of which exceed the threshold value, so they are selected into the first candidate feature set; while the information gain degree of the temperature parameter is only 0.08, so it is excluded. The first candidate feature set focuses on the time series change indicators in the dynamic environmental parameters that are strongly related to the event type, for example, in the landslide warning scenario, the hourly rainfall change rate (IG=0.21) and the slope inclination angle change amount (IG=0.17) are selected as the core dynamic features.

[0050] Step S123: Select the geological structure stability indicator and the altitude with an information gain degree greater than a second threshold value from the static geographic parameters as the second candidate feature set.

[0051] For example, the second threshold is set lower than the first threshold to retain more static features, for example, 0.10. The information gain calculation of the geological structure stability index must consider its spatial correlation with different event types. For example, in the case of earthquakes, this index shows a strong negative correlation with earthquake magnitude (IG = 0.25). The information gain of altitude in flood events is 0.12, reflecting that low-altitude areas are more vulnerable to disasters. The second candidate feature set is selected by screening key spatial attributes from the static parameters. For example, a region with a geological structure stability index of 0.35 (IG = 0.18) and an altitude of 200 meters (IG = 0.13) exceeds the second threshold and is therefore included in the candidate set.

[0052] Step S124: performing multicollinearity detection on the first candidate feature set and the second candidate feature set, deleting redundant features that have linear correlation with other features, and obtaining an optimized target feature parameter set.

[0053] For example, multicollinearity detection can evaluate the linear correlation between features by calculating the variance inflation factor (VIF). A VIF value greater than 10 indicates severe collinearity and requires the removal of redundant features. For example, the VIF of surface displacement and seismic wave propagation velocity in the first candidate feature set is 12.5, indicating a high linear correlation between the two, so seismic wave propagation velocity is removed; the VIF of altitude and slope angle in the second candidate feature set is 8.7, which is below the threshold and is therefore retained. The optimized target feature parameter set reduces the risk of model overfitting by eliminating redundant features, such as retaining the meteorological change rate, geological structure stability index, and altitude, while removing features that are collinear with surface displacement.

[0054] Step S125: matching the optimized target feature parameter set with a preset emergency event type-feature mapping table, dynamically adjusting feature weights according to the matching results, and generating final target feature parameters.

[0055] For example, a mapping table of emergency event types and features can be predefined by domain experts, for example, specifying key features and their importance weights for different event types. For example, in the mapping table for earthquake events, the weight of the geological structure stability index is 0.6, and the weight of the surface displacement is 0.4; in flood events, the weight of the meteorological change rate is 0.7, and the weight of the altitude is 0.3. During the matching process, the weight coefficients are dynamically adjusted based on the consistency between the parameters in the optimized feature set and the mapping table entries. For example, if the optimized features in a particular earthquake case include the geological structure stability index (with a mapping table weight of 0.6) and the meteorological change rate (not in the mapping table), the weight of the latter is automatically reduced to 0.1, while the weight of the former remains at 0.6. Ultimately, a weighted target feature parameter vector is generated for model training and prediction.

[0056] As an implementation, the step 200 of training the time series prediction model according to the dynamic environmental parameters in the standardized event feature set to obtain an event evolution prediction model comprises:

[0057] Step 210: dividing the dynamic environmental parameters into a training sequence and a verification sequence in chronological order, wherein the training sequence is used for model parameter updating, and the verification sequence is used for evaluating the prediction accuracy of the model.

[0058] For example, the chronological division of the dynamic environmental parameters can follow the principle of time series data dependence, ensuring that the training sequence and the verification sequence do not overlap on the time axis and maintain the continuity of event evolution. The training sequence usually accounts for 80% of the total time span, which is used for the model to learn the evolution law through historical data; the verification sequence accounts for the remaining 20%, which is used to verify the generalization ability of the model on unknown data. For example, in the typhoon event prediction scenario, 30 consecutive days of wind speed, air pressure, and sea temperature data are selected, with the first 24 days of data as the training sequence and the last 6 days of data as the verification sequence. Random sampling should be avoided during division to prevent future information leakage. The specific implementation is to strictly cut according to the time stamp sequence, for example, the dynamic environmental parameters from August 1, 2023 to August 24, 2023 are used as the training sequence, and the data from August 25 to August 30 are used as the verification sequence. Each data point in the training sequence contains multiple dynamic parameters, such as an instantaneous wind speed of 28 meters per second, an air pressure gradient of 5 hundred pascals per kilometer, and a sea temperature of 29 degrees Celsius. The verification sequence contains the true observed values of the same parameters at subsequent time points, which are used to compare with the model prediction results to calculate error indicators.

[0059] Step 220: constructing a hybrid model architecture containing a long short-term memory network and a convolutional neural network, and segmenting and sampling the training sequence through a sliding window mechanism to generate multiple time slice samples.

[0060] The Long Short-Term Memory Network (LSTM) is used to capture the long-term temporal dependencies of dynamic environmental parameters. Its hidden layer units regulate information flow through input, forget, and output gates. For example, when processing a 72-hour series of wind speeds, the LSTM can identify patterns in the correlation between periodic changes in wind speed and typhoon paths. The Convolutional Neural Network (CNN) uses a one-dimensional convolution kernel to extract local fluctuation features. For example, convolution operations on a 3-hour time window can capture sudden pressure drops. The sliding window mechanism truncates the training sequence with a fixed step size, for example, setting a window length of 24 hours and a sliding step size of 1 hour, dividing the training sequence into multiple consecutive or partially overlapping time segments. Each time segment sample contains the dynamic parameter matrix of all time points in the window. For example, a 24-hour window sample contains the wind speed sequence [20, 22, 25, …, 35] m / s, the air pressure sequence [1005, 1002, 998, …, 985] hPa, and the sea temperature sequence [28, 28.5, 29, …, 30] degrees Celsius, forming an input matrix of dimension 24×3 for feature learning by the hybrid model architecture.

[0061] Step 230: Add event type labels and intensity level labels to each time segment sample, input the hybrid model architecture for multi-task joint training, and obtain an initial prediction model.

[0062] For example, event type labels are matched based on classification annotations in the historical emergency event case library, for example, time segment samples are labeled as "typhoon," "flood," or "earthquake" types; intensity level labels use a discrete grading standard, for example, typhoon intensity is divided into levels 1-5, corresponding to central wind speed ranges of 17.2-24.4 m / s, 24.5-32.6 m / s, 32.7-41.4 m / s, 41.5-50.9 m / s, and greater than 51.0 m / s. Multi-task joint training refers to the model synchronously learning event type classification and intensity level regression tasks, which is achieved by setting a fully connected classification layer and regression layer in parallel at the end of the hybrid model architecture. For example, a 24-hour time segment sample labeled "Typhoon Type - Intensity Level 3" is input. After the LSTM-CNN hybrid model extracts the spatiotemporal features, the classification layer outputs the probability distribution of typhoons, floods, and earthquakes [0.92, 0.05, 0.03]. The regression layer outputs the intensity level prediction value 3.2, which is compared with the true label to calculate the cross entropy loss and mean square error loss to drive the update of model parameters.

[0063] Step 240: Calculate the mean square error and classification accuracy of the initial prediction model in each time segment through the verification sequence, and use an adaptive learning rate algorithm to optimize the model parameters until convergence.

[0064] The mean squared error (MSE) is used to evaluate the accuracy of the intensity level regression task, which is calculated by the mean of the squared difference between the predicted intensity level and the true level, for example, the true intensity level of a certain time segment is 4, and the model predicts the value to be 3.8, then the error of a single sample is (4-3.8)²=0.04; the classification accuracy is obtained by counting the proportion of samples that are correctly classified, for example, 92 out of 100 validation samples are correctly classified, then the accuracy is 92%. The adaptive learning rate algorithm such as the Adam optimizer dynamically adjusts the learning rate according to the first and second moments of the gradient, for example, the initial learning rate is set to 0.001, when the validation loss does not decrease for 5 consecutive epochs, the learning rate is decayed to 0.0001. The convergence condition is set to the validation loss fluctuation amplitude less than 1e-5 and lasting for 3 epochs, for example, the MSE of the typhoon intensity regression task gradually decreases from 0.25 to 0.18 and stabilizes around this value, while the classification accuracy increases from 85% to 92% and does not change significantly, the model training is completed.

[0065] Step 250: match the optimized model with the delay tolerance of real-time data stream, adjust the number of model layers and neurons according to the test results, and generate an event evolution prediction model that meets the real-time requirements.

[0066] The delay tolerance of real-time data stream refers to the maximum data processing and prediction time allowed by the system, for example, determined by the timeliness requirement of emergency response, for example, the flood prediction needs to complete the whole process from data input to result output within 10 seconds. The matching test measures the inference time and resource occupancy rate by simulating real-time data stream injection, for example, the test shows that the optimized model needs 12 seconds to process a 24-hour time segment, which exceeds the tolerance threshold. The adjustment measures include reducing the number of LSTM layers from 3 to 2 and reducing the number of CNN convolution kernels from 64 to 32, which shortens the inference time to 8 seconds. At the same time, verify the change of model accuracy, if the intensity regression MSE increases slightly from 0.18 to 0.20 but is still lower than the acceptable threshold 0.25, then the adjustment is effective. The final generated event evolution prediction model balances the computing resources and prediction accuracy, for example, it can process 5 time segment samples per second when deployed on an edge server, meeting the real-time warning demand 30 minutes before the flood peak arrives.

[0067] As an implementation manner, the training process of the hybrid model architecture further includes the following steps:

[0068] Step S2301: input the global time sequence features output by the long short-term memory network into the attention mechanism layer to generate event evolution contribution weights corresponding to each time step, and dynamically weight and sum the global time sequence features according to the contribution weights to obtain a time attention feature vector.

[0069] The attention mechanism layer uses learnable parameters to calculate the importance of each time-step feature to the current prediction objective. For example, when processing a 72-hour typhoon wind speed series, the model may assign higher weight to sudden changes in wind speed 24 hours before landfall. Specifically, the LSTM output dimension is, for example, a 24×128 global time series feature matrix. The attention layer generates a 24-dimensional weight vector [0.02, 0.03, …, 0.15], representing the contribution of each hourly feature to the predicted typhoon path. Dynamically weighted summation results in a 128-dimensional temporal attention feature vector, whose values ​​focus on the influence of key time periods. For example, a weight of 0.15 corresponds to the characteristics of the typhoon's eyewall replacement phase, highlighting key turning points in the evolution process.

[0070] Step S2302: Dimensionally align and splice the local fluctuation pattern extracted by the convolutional neural network with the temporal attention feature vector to form a spliced ​​feature matrix containing temporal global dependencies and spatial local details.

[0071] For example, the local fluctuation pattern extracted by the convolutional neural network is dimensioned to 24×64, representing the local variation characteristics of hourly data, such as a sudden drop in air pressure identified by a 3-hour convolutional window. The temporal attention feature vector is replicated and expanded to a 24×128 matrix to achieve dimensional alignment. The concatenation operation is performed along the feature channel dimension, generating a concatenated feature matrix of 24×(128+64)=24×192. This matrix simultaneously preserves long-term trends captured by the LSTM (such as the stability of a typhoon's directional movement) and short-term anomalies detected by the CNN (such as sudden storm surges). For example, the 12th-hour feature in the matrix combines the wind speed trend of the previous 12 hours with the pressure fluctuations of the current 3 hours, providing comprehensive spatiotemporal information input for the subsequent gated recurrent unit.

[0072] Step S2303: Input the concatenated feature matrix into the gated recurrent unit, control the feature transfer path by resetting the gate and updating the gate, filter out the noise data that is irrelevant to the diffusion path of the emergency event, and output the filtered high-dimensional spatiotemporal feature sequence.

[0073] The reset gate of the Gated Recurrent Unit (GRU) determines the degree of forgetting of historical features, while the update gate controls the proportion of new features incorporated. For example, when an outlier is detected in rainfall data at a certain time step due to a sensor failure, the reset gate shuts down the transmission of historical information at that time step to prevent noise from contaminating subsequent predictions. After inputting a 24×192 concatenated feature matrix, the GRU processes each time step and outputs a 24×256 high-dimensional spatiotemporal feature sequence, where the eigenvector values ​​corresponding to the time step of the anomalous data are significantly reduced. For example, in flood forecasting, false water level rise signals caused by instrument errors are suppressed, while the runoff increase caused by actual sustained rainfall is retained, ensuring the reliability of the output feature sequence.

[0074] Step S2304: Input the high-dimensional spatiotemporal feature sequence into the diffusion path prediction branch and the intensity level regression branch respectively, wherein the diffusion path prediction branch outputs the probability distribution map of each grid unit being diffusely covered through the softmax function, and the intensity level regression branch outputs the intensity level value at each moment in the preset time window through the linear activation function.

[0075] For example, the diffusion path prediction branch can map 256-dimensional features to geographic grid space, for example, dividing the target area into 1km×1km grid cells, each corresponding to an output node. The softmax function calculates the probability of each cell being covered by the event at the next time step. For example, the probability that the typhoon center will be located at grid (15,32) in the next 6 hours is 0.93. The intensity level regression branch converts the 256-dimensional features into scalar values ​​through a fully connected layer. For example, the output of the typhoon center pressure in the next 6 hours is 935 hPa, corresponding to intensity level 4. The parallel processing of the two branches enables the model to simultaneously predict the spatial diffusion range and temporal intensity changes of the event, for example, simultaneously outputting a wildfire spread direction probability map and a fire line intensity prediction value.

[0076] Step S2305: performing cross entropy calculation on the probability distribution map and the grid unit coordinates of the true diffusion path in the verification sequence, and performing mean square error calculation on the intensity level value and the level label annotated in the verification sequence, and fusing the two loss values ​​to generate a joint optimization target.

[0077] The cross-entropy loss quantifies the difference between the predicted probability distribution and the true diffusion path. For example, if the true path covers the grid (15, 32), and the model predicts a probability of 0.93 for that grid, the loss for each sample is -ln(0.93) = 0.072. The mean square error loss calculates the squared difference between the predicted intensity value and the true level. For example, if the true level is 4 and the predicted value is 3.8, the loss is (4-3.8)² = 0.04. The joint optimization objective integrates these two losses through a weighted summation. For example, setting a cross-entropy weight of 0.6 and a mean square error weight of 0.4 results in a total loss of 0.6 × 0.072 + 0.4 × 0.04 = 0.059. This design forces the model to balance spatial diffusion accuracy and intensity prediction accuracy during the optimization process, avoiding overfitting of either task.

[0078] Step S2306: According to the back-propagation error signal of the joint optimization objective, the weight parameters of the long short-term memory network, the convolutional neural network and the gated recurrent unit are synchronously adjusted until the classification accuracy of the diffusion path prediction branch and the error rate of the intensity level regression branch reach the convergence threshold at the same time.

[0079] During backpropagation, the error signal is propagated back through the diffusion path prediction branch and the intensity level regression branch to the GRU, CNN, and LSTM networks. The gradients of the parameters at each layer are calculated using the chain rule. For example, the gradient of the cross-entropy loss with respect to the GRU parameters indicates how to adjust the gating mechanism to better capture key features of the diffusion path; the gradient of the mean square error loss with respect to the CNN convolution kernel guides how to optimize the ability to detect local fluctuations. Parameters are updated using an adaptive learning rate algorithm. Model convergence is considered when the diffusion path prediction accuracy remains above 95% for five consecutive epochs and the intensity regression mean square error (MSE) is below 0.05. The optimized model achieved an F1 score of 0.92 for diffusion path prediction and a mean absolute error of 0.15 for intensity level prediction on the test set, meeting the accuracy requirements for emergency response decision-making.

[0080] As an embodiment, the step S300, constructing a geographic area matching model based on the static geographic parameters in the standardized event feature set, includes:

[0081] Step S310: converting the terrain distribution data in the static geographic parameters into a rasterized elevation map, and marking the grid cells corresponding to the infrastructure coordinates in the elevation map.

[0082] Exemplarily, the terrain distribution data can be rasterized by a Digital Elevation Model (DEM) to discretize the continuous geospatial into a two-dimensional grid matrix of fixed resolution, and each grid cell stores the elevation value of the corresponding geographical position. The grid resolution is set according to the emergency response accuracy requirement, for example, a grid size of 1 km x 1 km is adopted to generate a 300 x 200 grid matrix in the target area, and the elevation value of each cell is accurate to the order of 0.1 meters. The infrastructure coordinates are obtained based on a Geographic Information System (GIS) database, including the latitude and longitude information of key facilities such as hospitals, fire stations, and bridges, which are mapped to the grid cell index through a coordinate conversion algorithm. For example, a hospital is located at 120.35 degrees east longitude and 30.28 degrees north latitude, and after coordinate conversion, it corresponds to the 152nd row and 78th column cell of the grid matrix. This cell is marked as a "medical facility - tertiary A hospital" and is attached with the number of beds and ambulance capacity attributes. In the forest fire scenario, after the terrain distribution data is converted into a rasterized elevation map, the grid cell corresponding to the coordinates of the fire lookout tower is marked in red, forming a spatial association base of terrain and infrastructure.

[0083] Step S320: According to the historical emergency resource scheduling records, associate the emergency resource deployment information in the grid cell to generate an initial geographic layer.

[0084] Exemplarily, the historical emergency resource scheduling records are stored in a structured database, and the record content includes event type, resource type, scheduling time, quantity, and target grid cell code. The emergency resource deployment information is bound to the grid cell through spatial association operation, for example, the inventory of flood control sandbags is associated with the grid cells along the river bank, and the take-off and landing point of the fire helicopter is associated with the grid cells in the mountain area. The initial geographic layer adopts a multi-layer superposition structure, and the bottom layer is the rasterized elevation map, and the upper layer is the resource density heat map. For example, in the case of a typhoon disaster, the number of inflatable boats associated with the historical scheduling records in a certain coastal area is 50, and the inventory of life jackets is 2000, and when generating the resource distribution heat map, the color depth of this cell is proportional to the resource sufficiency rate. The resource sufficiency rate calculation formula is the ratio of the current inventory to the historical maximum demand. When the inventory of inflatable boats in a certain grid cell is 30 and the historical maximum demand is 50, the sufficiency rate is 60%, and the corresponding color coding is orange.

[0085] Step S330: Perform topological structure analysis on the initial geographic layer to identify key path nodes and potential risk areas, and add corresponding topological relationship labels.

[0086] For example, topological structure analysis can be implemented using graph theory algorithms, abstracting grid cells into nodes and inter-cell adjacency relationships into edges. Indicators such as node betweenness centrality and edge connectivity strength are calculated. Critical path nodes are grid cells essential for emergency evacuation or resource transportation, such as those located at river bridges and tunnel entrances, whose betweenness centrality exceeds a threshold of 0.7. Potential risk areas are clusters of grid cells with unstable geological structures or frequent historical disasters, such as areas with high landslide risk with slopes greater than 25 degrees or areas with liquefied soil layers. Topological relationship labels include categories such as "core hub node," "secondary transit node," and "high-risk isolation zone." For example, the grid cell containing a cross-sea bridge is labeled "core hub node - transportation network level 1," while the surrounding buffer zone cells are labeled "secondary transit node - transportation network level 2." In earthquake emergency scenarios, topological analysis can identify potential landslide areas on both sides of a fault zone. These areas are then labeled "high-risk isolation zone - geological stability level D" to restrict rescue route planning from passing through these areas.

[0087] Step S340: performing region segmentation on the initial geographic layer based on the topological relationship labels to obtain multiple sub-region templates, and configuring an independent data rendering channel for each sub-region template.

[0088] Exemplarily, regional segmentation can be achieved using the Voronoi Diagram algorithm, where Thiessen polygons are generated using key path nodes as seed points, and the grid cells covered by each polygon constitute a sub-region template. For example, sub-region templates generated using five core hospitals as seed points, each template contains 15-20 grid cells, covering a 10-kilometer radius around the hospital. The data rendering channel assigns an independent color mapping table, transparency parameters, and layer overlay order to each sub-region. For example, the "high-risk isolation zone" sub-region is filled with red translucent, and the "core hub node" sub-region is highlighted with a gold border. In flood disaster scenarios, the sub-region templates segmented along the river channel are configured with a blue gradient rendering channel to show the rising trend of the water level; the mountainous area template is configured with a gray contour rendering channel to highlight the impact of terrain on the direction of flood flow.

[0089] Step S350: Dynamically bind each sub-region template to the intensity level in the event classification result to generate a visualization layer template that can automatically adjust the display ratio according to the intensity level.

[0090] For example, dynamic binding can be achieved by establishing intensity level-display ratio mapping rules. For example, typhoon intensity levels 1-5 correspond to sub-area display ratios of 30%, 50%, 70%, 90%, and 100%, respectively. When the event classification result is "Typhoon-Intensity Level 4", only 90% of the core hub node sub-area template is displayed, hiding the edge low-risk areas; if the intensity level rises to level 5, all sub-area templates are displayed in full. Automatic adjustment of visualization layer templates is achieved through a spatial index acceleration algorithm, such as using an R-tree index to quickly retrieve the set of raster cells to be rendered. In a chemical leak incident, the sub-area with a radius of 3 kilometers around the leak source corresponding to intensity level 3 is highlighted, and the emergency resource deployment information only renders the location of the chemical defense forces within this range; when the predicted intensity rises to level 4, the display radius is expanded to 5 kilometers, and the wind rose diagram sub-area template is superimposed to indicate the direction of gas diffusion.

[0091] As an implementation method, the step S500, fusing the diffusion path and intensity level in the event classification result with the called visualization layer template to generate a dynamic visualization instruction set, includes:

[0092] Step S510: according to the diffusion path in the event classification result, a heat diffusion trajectory is drawn in the called visualization layer template, and intensity level changes are represented by color gradients.

[0093] For example, the heat diffusion trajectory can be generated based on the probability distribution of grid cells output by the diffusion path prediction model, and the inverse distance weighted interpolation algorithm is used to convert discrete probability values ​​into continuous color patches. The color gradient scheme follows industry standards, for example, red represents intensity level 5 (highest risk), orange is level 4, yellow is level 3, green is level 2, and blue is level 1. In the forest fire scenario, the grid cells covered by the diffusion path decrease from 0.9 to 0.1 according to the probability value of the fire spreading outward from the ignition point, and the corresponding color changes from dark red to light blue, forming a heat map of the spread of the fire. The trajectory drawing uses Bezier curves to smooth the transition between cells. For example, the center points of adjacent cells are connected as a curved path. The line width is proportional to the probability value. The line width of the highest probability path is set to 5 pixels, and the line width of the low probability path is 1 pixel.

[0094] Step S520: Identify the grid cells covered by the diffusion path, calculate the resource gap ratio based on the emergency resource deployment information, and generate a resource warning mark.

[0095] Exemplarily, the resource gap ratio calculation manner is, for example, (current resource inventory-predicted demand) / predicted demand x 100%, wherein the predicted demand is estimated according to the population density covered by the diffusion path and the historical per capita resource consumption. For example, a certain flood diffusion path covers 10 grid units, with a total population of 50,000 people, and according to historical data, each person needs 0.5 bottles of drinking water, and the predicted demand is 25,000 bottles; if the current inventory is only 18,000 bottles, the gap ratio is (1.8-2.5) / 2.5x100%=-28%, and a "drinking water shortage 28%" warning mark is generated. The resource warning mark is displayed in an icon superimposed manner, for example, a flashing exclamation mark icon is added above the grid unit with a gap of more than 20%, and the icon color is divided into three grades of red, orange and yellow according to the severity of the gap. In the earthquake rescue scene, a certain collapse area grid unit predicts that 50 hydraulic jacking devices are needed, and 30 devices are actually deployed, with a gap of 40%, and the warning mark is a red device icon with a "gap 40%" label.

[0096] Step S530: encode the heat diffusion track, intensity level color gradient and resource warning mark into vector graphics instructions.

[0097] Exemplarily, the vector graphics instructions are defined in a Scalable Vector Graphics (SVG) format, and the heat diffusion track is encoded as <path>Elements, attributes include d (path coordinates), stroke (color), stroke-width (line width); intensity level color gradient encoding is <lineargradient>element, defining the start and end colors and gradient direction; the resource warning mark code is <g>Combination elements, including Icon reference and <text>Label text. For example, the typhoon diffusion path instruction is <path d="M152,78 L155,80..." stroke="#FF4500" stroke-width="3" / > , the resource gap marking instruction is <g transform="translate(152,78)"> <img xlink:href="warning.png"> <text x="10" y="5"> Gap 28%< / text> < / g> The encoding process dynamically generates an instruction tree through the DOM operation library to ensure rendering efficiency and editability.

[0098] As an embodiment, the step S530 of encoding the heat diffusion trajectory, intensity level color gradient, and resource warning mark into vector graphics instructions includes:

[0099] Step S531: parse the continuous coordinate point sequence of the thermal diffusion trajectory, generate a piecewise Bezier curve control point set with transparency attenuation attributes according to the timestamp information of the diffusion path, and map the intensity level color gradient to the curve width gradient parameter to form a trajectory vector primitive that is traceable in the time dimension.

[0100] For example, the continuous coordinate point sequence of the heat diffusion trajectory consists of the latitude and longitude coordinates of the grid cell center point output by the event evolution prediction model. For example, the typhoon path prediction data includes a set of center point coordinates [(120.35, 30.28), (120.41, 30.31), …, (120.78, 30.65)] updated every hour. Timestamp information is bound to each coordinate point, recording the predicted time corresponding to the location, such as one point every hour from 09:00 to 12:00 on August 25, 2023. The transparency decay attribute can be dynamically calculated based on the interval between the timestamp and the current display time. For example, the trajectory transparency within 24 hours is set to decay linearly from 1.0 to 0.2, and data beyond 24 hours is completely transparent. The control point set of the piecewise Bezier curve can be generated, for example, by the cubic Bezier interpolation algorithm, converting adjacent coordinate points into curve segments containing two control points. For example, the curve control points between the coordinate points P0 (120.35, 30.28) and P1 (120.41, 30.31) are C0 (120.37, 30.29) and C1 (120.39, 30.30), forming a smooth path. When the intensity level color gradient is mapped to the curve width gradient parameter, the intensity level 1 is set to correspond to a line width of 2 pixels, the level 5 corresponds to a line width of 10 pixels, and the intermediate levels are interpolated according to linear proportions. The final generated trajectory vector primitive contains a coordinate sequence, transparency parameters and line width data, such as SVG path instructions <path d="M120.3530.28 C120.37 30.29 120.39 30.30 120.41 30.31…" stroke-opacity="0.8" stroke-width="5" / > , achieving visual expression that is traceable in the time dimension.

[0101] Step S532: extract the color gamut distribution range of the intensity level color gradient, discretize the continuous color levels into polygonal filling area definition instructions based on the grid unit boundary coordinates, and add a gradient color band attribute tag containing intensity value metadata to each polygon.

[0102] For example, the color gamut distribution range of the intensity level color gradient can be from red (RGB 255,0,0) to blue (RGB 0,0,255), divided into 5 equally spaced color levels. The grid cell boundary coordinates are obtained through geographic coordinate conversion. For example, the four vertex coordinates of a 1km×1km grid are (120.30,30.25), (120.30,30.26), (120.31,30.26), (120.31,30.25). The discretization process maps continuous color levels to grid cells. For example, the predicted intensity level 3 corresponds to RGB (255,165,0), and this color is filled into all grid cells marked as level 3. The fill area definition instruction uses SVG's <polygon>Elements, such as <polygon points="120.30,30.25 120.30,30.26 120.31,30.26120.31,30.25" fill="#FFA500" / > Gradient color band attribute labels are embedded with custom data attributes, such as adding data-intensity-level="3" and data-max-wind="41.4m / s", allowing the interactive interface to query intensity details. In the flood prediction scenario, grid cells with intensity level 2 are filled with light blue (RGB 173, 216, 230) and annotated with data-water-depth="1.2m", forming a complete instruction linking visualization and data.

[0103] Step S533: Decompose the icon elements of the resource warning mark into a set of basic geometric primitives, convert the warning level text into vector font outline data, and encode the material gap ratio numerical value into primitive scaling coefficient and rotation angle parameters.

[0104] For example, resource warning marker icons can adopt a standardized design. For example, the "medical supply shortage" icon is composed of a cross (two rectangles stacked orthogonally) and an exclamation point triangle. The decomposition process converts the cross into four line elements: <line x1="10" y1="5" x2="10" y2="15" / > and <line x1="5" y1="10" x2="15" y2="10" / > , the exclamation point triangle is converted to <path d="M10,0 L15,5 L5,5 Z" / > The warning level text "Level 3" is converted into vector outline data through the FreeType library to generate <path d="M20,30 L25,30…" / > A series of linear and curved instructions. The proportion of a material shortage is encoded as a scaling factor; for example, a 30% shortage causes the icon to be scaled to 130% of its original size. A rotation angle parameter dynamically adjusts the orientation of icons, for example, the wind direction indicator rotates 45 degrees to follow the predicted wind direction. In earthquake rescue scenarios, the exclamation point icon indicating a hydraulic equipment shortage is magnified 1.5 times and displayed at a 10-degree angle to enhance visual warning effectiveness.

[0105] Step S534: aligning the trajectory vector primitives, gradient color band attribute labels, and geometric primitive sets according to a spatial coordinate system, establishing a mapping relationship between a timeline control and vector element animation attributes, and generating an uncompressed original vector instruction set.

[0106] For example, the spatial coordinate system alignment can be achieved by uniformly adopting the WGS84 geographic coordinate system, and the coordinate parameters of all vector elements are converted to screen pixel coordinates. For example, the geographic coordinates (120.35, 30.28) are converted to canvas coordinates (532, 784) through the Mercator projection. The mapping relationship between the timeline control and the vector elements is implemented through JavaScript objects. For example, the transparency of the trajectory segment corresponding to the timestamp 2023-08-25T09:00:00 is defined as 0.8, and the transparency of the segment corresponding to 10:00:00 is defined as 0.7. The original vector instruction set contains unoptimized complete drawing commands, for example, it contains 5000 <polygon>Elements, 200 <path>Instructions and 150 <g>Combined elements, each element carries independent spatiotemporal attribute parameters. In a chemical leak visualization, each segment of the gas diffusion path is bound to a start time and concentration value, and sliding the timeline triggers the transparency and color changes of the corresponding segment.

[0107] Step S535: topologically optimizing the original vector instruction set, merging polygon filling instructions with the same attributes in adjacent grid cells, compressing repeated geometric primitive coordinate sequences, and generating a simplified composite vector instruction set.

[0108] For example, the topology optimization can use the Douglas-Peucker algorithm to simplify the polygon boundary coordinate points, for example, simplifying a complex polygon containing 100 vertices into 20 key points, keeping the shape error less than 1 pixel. Polygons with the same attributes are merged into a single drawing instruction, for example, 10 adjacent grid cells with intensity level 3 are merged into <path d="M120.30 30.25 H120.40 V30.30 H120.30 Z" / > , reducing the number of drawing calls. Geometric primitive coordinate sequence compression uses differential encoding. For example, the consecutive coordinate points (120.30, 30.25), (120.30, 30.26), and (120.31, 30.26) are encoded in the "+0.01 latitude + 0.0 longitude" incremental format. The optimized composite vector instruction set reduces its capacity by 60%, for example, from the original 50MB to 20MB, improving transmission and rendering efficiency. In the forest fire scene, the jagged polygons at the edge of the fire were optimized and converted into smooth curves, reducing the computational load of the GPU vertex shader.

[0109] Step S536: embedding an interactive response tag associated with the emergency resource deployment information into the composite vector instruction set, and adding a material type code and an inventory query interface to each resource warning tag.

[0110] For example, interactive response tags can be embedded in the form of HTML5 custom data attributes, such as adding data-resource-type="medical" data-inventory="1200" to the medical supplies icon element. The inventory query interface is implemented through AJAX request. When the user clicks the icon, fetch("api / inventory?type=medical&grid=152-78") is triggered to obtain real-time data. The material type code adopts the ISO standard. For example, "MED1" represents a first aid kit, and "WAT2" represents bottled drinking water. In the typhoon emergency scenario, the assault boat icon is attached with data-resource-type="boat" data-capacity="20 people". Clicking it will display the location and capacity list of all available ships within a 5-kilometer radius.

[0111] Step S537: Hierarchical indexing of the composite vector instruction set according to the rendering priority order of the visualization layer templates, to generate a final vector graphics instruction package containing layer stacking order and visibility control parameters.

[0112] Exemplarily, the rendering priority order can be set according to the visual importance rule, with the heat diffusion trajectory layer as the highest level (z-index = 1000), the resource warning mark second (z-index = 900), and the intensity level color band as the bottom layer (z-index = 800). The visibility control parameters are realized through CSS class names, for example, adding class = "geology-layer visibility-toggle" to the earthquake fault zone layer, so that the user can control the display state through the interface switch. The hierarchical indexing adopts a quadtree spatial indexing structure, for example, dividing the screen into four quadrants, and storing the vector instruction pointer of the corresponding region in each quadrant to accelerate data retrieval during local rendering. The final vector graphics instruction package is packaged in JSON format, containing layer definition, spatial indexing, and instruction data, for example, { "layers": [{"name": "heat-path", "zIndex": 1000,"commands": [...]}]} for rendering engine parsing and execution.

[0113] Step S540: Dynamically adjusting the rendering precision of the vector graphics instruction according to the resolution of the interactive interface, and adding a time axis control to realize the historical playback and prediction deduction of the evolution trend graph.

[0114] Exemplarily, the rendering precision adjustment can be realized through the Level of Detail (LOD) algorithm. When the interactive interface is zoomed to the full area view, a simplified version of the vector instruction is used (such as reducing the number of path nodes and merging adjacent color regions); when zoomed in to the grid unit level, high-precision instructions are loaded (such as displaying complete path nodes and resource annotation details). The time axis control integrates HTML5 <input type="range"> elements, binds timestamp metadata, and aligns the slider step with the prediction time window. For example, in a 72-hour typhoon prediction, the time axis is moved every 1 hour, triggering the vector instruction update function to redraw the heat diffusion trajectory and resource state at the corresponding time. The historical playback function is realized by caching the instruction set at past time points, and when the user drags the time axis to the historical position, the corresponding instructions are extracted from the cache queue for re-rendering.

[0115] Step S550: Converting the vector graphics instruction into a pixel data stream through a parallel rendering engine, and outputting it to the interactive interface for real-time refreshing.

[0116] For example, the parallel rendering engine can be implemented based on WebGL technology, parsing SVG instructions into WebGL Shader Language (GLSL), and calculating vertex coordinates and fragment colors in parallel through the graphics processing unit (GPU). For example, the vector path of the heat diffusion trajectory is converted into a triangle mesh vertex buffer, and the color gradient is generated by the fragment shader based on the UV coordinate interpolation. The pixel data stream is output through a double buffering mechanism, and the front and back buffers are swapped immediately after the current frame rendering is completed to ensure that the refresh rate is not less than 60fps. In the mobile interactive interface, the engine automatically switches to Canvas 2D rendering mode, pre-generates a static frame sequence through an off-screen canvas, and loads frame data on demand when sliding the timeline, balancing performance and visual effects. In the chemical leak simulation, the real-time refresh function receives the latest sensor data and updates the vector instructions every 5 seconds, allowing users to observe the dynamic changes in the gas diffusion boundary.

[0117] As an embodiment, the step S550 of converting the vector graphics instruction into a pixel data stream by a parallel rendering engine and outputting the pixel data stream to the interactive interface for real-time refresh may include:

[0118] Step S551: Split the vector graphics instructions into a heat diffusion trajectory vector sub-instruction set, an intensity level color gradient vector sub-instruction set, and a resource warning mark vector sub-instruction set according to the layer type.

[0119] For example, the layer splitting is based on the layer definition in the vector instruction package, for example, all layers with z-index=1000 are extracted from the JSON instruction package. <path>Elements are classified into the heat diffusion trajectory subset, with z-index = 800 <polygon>Elements are classified into the color gradient subset. Resource warning marker subset filters include data-resource-type attributes <g>Elements. In GPU memory allocation, the trajectory subset is stored in Buffer Object 0, the color gradient subset is stored in Buffer Object 1, and the marker subset is stored in Buffer Object 2, realizing data isolation and parallel access. For example, in a typhoon visualization scene, the trajectory subset contains 1200 Bezier curve instructions, the color subset contains 5000 polygons, and the marker subset contains 200 icon combinations.

[0120] Step S552: According to the spatial resolution of the current window of the interactive interface, an independent graphics rendering pipeline is allocated for each vector sub-instruction set, and the rendering tasks of each pipeline are mapped to the parallel computing units of the GPU.

[0121] Exemplarily, the graphics rendering pipeline is divided into a geometry processing pipeline and a fragment processing pipeline, the thermal diffusion trajectory subset is allocated to pipeline 0, the Bezier curve is processed using a surface subdivision shader; the color gradient subset is allocated to pipeline 1, and the polygon filling is batch processed using a compute shader; the resource marker subset is allocated to pipeline 2, and an icon instance is generated by calling a geometry shader. The GPU computing unit is divided according to the NVIDIA CUDA core architecture, and 5000 polygon filling tasks are decomposed into parallel computing tasks of 512 threads per block. At a resolution of 4K, pipeline 0 allocates 8 SM units to process trajectory anti-aliasing, and pipeline 1 occupies 12 SM units for color gamut conversion.

[0122] Step S553: Perform Bezier curve rasterization processing on the thermal diffusion trajectory vector sub-instruction set to generate a trajectory pixel array with transparency gradient properties, and dynamically clip the trajectory segments that exceed the time window according to the current progress parameter of the time axis control.

[0123] Exemplarily, the Bezier curve rasterization can use the Loop-Blinn algorithm to convert a cubic Bezier curve into a triangle strip, for example, the instruction <path d="M0,0 C10,20 30,40 50,50" / > is decomposed into 32 triangular facets. The transparency gradient is calculated by a fragment shader according to the curve parameter t value, for example, t=0 alpha=1.0, t=1 alpha=0.2. The time axis clipping function is realized by setting the stencil buffer, when the time window is set to 24 hours, only the trajectory segments with t∈[0,0.8] (corresponding to the first 24 hours of data) are retained, and the subsequent part is excluded by stencil test. In flood evolution visualization, the 72-hour predicted trajectory is clipped to the data within the current 48 hours, and the exceeding part is not involved in rasterization.

[0124] Step S554: Perform color gamut space conversion on the intensity level color gradient vector sub-instruction set to map the HSL color model to an RGB pixel matrix for screen display, and add an anti-aliasing processing color transition zone combined with the grid cell boundary data.

[0125] For example, HSL to RGB conversion can be implemented in the fragment shader, for example, converting HSL(30°, 100%, 50%) to RGB(255, 128, 0). Anti-aliasing can be performed using multi-sample anti-aliasing (MSAA 4x), calculating color coverage at four subsample points per pixel. Color transition bands are generated through linear interpolation, for example, a 2-pixel gradient is generated at the grid boundary between intensity level 3 (HSL(60°, 100%, 50%)) and level 4 (HSL(30°, 100%, 50%)). In earthquake intensity visualization, a smooth transition is added at the boundary between the IX degree zone (HSL(0°, 100%, 50%)) and the X degree zone (HSL(30°, 100%, 50%)) to avoid jagged edges on color blocks.

[0126] Step S555: Decompose the resource warning mark vector sub-instruction set into icon outline geometry data and text annotation data, use a geometry shader to generate a multi-resolution icon bitmap, and use a font rendering engine to generate an adaptive size warning text map.

[0127] For example, after the icon outline geometry data is input into the geometry shader, a Level of Dimension (LOD) model is dynamically generated based on the viewing distance. Specifically, for example, when the icon screen size is less than 16×16 pixels, a simplified 8-gon version is generated; when it is larger than 64×64 pixels, a full 128-gon version is generated. The font rendering engine uses Signed Distance Field (SDF) technology to generate a 512×512 pixel distance field map for the "28% gap" text. The fragment shader then smooths the edges when zooming. For mobile display, the icon bitmap is pre-generated in three resolutions: 256×256, 128×128, and 64×64, which automatically switch based on the zoom level.

[0128] Step S556: Detect the output delay status of each graphics rendering pipeline. When the difference in the generation progress of the track pixel matrix and the color transition band exceeds a threshold, start the pixel cache synchronization mechanism to align the spatial coordinate reference points of each layer.

[0129] For example, latency can be monitored using an OpenGL fence object (GL_FENCE), assuming a threshold of ΔT = 16ms (corresponding to 60fps). When pipeline 0's trajectory rendering takes 18ms and pipeline 1's color band rendering takes 12ms, the synchronization mechanism is triggered to pause pipeline 1's output, waiting for pipeline 0 to complete the trajectory rasterization for the current frame. Spatial coordinate reference point alignment is implemented using atomic counters, ensuring that pixel data for all layers is composited based on the same world coordinate origin (e.g., the screen center (0,0)). In the typhoon eyewall displacement animation, the coordinate system is automatically resynchronized if the offset error between the trajectory layer and the color band layer exceeds 2 pixels.

[0130] Step S557: The synchronized track pixel matrix, RGB pixel matrix and icon bitmap are layer-fused, and the display priority conflict of the overlapping area of ​​the graphic elements is resolved according to the depth buffer data to generate complete frame buffer data.

[0131] For example, layer fusion uses the alpha blending formula: C out =C src *α src +C dst *(1-α src The depth buffer test uses GL_DEPTH_TEST and sets the depth to 0.5 for the track layer, 0.7 for the icon layer, and 0.3 for the color band layer. This ensures that the color band bottom layer is visible and the icon covers the track. In the area where the medical resource icon overlaps with the flood band, the depth test ensures that the icon always appears in the foreground. The frame buffer object (FBO) stores the fused RGBA pixel data at a resolution of 3840×2160, 32-bit color depth, and occupies 31.6MB of video memory.

[0132] Step S558: cutting the frame buffer data into pixel data streams matching the vertical synchronization signal according to the refresh rate of the interactive interface, and sending them to the video memory mapping area of ​​the display device through the multi-threaded transmission channel.

[0133] For example, the vertical sync signal interval is 16.67ms (60Hz refresh rate), and frame buffer data is divided into four horizontal strips (540 lines each) and transmitted in parallel via PCIe 4.0×16 lanes. Memory mapping uses NVIDIA GPUDirect technology, writing directly to the graphics card's Frame Buffer Object. In an 8K resolution scenario, with four-way SLI parallel transmission enabled, each processing 2160×3840 pixel data, the total bandwidth requirement is 12.8GB / s.

[0134] Step S559: monitor the view transformation parameters triggered by user interaction events in real time. When a zoom or pan operation is detected, interrupt the current pixel data stream transmission and give priority to rendering the low-precision preview frame in the center area of ​​the viewport.

[0135] For example, the view transformation parameters can be captured through the WPF Render Transform matrix, for example, setting the zoom factor to be greater than 1.5 triggers the low-precision mode. The preview frame assumes binary downsampling, compresses the 4K frame buffer to a resolution of 1024×576, and sets the texture filtering mode to NEAREST. The central area is defined as the central 50% of the viewport, and computing resources are allocated first to render high-precision content in this area. When the user zooms in and out of the map quickly, the system prioritizes displaying the downsampled overview of the full map, and progressively loads high-definition details after the operation stops.

[0136] Step S5510: When network bandwidth fluctuations cause data transmission delays, a motion blur compensation image is generated based on historical frame buffer data to fill the display abnormal area until the new pixel data stream completely covers the current window.

[0137] For example, motion blur compensation can generate a dynamic blur effect by calculating the pixel displacement vector through the Optical Flow of the previous two frames. For example, in real-time typhoon path tracking, when the new data is delayed by 200ms, based on the previous frame position (120.50, 30.40) and the speed of 25km / h, it is predicted that the current arrival time is (120.53, 30.42), and a directional blur filter is added to this area. Abnormal area detection compares the SSIM structural similarity of the frame sequence, and areas with a difference of more than 0.3 trigger compensation rendering. The compensated image is compressed using JPEG-XS encoding, and the bit rate is reduced to 30% of the original data, giving priority to ensuring the visual continuity of key areas.

[0138] As an implementation manner, the method provided in the embodiment of the present invention may further include the following steps:

[0139] Step S600: In response to an area selection instruction triggered by the user through the interactive interface, a target sub-area is intercepted from the visualization layer template according to the coordinate range contained in the area selection instruction, and the terrain distribution, infrastructure coordinates and emergency resource deployment information corresponding to the target sub-area are extracted.

[0140] Exemplarily, the area selection command captures the geographic coordinate range entered by the user through the rectangular selection tool or free polygon drawing tool of the interactive interface, for example, the user selects a rectangular area of ​​120.30°-120.45° east longitude and 30.25°-30.35° north latitude on the map interface. The system extracts the corresponding rasterized elevation map subset from the visualization layer template based on the coordinate range, for example, intercepting the sub-matrix with row numbers 150-200 and column numbers 80-120 in the raster matrix as the target sub-area. The terrain distribution data extraction includes the elevation value, slope and aspect parameters within the sub-area, such as a continuously changing gradient from 50 meters to 300 meters above sea level. The infrastructure coordinate extraction covers the longitude and latitude positioning information of all hospitals, fire stations and transportation hubs within the selected range. For example, a tertiary hospital is located at 120.38° east longitude and 30.30° north latitude, and its grid cell index is (175,95). Emergency resource deployment information is linked to the coordinates of various infrastructure facilities. For example, the hospital's corresponding medical supply inventory includes 1,200 first aid kits and 800 bags of plasma. In a forest fire emergency scenario, the target subregion extracts the coordinates of the fire isolation zone within a 10-kilometer radius of the fire site and the capacity of the fire water reservoir, forming a basic spatial dataset for emergency decision-making.

[0141] Step S700: superimpose the terrain distribution of the target sub-region with real-time meteorological parameters in the current environmental data stream to generate an enhanced terrain layer with wind speed vector arrows and precipitation intensity contour lines.

[0142] Illustratively, real-time meteorological parameters can be updated by Internet of Things sensors at a frequency of seconds, including wind speed, wind direction, precipitation, and temperature data. The spatial superposition of terrain distribution data and meteorological parameters uses a grid cell binding method, for example, each 1km x 1km grid cell is associated with current wind speed 12m / s, wind direction 270°, and hourly precipitation 15mm meteorological observation values. Wind speed vector arrows are generated by SVG <marker>Elements are implemented, with arrow length proportional to wind speed and direction aligned with wind angle. For example, a wind speed of 12m / s corresponds to an arrow length of 40 pixels, and a wind direction of 270° points due west. Precipitation intensity contours are generated based on the inverse distance weighted interpolation algorithm, converting discrete precipitation observations into continuous contour lines. For example, three contour lines of 5mm, 10mm, and 15mm are drawn in the target sub-area, with line width increasing with intensity. The enhanced terrain layer integrates elevation color scales, vector arrows, and contour lines. For example, in the 3D rendering view, the area at an altitude of 300 meters is displayed in dark brown, with the northwest wind speed arrow and the 15mm contour line outlined in red and bold, forming a visual foundation for the collaborative expression of multi-dimensional data.

[0143] Step S800: Locate the building locations affected by the diffusion path in the enhanced terrain layer according to the infrastructure coordinates, retrieve the material inventory in the associated emergency resource deployment information, and generate a resource distribution overlay map with icon size indicating inventory adequacy.

[0144] For example, buildings affected by the diffusion path are identified using a spatial intersection algorithm. For example, grid cells covered by the heat diffusion trajectory are compared with the hospital's coordinates for a point-to-surface containment test. If the probability of diffusion in the grid cell containing the hospital exceeds 50%, the cell is marked as affected. Material inventory adequacy is calculated as the ratio of current inventory to historical peak demand. For example, a fire station with 200 fire extinguishers and a historical peak demand of 300 fire extinguishers has an adequacy of 66.7%. Icon size mapping rules set an adequacy of 80% or higher to the original size, and reduce the size by 15% for every 10% decrease. For example, an icon with an adequacy of 66.7% is scaled to 77.5% of its original size. The resource distribution overlay uses a layered symbology. For example, hospital icons are represented by a cross with a gradient size from 30×30 pixels to 15×15 pixels; fire stations are represented by a flame icon, whose size dynamically adjusts based on the fire extinguisher inventory. In the event of a chemical leak, the gas mask inventory icon for the affected chemical plant is reduced to 60% of its original size, with a red exclamation mark superimposed below the icon.

[0145] Step S900: Based on the spatial intersection area of ​​the resource distribution coverage map and the heat diffusion trajectory, calculate the redundant coordinate points of emergency resources not covered by the diffusion path, and convert them into resource supply path planning points that can be dynamically scheduled.

[0146] Exemplarily, the spatial intersection area can be determined by the logical AND operation of grid cells. For example, when the grid cell where the icon is located in the resource distribution coverage map is not covered by any cell with a probability of more than 30% in the thermal diffusion trajectory, it is determined to be a redundant resource point. After the redundant coordinate points are extracted, they are converted into a set of longitude and latitude coordinates of the path planning points. For example, a material warehouse is located at 120.40° east longitude and 30.32° north latitude. The diffusion probability within 1 km around it is less than 10%, and it is marked as a schedulable starting point. The attributes of the path planning point include resource type, inventory and maximum transportation carrying capacity. For example, the warehouse point is attached with {"type":"medical","capacity":"5-ton trucks×3 vehicles"}. In the flood scenario, the coordinates of the emergency material storage depot that was not flooded are converted into the starting point of the ship supply route, generating a scheduling task node for transporting sandbags to the disaster-stricken area.

[0147] Step S1000: Topologically match the resource replenishment path planning points with the transportation network data in the terrain distribution of the target sub-area, eliminate the path nodes marked as inaccessible by the surface displacement data, and generate an optimized distribution route set with detour suggestion marks.

[0148] Exemplarily, traffic network data includes, for example, road grade, width, and real-time traffic status. For example, expressways, national roads, and county roads are coded as L1, L2, and L3 levels, respectively. During topological matching, the Dijkstra algorithm can be used to calculate the shortest path from the planned point to the disaster-stricken area. For example, the optimal route from the warehouse to the hospital is to drive 8 kilometers along the L1 road and then turn onto the L2 road for 3 kilometers. Surface displacement data is monitored using InSAR technology, and road sections with displacements exceeding 50 cm are marked as inaccessible nodes. Detour suggestions are generated based on the A* algorithm. For example, when the displacement of a bridge in the original path reaches 80 cm, the system recommends detouring to an alternative road 5 kilometers away, and adds a yellow dotted arrow and a "recommended detour" text annotation to the route map. The optimized delivery route set is output as a GPX format file, containing a sequence of path points, detour prompts, and estimated travel time, for import and execution by navigation devices.

[0149] Step S1100: Based on the length and surface displacement of each route in the optimized delivery route set, a route safety level color bar is superimposed on the enhanced terrain layer, and the travel speed of the path with the highest safety level in the delivery route set is compared with the thermal diffusion trajectory to generate a resource scheduling timeliness warning signal.

[0150] For example, a route's safety rating can be calculated using a linear weighted formula: safety score = 0.6 × (1 - displacement / 100 cm) + 0.4 × (1 - route length / 50 km). The result is mapped to three color bars: red (0-0.3), yellow (0.3-0.6), and green (0.6-1). For example, a 20 km long route with a maximum displacement of 30 cm has a safety score of 0.71, displayed as a green bar. The speed is compared to the speed predicted by the diffusion model and the average vehicle speed. For example, if a thermal diffusion trajectory advances 2 km per hour, while the optimal delivery route speed is 60 km / h, a time-sensitive warning signal ("3 hours remaining safe time") is generated. This warning signal is encoded as a JSON message containing the route ID, remaining time threshold, and recommended action, such as {"route_id":"R002", "time_left":"180min", "action":"priority dispatch"}, and is distributed to the command terminal via a message queue.

[0151] Step S1200: The resource scheduling timeliness warning signal is associated with the coordinates of the new monitoring point in the real-time environmental data stream in real time. When the new monitoring point falls into the warning signal coverage area, the recalculation of the optimized distribution route set is automatically triggered, and the view zoom ratio of the current enhanced terrain layer is locked to prevent path display overlap.

[0152] For example, the coordinates of new monitoring points can be uploaded in real time through drone inspections or ground sensors, such as adding a new smoke concentration monitoring point at 120.42° east longitude and 30.31° north latitude. Spatial association judgment uses a point-surface inclusion algorithm. If the point is located in the buffer zone covered by any early warning signal (such as a radius of 2km), the path recalculation is triggered. The view zoom ratio is locked by saving the center coordinates and zoom level of the current viewport, for example, maintaining a scale of 1:50000 to ensure that the path lines and icons are not displaced during the redrawing process. In the typhoon path update scenario, the addition of a monitoring point causes the original optimal path to enter a level 10 wind circle. The system immediately replans a detour route and keeps the map view stable to avoid operational interference.

[0153] Step S1300: In response to the user's click operation on the resource distribution overlay map, dynamically filter and click all redundant coordinate points of emergency resources within the preset coordinate range, and drive the interactive interface to highlight the icons of materials that can be dispatched with priority in a pulse flashing mode.

[0154] Exemplarily, the preset range of click coordinates is defined as a circular area with a radius of 5km centered on the click point. For example, when the user clicks 120.38° east longitude and 30.29° north latitude, the coordinates of all warehouses with inventory adequacy ≥ 70% within the range are filtered. The pulse flashing mode is implemented through CSS animation. The icon transparency changes periodically between 0.3 and 1.0 at a frequency of 2Hz, and the border color is synchronously switched to gold. The highlighting strategy follows the principle of spatial proximity and inventory priority. For example, among the 10 candidate warehouses, the icons of the three warehouses closest to the disaster site continue to flash, and the rest are adjusted to a translucent state. In the dispatch of epidemic materials, after clicking the icon of the cabin hospital, the icons of suppliers with sufficient inventory of protective clothing within a surrounding 5km will enter a flashing state, prompting the commander to contact them first.

[0155] Step S1400: When it is detected that the intensity level color gradient of the heat diffusion trajectory changes across levels, the safety level color bars of all delivery routes intersecting with the changed area in the enhanced terrain layer are refreshed, and the priority sorting of the resource scheduling plan list is adjusted according to the refreshed color bars.

[0156] For example, cross-level changes can be determined based on a difference of ≥ 2 levels in intensity levels between adjacent grid cells, e.g., a sudden change from Level 2 (yellow) to Level 4 (orange). Intersecting delivery routes are identified through spatial overlay analysis. For example, if a 3km section of an original safe green route enters an intensity level 4 area, this triggers the color bar of that section to be updated to orange. Resource scheduling plans are prioritized dynamically based on the path safety score. For example, a route with a score of 0.7 in the original plan list is demoted to third place, while a newly calculated route with a score of 0.85 is promoted to first place. Plan list updates are pushed to mobile devices in real time via WebSocket, ensuring that field personnel receive the latest action guidelines.

[0157] Step S1500: When synchronously playing the evolution process of resource scheduling timeliness warning signals, optimized distribution route sets, and heat diffusion trajectories in the interactive interface, interframe tween animation is dynamically inserted according to the real-time movement speed of the graphic elements to maintain the visual continuity of the collaborative evolution of multiple layers.

[0158] For example, inter-frame interpolation animation can use a linear interpolation algorithm. For example, the heat diffusion trajectory moves 5 pixels per frame, and 3 intermediate positions are inserted between two frames to make the trajectory advancement process smooth and without jumps. The position update frequency of the resource scheduling route icon is synchronized with the vehicle GPS signal. When the data transmission interval exceeds 500ms, the intermediate trajectory points are calculated based on the speed vector. Visual continuity is guaranteed by a unified time base. The animation progress of all layers is bound to the same timer, for example, a global redraw is triggered every 100ms. In the visualization of the flood peak evolution, the resource transportation vehicle icon moves dynamically along the path, and the proportional relationship between its displacement speed and the flood diffusion speed is adjusted through the animation frame rate to ensure that the commander can intuitively perceive the urgency of the time window.

[0159] As an embodiment, the method further includes a model online updating process, which may specifically include the following steps:

[0160] Step S1600: monitor the prediction error of the event evolution prediction model on the latest environmental data stream in real time, and trigger the model retraining process when the prediction error exceeds a preset tolerance.

[0161] For example, the prediction error of the event evolution prediction model is quantified by calculating the root mean square error (RMSE) between the model output and the actual observation value. Preset tolerances set dynamic thresholds based on different emergency event types. For example, in a typhoon path prediction scenario, when the predicted center position deviation for three consecutive time segments exceeds 15 kilometers (RMSE ≥ 15km), the prediction error is determined to have exceeded the tolerance threshold, triggering the model retraining process. The real-time monitoring module is embedded in the data pipeline processing node. Upon receiving the latest batch of environmental data streams (such as wind speed and air pressure data updated every minute), it immediately performs error calculations and compares them with the threshold. In forest fire spread prediction, if the relative error between the actual fire front advance speed and the model prediction value exceeds 20% for five consecutive times, the system automatically generates a model failure alarm log and initiates the parameter update task queue.

[0162] Step S1700: extract continuous time segments from the current environmental data stream as new training data, and perform distribution consistency detection with the historical monitoring data set.

[0163] For example, continuous time segments can be captured using a sliding window mechanism, with the window length aligned with the model input dimension. For example, a typhoon prediction model requires 72 hours of continuous data as input. The newly added training data captures the sea temperature and wind speed matrices updated every second in the last 72 hours and is stored as a tensor structure with a shape of (72×3600, 12). Distribution consistency detection is achieved by comparing the statistical characteristics of key parameters of the newly added data with those of the historical data. For example, if the mean of the air pressure value in the historical data set is 1013 hPa and the standard deviation is 5 hPa, if the newly added data has a distribution deviation of 1018 hPa and a standard deviation of 12 hPa, the data anomaly flag is triggered. In the flood prediction scenario, the proportion of samples in the newly added data with hourly rainfall exceeding 200% of the historical maximum record is 15%, indicating that the rainfall intensity distribution has changed significantly and requires subsequent processing.

[0164] As an implementation manner, in step S1700, the distribution consistency detection process includes:

[0165] Step S1710: Calculate the KL divergence of the newly added training data and the historical monitoring data on the dynamic environment parameters to determine whether the data distribution has significantly shifted.

[0166] The KL divergence (Kullback-Leibler Divergence) measures the asymmetric difference between two probability distributions. Dynamic environmental parameters select the features that have the greatest impact on model predictions. For example, in typhoon forecasting, three parameters, central pressure, maximum wind speed, and moving direction angle, are selected and the KL divergence of their probability density functions is calculated. The significance threshold is set to KL ≥ 0.1. When the KL value of the pressure distribution of the newly added data reaches 0.15, a significant shift is considered to have occurred. For flood forecasting, for example, the KL value of the hourly rainfall distribution in the newly added dataset is 0.25, significantly higher than the historical data distribution, triggering the data augmentation process.

[0167] Step S1720: If a significant deviation occurs, the data augmentation process is started to synthesize simulated data consistent with the historical distribution through the generative adversarial network.

[0168] For example, the generator of a Generative Adversarial Network (GAN) learns the distribution characteristics of historical data. For example, in a typhoon scenario, it generates simulated data that conforms to the historical mean pressure of 1013 hPa and a Weibull distribution of wind speed with shape parameter k=2. The discriminator network distinguishes between real and synthetic samples with a confidence level of 0.95, and training stops when the discriminator's error rate for generated data exceeds 30%. In a rainstorm forecasting case, the GAN generated 1,000 sets of hourly rainfall data, whose distribution mean and variance differed by less than 5% from the historical data, effectively compensating for the distribution bias of the newly added data.

[0169] Step S1730: Perform physical constraint verification on the simulation data to eliminate abnormal samples that do not conform to the actual environment rules.

[0170] For example, physical constraint verification can define a reasonable range of parameters based on domain knowledge. For example, the air pressure at the center of a typhoon must not be lower than 870 hPa or higher than 1100 hPa, and the wind speed and air pressure must satisfy the gradient wind equation. The symbolic regression algorithm is used to detect contradictory items in the generated data. For example, in a simulated sample, a wind speed of 50 m / s corresponds to an air pressure of 980 hPa, which violates the wind speed-air pressure empirical formula Vmax=14.5√(1010-Pmin), and is marked as an abnormal sample. In the flood data verification, the generated hourly rainfall sample of 250 mm was detected to be in an area with a terrain slope of >30 degrees. However, the actual soil saturation runoff time at this slope is less than 1 hour, and effective runoff cannot be generated, so it is eliminated.

[0171] Step S1740: Mix the verified simulated data with the newly added training data to generate a balanced training data set for model retraining.

[0172] For example, the mixing ratio can be dynamically adjusted based on the KL divergence value. When the KL value for newly added data is 0.15, a balanced dataset is constructed with a ratio of 6:2:2 for historical data: newly added data: simulated data. The SMOTE (Synthetic Minority Oversampling Technique) algorithm is used to balance the dataset, oversampling minority samples (such as super typhoon data). When training an earthquake prediction model, simulated magnitude 7 earthquake data is mixed with actual newly added magnitude 5 earthquake data to shift the magnitude distribution from a skewed distribution to a uniform distribution, improving the model's ability to predict rare events.

[0173] Step S1750: After the model update is completed, the performance improvement of the updated model is evaluated by comparing the prediction results of the new and old models on the validation set.

[0174] For example, the validation set was constructed using time series cross-validation, including data from the previous 30 days that had not been used in training. Performance evaluation metrics included forecast accuracy improvement (e.g., typhoon path error decreased from 15km to 12km, a 20% improvement) and classification F1 scores (e.g., F1 for rainstorm intensity prediction increased from 0.85 to 0.89). In the forest fire case study, the updated model's prediction accuracy for fire spread increased from 78% to 85%, and the false alarm rate decreased by 3 percentage points. The system generates a performance report and stores it in the model repository.

[0175] Step S1800: dynamic weights are assigned to newly detected training data, and the parameters of the event evolution prediction model are updated using an incremental learning algorithm.

[0176] For example, dynamic weight allocation is based on the inverse of data timeliness and prediction error, with the weight of data from the last 72 hours set at 1.2 and historical data downgraded to 0.8. Incremental learning utilizes the Online Sequential Extreme Learning Machine (OS-ELM) algorithm, which updates only the network output layer weight matrix while preserving the hidden layer's feature extraction capabilities. For example, the typhoon prediction model has a fixed number of 500 hidden layer nodes. New data is used to adjust the output layer weights at a learning rate of 0.01. This parameter update takes no more than 200ms on the GPU, meeting real-time requirements.

[0177] Step S1900: Perform compatibility testing on the updated model and the geographic area matching model to ensure that the parameter mapping relationship of the visualization layer template is adjusted synchronously.

[0178] For example, the compatibility test focuses on verifying the spatial alignment accuracy between the model output and the layer elements. For example, the coordinate system of the diffusion path raster unit output by the updated model must be consistent with the tile index of the geographic area matching model. When the resolution of the path probability matrix output by the model is increased from 1km to 500m, the geographic area matching model automatically adjusts the raster division rules and subdivides the original raster unit into 4 sub-units for parameter mapping. During the testing phase, the system injects 100 sets of test cases, requiring that the deviation between the path coverage raster and the layer display coordinates does not exceed 2 pixels, otherwise the coordinate conversion parameter calibration process will be triggered.

[0179] Step S2000: Gradually deploy the updated model to the production environment through the grayscale release mechanism, and prompt the model version change information in the interactive interface.

[0180] For example, the grayscale release utilizes traffic diversion technology. For example, on the first day, 5% of prediction requests are routed to the new model, and its prediction error and system load are continuously monitored. Once the new model performs stably within 24 hours (e.g., its error rate is less than 90% of the old model), the traffic diversion ratio is gradually increased to 100%. The UI adds a version identifier (e.g., "Typhoon Forecast Model V2.1.5") to the upper-right status bar, and displays the update log via a tooltip, including performance improvement data and key parameter changes. Users can click on the version number to view a comparison chart of the new and old model predictions, for example, a side-by-side display of the difference between the 72-hour path forecasts for the same typhoon event using the V2.1.4 and V2.1.5 models.

[0181] Please refer to Figure 3 , is a block diagram of the structure of the terminal device 120 of the present invention. The terminal device 120 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a random access memory (RAM) 1003. The RAM 1003 may also store various programs and data required for the operation of the terminal device 120. The computing unit 1001, ROM 1002, and RAM 1003 are connected to each other via a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0182] Multiple components in the terminal device 120 are connected to the I / O interface 1005, including: an input unit 1006, an output unit 1007, a storage unit 1008, and a communication unit 1009. The input unit 1006 can be any type of device that can input information to the terminal device 120. The input unit 1006 can receive input digital or character information and generate key signal input related to user settings and / or function control of the server, and can include but is not limited to a mouse, a keyboard, a touch screen, a trackpad, a trackball, a joystick, a microphone, and / or a remote control. The output unit 1007 can be any type of device that can present information, and can include but is not limited to a display, a speaker, a video / audio output terminal, a vibrator, and / or a printer. The storage unit 1008 can include but is not limited to a magnetic disk and an optical disk. The communication unit 1009 allows the terminal device 120 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks, and may include but is not limited to a modem, a network card, an infrared communication device, a wireless communication transceiver and / or a chipset, such as a Bluetooth™ device, an 802.11 device, a WiFi device, a WiMax device, a cellular communication device and / or the like.

[0183] The computing unit 1001 can be various general-purpose and / or specialized processing components with processing and computing capabilities. Some examples of the computing unit 1001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 1001 performs the various methods and processes described above, such as the emergency event visualization collaboration method. For example, in some embodiments, the emergency event visualization collaboration method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program can be loaded and / or installed on the terminal device 120 via the ROM 1002 and / or the communication unit 1009. When the computer program is loaded into the RAM 1003 and executed by the computing unit 1001, one or more steps of the emergency event visualization collaboration method described above can be performed. Alternatively, in other embodiments, the computing unit 1001 may be configured to execute the emergency event visualization collaboration method in any other appropriate manner (eg, by means of firmware).

[0184] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0185] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0186] It should be understood that various forms of flow shown above can be used, with steps reordered, added, or removed. For example, various steps recited in the application can be performed in parallel, in series, or in a different order, without limitation herein, so long as the desired results of the technology disclosed herein are achieved.

[0187] While embodiments or examples of this application have been described with reference to the figures, it will be understood by those skilled in the art that the foregoing and various other embodiments or examples can be made and implemented in various ways. The foregoing and other embodiments or examples of this application can be practiced by employing the specific designs described herein, by employing other designs, or by employing a combination of designs that are known or to be developed. The disclosures of the patents, patent applications, and publications referenced in this disclosure are hereby incorporated by reference in their entirety.< / marker> < / g> < / polygon> < / path> < / g> < / path> < / polygon> < / polygon> < / text> < / g> < / lineargradient> < / path>

Claims

1. A visual collaborative method for emergency events, characterized in that: The method comprises: Acquire a historical monitoring dataset of emergency events in a target area, and preprocess the historical monitoring dataset to generate a standardized event feature set; the standardized event feature set includes dynamic environmental parameters and static geographic parameters associated with multiple emergency event types; Training a time series prediction model based on the dynamic environmental parameters in the standardized event feature set to obtain an event evolution prediction model; the output of the event evolution prediction model includes the diffusion path and intensity level of the emergency event within a preset time window; Converting the terrain distribution data in the static geographic parameters into a rasterized elevation map, and marking the grid cells corresponding to the infrastructure coordinates in the elevation map; Correlating emergency resource deployment information within the grid cells based on historical emergency resource dispatch records to generate an initial geographic layer; Performing a topological structure analysis on the initial geographic layer, identifying key path nodes and potential risk areas, and adding corresponding topological relationship labels; Performing regional segmentation on the initial geographic layer based on the topological relationship labels to obtain multiple sub-region templates, and configuring an independent data rendering channel for each sub-region template; Dynamically bind each sub-region template to the intensity level in the event classification result to generate a visualization layer template that automatically adjusts the display scale according to the intensity level. The visualization layer template includes terrain distribution, infrastructure coordinates, and emergency resource deployment information. Collecting the current environmental data stream of the target area in real time, classifying the current environmental data stream using the event evolution prediction model to obtain an event classification result, and calling the corresponding visualization layer template according to the event classification result; According to the diffusion path in the event classification result, a heat diffusion trajectory is drawn in the called visualization layer template, and the intensity level change is represented by a color gradient; Identifying grid cells covered by the diffusion path, and calculating the resource gap ratio based on the emergency resource deployment information to generate a resource warning mark; Analyzing the continuous coordinate point sequence of the heat diffusion trajectory, generating a piecewise Bezier curve control point set with transparency attenuation properties based on the timestamp information of the diffusion path, and mapping the intensity level color gradient to the curve width gradient parameter to form a trajectory vector primitive that is traceable in the time dimension; Extracting the color gamut distribution range of the intensity level color gradient, discretizing the continuous color scale into polygonal fill area definition instructions based on the grid cell boundary coordinates, and adding a gradient color band attribute tag containing intensity value metadata to each polygon; Decomposing the icon elements of the resource warning mark into a set of basic geometric primitives, converting the warning level text into vector font outline data, and encoding the material gap ratio numerical value into primitive scaling coefficients and rotation angle parameters; Aligning the trajectory vector primitives, gradient color band attribute labels, and geometric primitive sets according to a spatial coordinate system, establishing a mapping relationship between a timeline control and vector element animation attributes, and generating an uncompressed original vector instruction set; Optimizing the topological structure of the original vector instruction set, merging polygon filling instructions with the same attributes in adjacent grid cells, compressing repeated geometric primitive coordinate sequences, and generating a simplified composite vector instruction set; Embedding an interactive response tag associated with emergency resource deployment information in the composite vector instruction set, and attaching a material type code and an inventory query interface to each resource warning tag; According to the rendering priority of the visualization layer template, the composite vector instruction set is hierarchically indexed and marked to generate the final vector graphics instruction packet containing the layer overlay order and visibility control parameters; Dynamically adjust the rendering accuracy of vector graphics instructions according to the resolution of the interactive interface, and add timeline controls to enable historical playback and predictive deduction of evolutionary situation maps; The vector graphics instructions are converted into pixel data streams through a parallel rendering engine and output to the interactive interface for real-time refresh, and an emergency event evolution situation diagram is rendered and output through the interactive interface.

2. The method according to claim 1, characterized in that The preprocessing of the historical monitoring data set to generate a standardized event feature set includes: Interpolate and fill missing data in the historical monitoring data set, and filter noise data according to a preset abnormal threshold to generate a cleaned monitoring data subset; Extracting target characteristic parameters that are strongly correlated with the type of emergency event from the cleaned monitoring data subset; the target characteristic parameters include meteorological change rate and surface displacement in dynamic environmental parameters, and altitude and geological structure stability index in static geographic parameters; According to the physical constraints corresponding to each type of emergency event, the target feature parameters are normalized to generate the standardized event feature set; wherein, the normalization includes mapping parameters of different dimensions to a unified numerical interval and matching the corresponding nonlinear scaling function according to the event type.

3. The method according to claim 2, characterized in that The extraction of target feature parameters that are strongly related to the type of emergency event includes: Calculate the information gain of each characteristic parameter in the cleaned monitoring data subset according to the event type labels marked in the historical emergency event case library; Selecting the meteorological change rate and the surface displacement whose information gain is greater than a first threshold from the dynamic environmental parameters as the first candidate feature set; Selecting geological structure stability indicators and altitudes with information gain greater than a second threshold from the static geographic parameters as the second candidate feature set; Performing multicollinearity detection on the first candidate feature set and the second candidate feature set, deleting redundant features that are linearly correlated with other features, and obtaining an optimized target feature parameter set; The optimized target feature parameter set is matched with a preset emergency event type-feature mapping table, and the feature weights are dynamically adjusted according to the matching results to generate the final target feature parameters.

4. The method according to claim 1, wherein The training of a time series prediction model based on the dynamic environment parameters in the standardized event feature set to obtain an event evolution prediction model includes: Dividing the dynamic environment parameters into a training sequence and a verification sequence in chronological order, wherein the training sequence is used for updating model parameters, and the verification sequence is used for evaluating model prediction accuracy; Constructing a hybrid model architecture including a long short-term memory network and a convolutional neural network, and performing segmented sampling on the training sequence through a sliding window mechanism to generate multiple time segment samples; Add event type labels and intensity level labels to each time segment sample, input the hybrid model architecture for multi-task joint training, and obtain an initial prediction model; The mean square error and classification accuracy of the initial prediction model at each time segment are calculated through the verification sequence, and the model parameters are optimized using an adaptive learning rate algorithm until convergence; The optimized model is tested to match the delay tolerance of the real-time data stream, and the number of model layers and neurons is adjusted according to the test results to generate an event evolution prediction model that meets real-time requirements.

5. The method according to claim 4, characterized in that The training process of the hybrid model architecture also includes: Inputting the global temporal features output by the long short-term memory network into the attention mechanism layer, generating the event evolution contribution weight corresponding to each time step, and dynamically weighting and summing the global temporal features according to the contribution weight to obtain a temporal attention feature vector; Aligning the local fluctuation pattern extracted by the convolutional neural network with the temporal attention feature vector to form a concatenated feature matrix containing temporal global dependencies and spatial local details; Inputting the spliced ​​feature matrix into a gated recurrent unit, controlling the feature transfer path by resetting the gate and updating the gate, filtering out noise data irrelevant to the diffusion path of the emergency event, and outputting a filtered high-dimensional spatiotemporal feature sequence; The high-dimensional spatiotemporal feature sequence is input into the diffusion path prediction branch and the intensity level regression branch respectively, wherein the diffusion path prediction branch outputs a probability distribution map of each grid unit being diffusely covered by the softmax function, and the intensity level regression branch outputs the intensity level value at each moment in a preset time window by using a linear activation function; A cross entropy calculation is performed on the probability distribution map and the grid unit coordinates of the true diffusion path in the verification sequence. At the same time, a mean square error calculation is performed on the intensity level value and the level label annotated in the verification sequence. The two loss values ​​are combined to generate a joint optimization target. According to the joint optimization objective, the error signal is back-propagated, and the weight parameters of the long short-term memory network, convolutional neural network, and gated recurrent unit are synchronously adjusted until the classification accuracy of the diffusion path prediction branch and the error rate of the intensity level regression branch simultaneously reach the convergence threshold.

6. The method according to claim 1, characterized in that The method further comprises: In response to a region selection instruction triggered by a user through an interactive interface, intercepting a target subregion from the visualization layer template according to a coordinate range contained in the region selection instruction, and extracting terrain distribution, infrastructure coordinates, and emergency resource deployment information corresponding to the target subregion; Superimposing the terrain distribution of the target sub-area with the real-time meteorological parameters in the current environmental data stream to generate an enhanced terrain layer with wind speed vector arrows and precipitation intensity contour lines; Locating the locations of buildings affected by the diffusion path in the enhanced terrain layer according to the infrastructure coordinates, retrieving the material inventory in the associated emergency resource deployment information, and generating a resource distribution overlay map that indicates the inventory adequacy using icon size; Based on the spatial intersection area of ​​the resource distribution coverage map and the heat diffusion trajectory, redundant coordinate points of emergency resources not covered by the diffusion path are calculated and converted into resource replenishment path planning points that can be dynamically scheduled; Topologically match the resource replenishment route planning points with the transportation network data in the terrain distribution of the target sub-area, eliminate the path nodes marked as impassable by the surface displacement data, and generate an optimized delivery route set with detour suggestion marks; Based on the length and surface displacement of each route in the optimized delivery route set, a route safety level color bar is superimposed on the enhanced terrain layer, and the travel speed of the path with the highest safety level in the delivery route set is compared with the heat diffusion trajectory to generate a resource scheduling timeliness warning signal; The resource scheduling timeliness warning signal is associated with the coordinates of the new monitoring point in the real-time environmental data stream in real time. When the new monitoring point falls into the coverage area of ​​the warning signal, the optimized delivery route set is automatically recalculated, and the current enhanced terrain layer view zoom ratio is locked to prevent the path display from overlapping. In response to a user's click on the resource distribution overlay, the system dynamically filters and clicks all redundant coordinate points of emergency resources within a preset coordinate range, and drives the interactive interface to highlight the icons of resources that can be dispatched with priority in a pulse flashing mode; When a cross-level change in the intensity level color gradient of the heat diffusion trajectory is detected, the safety level color bars of all delivery routes intersecting the changed area in the enhanced terrain layer are refreshed, and the priority ranking of the resource scheduling plan list is adjusted according to the refreshed color bars; When synchronously playing the evolution process of resource scheduling time warning signals, optimized distribution route sets, and heat diffusion trajectories in the interactive interface, interframe interpolation animation is dynamically inserted according to the real-time movement speed of graphic elements to maintain the visual continuity of the collaborative evolution of multiple layers.

7. The method according to claim 1, characterized in that The method also includes a model online updating process, including: Real-time monitoring of the prediction error of the event evolution prediction model on the latest environmental data stream, and triggering a model retraining process when the prediction error exceeds a preset tolerance; Extract continuous time segments from the current environmental data stream as new training data and perform distribution consistency checks with historical monitoring data sets; Assigning dynamic weights to newly detected training data and using an incremental learning algorithm to update the parameters of the event evolution prediction model; Perform compatibility tests on the updated model and the geographic region matching model to ensure that the parameter mapping relationship of the visualization layer template is adjusted synchronously; The updated model is gradually deployed to the production environment through the grayscale release mechanism, and the model version change information is prompted in the interactive interface.

8. A terminal device, characterized in that: include: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

Images