Mountain torrent early warning method and system based on digital twinborn technology

Abstract

The invention belongs to the technical field of mountain torrent early warning, and particularly discloses a mountain torrent early warning method and system based on a digital twinborn technology. According to the method, the data offset between prediction parameters and real-time data of the digital twin model is calculated in a continuous period, when the data offset continuously exceeds a preset offset threshold value, it is judged that an offset trend is formed, and early warning is triggered by recognizing the offset trend. Determining a key parameter from the plurality of prediction parameters forming the offset trend to generate a risk prediction parameter; and a plurality of early warning schemes are generated in combination with the selectable parameters, and evacuation suggestion information is generated and output through a multi-scheme evaluation and preferential mechanism. According to the method, early warning is triggered by recognizing the deviation trend, the mountain torrent disaster precursor is captured, the false alarm rate and the missing report rate are reduced, the prediction accuracy is improved through multi-scheme evaluation and a preferential mechanism, and the actual combat value and the guiding effect of early warning are enhanced.

Description

Technical Field

[0001] This invention belongs to the field of flash flood early warning technology, specifically relating to a flash flood early warning method and system based on digital twin technology. Background Technology

[0002] Flash floods are mostly caused by short-term heavy rainfall or continuous torrential rain, and are often accompanied by secondary disasters with huge destructive power such as landslides and debris flows. They are characterized by their suddenness and high risk of disaster. Current technologies can reveal the macro-laws of flash flood occurrence through statistical analysis of historical disaster data, which can guide long-term flash flood planning and management. However, they cannot predict flash flood events in real time and lack the ability to predict the time, specific location and scope of impact of disasters, making it difficult to meet the immediacy requirements of disaster early warning.

[0003] Furthermore, in existing technologies, the use of automated monitoring systems deployed in locations prone to flash floods to monitor flash floods often overlooks the complexity and nonlinearity of the flood generation process. Alarms are only triggered when rainfall or water level exceeds preset values. At the same time, it is difficult to integrate multi-source heterogeneous data such as rainfall, water level, and soil moisture content, resulting in insufficient accuracy of early warnings.

[0004] Furthermore, existing flash flood monitoring technologies struggle to predict the formation and development of flood peaks based on changes in real-time monitoring parameters, and the lack of forward-looking early warning information makes it difficult to promptly notify downstream personnel to evacuate and conduct emergency responses.

[0005] In view of this, the present invention proposes a flash flood early warning method and system based on digital twin technology. Summary of the Invention

[0006] The purpose of this invention is to provide a flash flood early warning method and system based on digital twin technology, which solves the problem that the existing technology cannot analyze the changing trend of flash floods based on actual measured parameters, resulting in inaccurate flash flood prediction and early warning.

[0007] To achieve the above-mentioned objectives, the technical solution adopted by the present invention is as follows:

[0008] A flash flood early warning method based on digital twin technology includes the following steps:

[0009] When a deviation trend is determined to be forming, a flash flood warning will be issued;

[0010] Among them, the offset trend is formed when the data offset between the predicted parameters generated based on the digital twin model and the real-time data exceeds a preset offset threshold for a continuous preset number of periods.

[0011] The data offset is the difference between the prediction parameter and the real-time data corresponding to its time point; and the conditions for determining the formation of an offset trend also include: the data offset calculated in the current period changes in the same direction as the data offset calculated in the previous period.

[0012] The implementation of flash flood warning includes: determining key parameters from multiple prediction parameters that form a deviation trend; generating risk prediction parameters based on the key parameters; and generating and outputting evacuation suggestion information based on the risk prediction parameters.

[0013] The generation of prediction parameters includes: acquiring the instantaneous water flow change between real-time data corresponding to two adjacent retrieval time points in a cycle; and using the instantaneous water flow change as input, processing it through a digital twin model to generate prediction parameters.

[0014] Preferably, the acquisition of instantaneous water flow changes includes:

[0015] Extract real-time data corresponding to two adjacent retrieval time points; calculate the average water flow offset based on the real-time data corresponding to the two retrieval time points, and use the average water flow offset as the instantaneous water flow change.

[0016] Preferably, the method further includes:

[0017] Acquire monitoring data for areas prone to flash floods. The monitoring data includes rainfall data, soil moisture data, and hydrological data, which are both historical and real-time data. Based on the monitoring data, generate multiple retrieval time points sequentially from the initial time point, with a preset sampling time interval as the cycle.

[0018] Preferably, determining the key parameters from the multiple prediction parameters that form the offset trend includes:

[0019] The last prediction parameter that forms the offset trend is identified as the key parameter; the steps to generate risk prediction parameters include: constructing a prediction curve based on one or more key parameters; and extracting parameters within a preset prediction period from the prediction curve as risk prediction parameters.

[0020] Preferably, the process of generating and outputting evacuation suggestion information further includes:

[0021] The risk prediction parameters are combined with optional parameters, including water flow disturbance parameters and airflow disturbance parameters of the prediction area, to generate multiple evaluation parameter sets, where each evaluation parameter set corresponds to an early warning scheme. By calculating the deviation between the evaluation parameter set corresponding to each early warning scheme and the reference data, the early warning scheme with the smallest deviation value is selected as the standard scheme, and evacuation recommendation information is generated and output based on the standard scheme.

[0022] Preferably, the evacuation suggestion information includes:

[0023] The airflow evacuation plan and water flow evacuation plan are generated based on the sensor distribution. The airflow evacuation plan includes the delineation information of the odor distribution area and the delineation information of the alarm area, while the water flow evacuation plan includes the water inlet evacuation information and the water return evacuation information.

[0024] A flash flood early warning system based on digital twin technology includes: a digital twin model, a trend monitoring module, a risk early warning module, and an information output module;

[0025] The trend monitoring module is used to periodically acquire real-time data and receive prediction parameters generated by the digital twin model. By calculating the data offset between the prediction parameters and the real-time data, it determines that an offset trend has been formed when the data offset exceeds a preset offset threshold continuously within a preset number of periods.

[0026] The risk warning module is used to determine the deviation trend for the response trend monitoring module, and to perform the following steps: determine key parameters from multiple prediction parameters that form the deviation trend; generate risk prediction parameters based on the key parameters; combine the risk prediction parameters with optional parameters to generate multiple sets of evaluation parameters corresponding to different warning schemes; select a standard scheme by comparing multiple sets of evaluation parameters with reference data; and generate evacuation recommendation information based on the standard scheme.

[0027] The information output module is used to output evacuation advice information generated by the risk warning module.

[0028] Preferably, the trend monitoring module is further used to acquire the instantaneous water flow change between real-time data corresponding to two adjacent retrieval time points, and provide the instantaneous water flow change to the digital twin model to generate prediction parameters.

[0029] Preferably, determining the key parameters from the multiple prediction parameters that form the offset trend includes:

[0030] The last prediction parameter that forms the deviation trend is identified as the key parameter; the risk warning module generates risk prediction parameters by constructing a prediction curve based on one or more key parameters, and extracting parameters within a preset prediction period from the prediction curve as risk prediction parameters.

[0031] Preferably, the optional parameters include water flow disturbance parameters and airflow disturbance parameters for the prediction area;

[0032] The evacuation suggestion information includes airflow evacuation plans and water flow evacuation plans generated based on sensor distribution; wherein, the airflow evacuation plan includes the delineation information of odor distribution area and alarm area, and the water flow evacuation plan includes water inlet evacuation information and water return evacuation information.

[0033] Beneficial effects

[0034] This invention calculates the data offset between the predicted parameters of the digital twin model and the real-time data over a continuous period. When the data offset continuously exceeds a preset offset threshold, an offset trend is determined. By identifying the offset trend, an early warning is triggered, thereby capturing the precursors of flash floods, reducing the false alarm rate and the missed alarm rate, and improving the stability of the early warning.

[0035] After determining the deviation trend, this invention constructs a prediction curve based on key parameters to generate risk prediction parameters. Then, it combines optional parameters to generate multiple early warning schemes. By comparing with reference data, the early warning scheme with the smallest deviation value is selected as the standard scheme. After comprehensively considering various environmental factors and dynamically correcting the prediction, the accuracy and scientific nature of the prediction are improved.

[0036] This invention generates evacuation advice information based on standard schemes, including airflow evacuation schemes and water flow evacuation schemes, and further refines them into information on the delineation of odor distribution areas, alarm areas, water inflow evacuation information, and water return evacuation information, so as to provide a kind of refined evacuation guidance for different risk media, which can guide on-site personnel to evacuate quickly, orderly and safely, and enhance the guiding role of early warning information. Attached Figure Description

[0037] Figure 1 This is a flowchart of the method for issuing a flash flood warning when a deviation trend is determined, provided by the present invention.

[0038] Figure 2 This is a flowchart of the method for generating and outputting evacuation suggestion information provided by the present invention;

[0039] Figure 3 This is a system structure diagram provided by the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are merely for explaining the invention and are not intended to limit the scope of protection of the invention.

[0041] Example 1

[0042] Please refer to Figures 1-2 As shown, this embodiment discloses a flash flood early warning method based on digital twin technology, applicable to monitoring and early warning scenarios of sudden floods in regional mountains, valleys, and rivers, and specifically includes the following steps:

[0043] Acquire monitoring data of areas prone to flash floods and define it as basic data. Then, divide the basic data into historical data and real-time data to construct a high-fidelity digital twin environment that can accurately map the physical state of the real world.

[0044] The monitoring data includes rainfall data, which serves as the trigger for flash floods; soil moisture data, which affects the generation and confluence of surface runoff; and hydrological data, such as water level, flow velocity, and flow rate, which directly characterize the current carrying and transport status of the river channel.

[0045] Among them, the data collected before the preset initial time point is used as historical data to calibrate and standardize the various physical parameters and response relationships within the digital twin environment, thereby ensuring that the digital twin environment can accurately reproduce the past hydrogeological response patterns of the target area; the latest data collected by the field sensor network at each retrieval time point from the initial time point is used as real-time data to drive the digital twin environment to perform real-time status updates and future trend prediction calculations.

[0046] Furthermore, after determining the initial time point, a dynamic sampling time interval is set based on the current warning response level and weather conditions; and multiple periodic sampling time points are generated sequentially from the initial time point, with the sampling time interval as the cycle. Under high-level warnings or severe weather, the dynamic sampling time interval is shortened to achieve higher frequency monitoring.

[0047] Within a cycle consisting of two consecutive periodic data collection points, real-time hydrological data corresponding to the two time points are acquired. By calculating the net change in real-time hydrological data within this cycle, a value reflecting the overall trend of hydrological status changes during that period is obtained. This value is defined as the net change in hydrological status within the cycle. This net change filters out the instantaneous reading noise that may exist in a single sensor and can serve as a direct input to drive the digital twin environment to perform the next time step extrapolation.

[0048] Furthermore, the net change in hydrological state within the period is input into the digital twin environment. The digital twin environment performs extrapolation and calculation based on the physical parameters and response relationships of the river roughness based on the Manning formula and the soil infiltration rate based on the interflow model, which have been calibrated within it, thereby generating predictive parameters that characterize the hydrological state at the next moment.

[0049] Specifically, by acquiring real-time data corresponding to the time points of the prediction parameters, the difference between the prediction parameters and the real-time data is calculated, and this difference is determined as the data offset. The change of this data offset is continuously tracked within the sliding observation window in order to continuously monitor the accuracy of the inference and identify signs of risk.

[0050] When the absolute value of the data offset exceeds the preset offset threshold for a consecutive preset number of periods, and the data offset calculated in the current period changes in the same direction as the data offset in the previous period, thus showing a continuous and unidirectional deviation, an offset trend is determined to be formed.

[0051] Among them, a continuous and unidirectional deviation trend is considered a strong signal that the physical world has undergone drastic changes that cannot be explained by existing environmental parameters, such as the formation of a landslide dam upstream or a local breach of a levee. Therefore, this deviation trend is identified as a peak trend with high risk indicative significance. After the peak trend is identified, the last prediction parameter that forms the trend is determined as the key parameter. This parameter represents the final state of the abnormal trend at the current moment and contains the latest information on the cumulative deviation. It is the most effective and relevant benchmark for extrapolating and predicting future risks.

[0052] Furthermore, based on the determined key parameters, extrapolation calculations are performed using pre-set extrapolation rules such as water balance equations and river evolution models based on hydrological and geological principles, or by analyzing the periodic and trend characteristics in historical data sequences to construct one or more prediction curves describing the possible evolution paths of hydrological parameters such as water level and flow rate in the future. The prediction curves include the fitted values ​​of hydrological parameters at the current time point.

[0053] In this implementation, the next 2 hours are used as the preset prediction period, and parameters within this period are extracted from the prediction curve. The extracted parameters are used as risk prediction parameters, which serve as the core basis for assessing future flash flood risks.

[0054] Specifically, to address the extreme uncertainties in the development of flash floods, optional parameters representing extreme or uncertain conditions are further defined. These include water flow disturbance parameters that simulate scenarios such as the sudden opening of upstream sluice gates or partial blockage of the river channel due to landslides, which have a severe impact on the flow state, and airflow disturbance parameters that simulate sudden increases in local rainfall intensity caused by short-term severe convective weather, which have a severe impact on rainfall conditions. Different values ​​of risk prediction parameters and optional parameters are combined to generate multiple sets of evaluation parameters. Each set of evaluation parameters contains the hydrological parameter fitting values ​​of the corresponding prediction curve at the current time point.

[0055] Each set of assessment parameters corresponds to a complete early warning scheme for flash flood risk response that covers specific uncertainties. In order to select the scheme that is closest to the actual possibility from multiple early warning schemes, the latest real-time monitoring data independent of the prediction process is obtained as reference data. For each early warning scheme, the degree of fit between its corresponding assessment parameter set and the reference data is calculated. Based on the fitting value of the hydrological parameters at the current time point in the assessment parameter set, the reference data is compared. Finally, the early warning scheme with the highest degree of fit and closest to the actual possibility is selected as the standard scheme.

[0056] The formula for calculating the degree of fit is as follows:

[0057]

[0058] In the formula, The degree of agreement refers to the consistency between the early warning scheme represented by the set of evaluation parameters and the real-time reference data; the set of evaluation parameters... This includes risk prediction parameters and disturbance parameters; Let i represent the i-th evaluation parameter, which means the i-th evaluation parameter in the set of evaluation parameters. Parameter values, such as predicted peak flood level or flow rate; reference data This is the latest real-time monitoring data; This represents the i-th reference data, meaning the reference data that is related to... The corresponding number One real-time monitoring value; This represents the weighting coefficient, which means the weighting coefficient of the first generation. The weights of each parameter are used to represent the importance of different parameters in evaluating the overall fit, and satisfy the following conditions: ; This indicates the number of parameters, which means the total number of parameters contained in the evaluation parameter set and the reference data.

[0059] Furthermore, based on the standard scheme, the risk prediction parameters contained therein are compared with the real-time parameters extracted from the real-time collected monitoring data. When the comparison results show that the difference between the risk prediction parameters and the real-time parameters continues to widen or has reached a dangerous level, evacuation advice information is generated and output.

[0060] Specifically, to ensure the high operability of the recommended information, based on the flood inundation range, flow velocity and possible diffusion paths of harmful gases predicted by the standard plan, and combined with the pre-set geographical information such as terrain, road network and shelter location, evacuation recommendation information is dynamically generated. The evacuation recommendation information specifically includes airflow evacuation plans and water flow evacuation plans for different types of secondary disasters.

[0061] The airflow evacuation plan is an evacuation strategy designed to address the risk of harmful gas diffusion, such as methane and chemical leaks, that may accompany flash floods. It includes information on the delineation of odor distribution areas and corresponding alarm zones. The water flow evacuation plan is an evacuation strategy designed to address the risks of flooding and water flow impact. It includes inflow evacuation information to guide evacuation during the rising flood phase and backflow evacuation information to guide evacuation during the receding flood phase to avoid secondary risks such as silt instability and structure collapse. Inflow evacuation information includes predictions of inundation range, water depth, flow velocity, and safe evacuation routes. Backflow evacuation information includes the delineation of danger zones and safe behavior guidelines. All evacuation recommendations are ultimately visualized in a graphical manner to ensure the intuitiveness and ease of understanding of the warning information.

[0062] Example 2

[0063] Please refer to Figure 3 As shown, this embodiment discloses a flash flood early warning system based on digital twin technology, specifically including a digital twin environment and the following modules:

[0064] Digital twin environments are constructed by deploying various sensors such as rain gauges, water level gauges, soil moisture sensors, and flow meters in areas prone to flash floods and collecting long-term historical data including rainfall, soil moisture, and hydrological data. Through data-driven or physical mechanism-data fusion methods, a high-fidelity environment is built to simulate the dynamic evolution of water flow in the area under different meteorological and hydrological conditions.

[0065] A digital twin environment can receive external inputs and output predicted parameters of the hydrological state at a future point in time.

[0066] The trend monitoring module is used to continuously monitor the deviation between the real world and the digital twin environment and identify deviation trends that indicate risks. It continuously generates a series of periodic sampling time points starting from the initial time point by using a dynamic sampling time interval. In each cycle consisting of two adjacent periodic sampling time points, it acquires the real-time data corresponding to these two time points. The real-time data comes from various sensors deployed in areas prone to flash floods.

[0067] Based on the real-time data corresponding to two periodically retrieved time points, the net change in hydrological status within the period is calculated. The net change in hydrological status within the period is then used as input and provided to the digital twin environment. The digital twin environment performs calculations based on the received net change in hydrological status within the period to generate prediction parameters for the hydrological status at the next time point. The trend monitoring module receives these prediction parameters and obtains the real-time data corresponding to the time point of these prediction parameters.

[0068] The trend monitoring module calculates the difference between the predicted parameters and the real-time data as the data offset, compares the data offset with a preset offset threshold, and then determines the trend.

[0069] When it is determined that the calculated data offset exceeds the preset offset threshold within a consecutive preset number of periods, and the data offset calculated in the current period changes in the same direction as the data offset calculated in the previous period, then it is determined that a clear offset trend has been formed.

[0070] A clear deviation trend indicates that the discrepancy between the digital twin environment and the real world is continuing and expanding in the same direction, which is an important signal of the increasing risk of flash floods. Once a deviation trend is determined, the trend monitoring module immediately transmits relevant data, including multiple prediction parameters that indicate the deviation trend, to the risk warning module to initiate the flash flood warning process.

[0071] The risk warning module is used to generate specific warning plans and evacuation suggestions after receiving the offset trend formation signal from the trend monitoring module. By executing key parameter determination, the last predicted parameter that best represents the latest state before the risk approaches is determined from multiple predicted parameters that form the offset trend, and risk prediction parameters are generated based on the key parameter.

[0072] The risk warning module constructs a prediction curve based on one or more identified key parameters during the entire offset trend period through curve fitting or time series prediction algorithms. This curve extrapolates the identified offset trend. Parameters within a preset prediction period for possible risk evolution in the future are extracted from this prediction curve as risk prediction parameters, and multiple options are evaluated and optimized.

[0073] Specifically, the multi-scheme evaluation and selection includes: combining the generated risk prediction parameters with optional parameters, including water flow disturbance parameters and airflow disturbance parameters of the prediction area, to simulate different natural disturbances or emergency intervention measures;

[0074] Each combination generates an evaluation parameter set, and each evaluation parameter set corresponds to an early warning scheme. The latest real-time monitoring data, independent of the prediction process, is obtained as reference data. For each early warning scheme, the degree of consistency between its corresponding evaluation parameter set and the reference data is calculated. The early warning scheme with the highest degree of consistency is selected as the standard scheme to ensure that the optimal early warning scheme is generated after comprehensively considering future risks and multiple possibilities. Based on the selected standard scheme, specific evacuation advice information is generated.

[0075] The information output module is used to receive evacuation advice information and transmit it accurately and promptly to relevant personnel or response centers. After receiving the evacuation advice information, it formats it and pushes it to the designated output terminal, such as the display screen of the early warning center, the mobile device of emergency management personnel, or release it to the public through public broadcasting systems and social media. The content of the evacuation advice information is generated based on the sensor distribution of the predicted area and is highly targeted.

[0076] Evacuation advice may include: airflow evacuation plans, which may include information on the delineation of odor distribution areas based on data from toxic and harmful gas sensors or air quality sensors, as well as information on the delineation of alarm areas requiring immediate alerts to address potential secondary environmental risks associated with flash floods; and water flow evacuation plans, which include clear instructions on how to avoid main flood paths during inflow evacuation and instructions on how to avoid dangerous backflow areas that may be formed due to terrain or obstacles during backflow evacuation.

[0077] By constructing an early warning system that integrates monitoring, analysis, decision-making, and dissemination, we can identify deviations between the digital twin environment and reality to capture risk signals in advance. Then, through evaluation, we can select the best response strategies and generate specific evacuation recommendations.

[0078] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from it. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A flash flood early warning method based on digital twin technology, characterized in that, Includes the following steps: When a deviation trend is determined to form, a flash flood warning will be issued; Among them, the offset trend is formed when the data offset between the predicted parameters generated based on the digital twin model and the real-time data exceeds a preset offset threshold for a continuous preset number of periods. The data offset is the difference between the prediction parameter and the real-time data corresponding to its time point; and the conditions for determining the formation of an offset trend also include: the data offset calculated in the current period changes in the same direction as the data offset calculated in the previous period. The implementation of flash flood warning includes: determining key parameters from multiple prediction parameters that form a deviation trend; generating risk prediction parameters based on the key parameters; and generating and outputting evacuation suggestion information based on the risk prediction parameters. The generation of prediction parameters includes: acquiring the instantaneous water flow change between real-time data corresponding to two adjacent retrieval time points in a cycle; and using the instantaneous water flow change as input, processing it through a digital twin model to generate prediction parameters.

2. The flash flood early warning method based on digital twin technology according to claim 1, characterized in that, The acquisition of instantaneous water flow changes includes: Extract real-time data corresponding to two adjacent retrieval time points; calculate the average water flow offset based on the real-time data corresponding to the two retrieval time points, and use the average water flow offset as the instantaneous water flow change.

3. The flash flood early warning method based on digital twin technology according to claim 1, characterized in that, The method further includes: Acquire monitoring data for areas prone to flash floods. The monitoring data includes rainfall data, soil moisture data, and hydrological data, which are both historical and real-time data. Based on the monitoring data, generate multiple retrieval time points sequentially from the initial time point, with a preset sampling time interval as the cycle.

4. A flash flood early warning method based on digital twin technology according to claim 1, characterized in that, The process of determining key parameters from multiple prediction parameters that form the offset trend includes: The last prediction parameter that forms the offset trend is identified as the key parameter; the steps to generate risk prediction parameters include: constructing a prediction curve based on one or more key parameters; and extracting parameters within a preset prediction period from the prediction curve as risk prediction parameters.

5. A flash flood early warning method based on digital twin technology according to claim 1, characterized in that, Before generating and outputting evacuation suggestion information, the following also includes: The risk prediction parameters are combined with optional parameters, including water flow disturbance parameters and airflow disturbance parameters of the prediction area, to generate multiple evaluation parameter sets, where each evaluation parameter set corresponds to an early warning scheme. By calculating the deviation between the evaluation parameter set corresponding to each early warning scheme and the reference data, the early warning scheme with the smallest deviation value is selected as the standard scheme, and evacuation recommendation information is generated and output based on the standard scheme.

6. A flash flood early warning method based on digital twin technology according to claim 5, characterized in that, The evacuation advice information includes: The airflow evacuation plan and water flow evacuation plan are generated based on the sensor distribution. The airflow evacuation plan includes the delineation information of the odor distribution area and the delineation information of the alarm area, while the water flow evacuation plan includes the water inlet evacuation information and the water return evacuation information.

7. A flash flood early warning system based on digital twin technology, characterized in that, include: Digital twin model, trend monitoring module, risk warning module, and information output module; The trend monitoring module is used to periodically acquire real-time data and receive prediction parameters generated by the digital twin model. By calculating the data offset between the prediction parameters and the real-time data, it determines that an offset trend has been formed when the data offset exceeds a preset offset threshold continuously within a preset number of periods. The risk warning module is used to determine the deviation trend for the response trend monitoring module and to determine the key parameters from multiple prediction parameters that form the deviation trend. Generate risk prediction parameters based on key parameters; Risk prediction parameters are combined with optional parameters to generate multiple sets of evaluation parameters corresponding to different early warning schemes; by comparing multiple sets of evaluation parameters with reference data, a standard scheme is selected. Evacuation recommendations are generated based on standard protocols; The information output module is used to output evacuation advice information generated by the risk warning module.

8. A flash flood early warning system based on digital twin technology according to claim 7, characterized in that, The trend monitoring module is also used to acquire the instantaneous water flow change between real-time data corresponding to two adjacent retrieval time points, and provide the instantaneous water flow change to the digital twin model to generate prediction parameters.

9. A flash flood early warning system based on digital twin technology according to claim 7, characterized in that, The process of determining key parameters from multiple prediction parameters that form the offset trend includes: The last prediction parameter that forms the deviation trend is identified as the key parameter; the risk warning module generates risk prediction parameters by constructing a prediction curve based on one or more key parameters, and extracting parameters within a preset prediction period from the prediction curve as risk prediction parameters.

10. A flash flood early warning system based on digital twin technology according to claim 7, characterized in that, The optional parameters include water flow disturbance parameters and airflow disturbance parameters for the prediction area; The evacuation suggestion information includes airflow evacuation plans and water flow evacuation plans generated based on sensor distribution; wherein, the airflow evacuation plan includes the delineation information of odor distribution area and alarm area, and the water flow evacuation plan includes water inlet evacuation information and water return evacuation information.

Images