Intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas

Abstract

The present invention belongs to the technical field of water conservancy project monitoring. The present invention discloses an intelligent perception and multi-stage progressive early warning system for flood storage and detention area risks, including a flood storage and detention area perception module, which collects multi-source heterogeneous data covering the flood storage and detention area, constructs a unified spatiotemporal coding framework, maps the multi-source heterogeneous data to a standardized geographic information grid, and adopts a confidence control mechanism to form a dynamic storage and detention pressure map; a linkage instability propagation module, which constructs a flood diversion driven situation flow field based on the dynamic storage and detention pressure map, and continuously tracks the topological disturbance characteristics in the flood diversion driven situation flow field, and identifies the linkage instability propagation chain and the instability propagation path; a risk phase change identification module, which dynamically tracks the derivative of the instability propagation tension gradient over time according to the instability propagation chain and the instability propagation path, and combines the center of gravity migration trend and edge activation frequency in the flood diversion driven situation flow field; and provides strong support for scientific decision-making in complex flood environments.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy project monitoring, and more specifically, to an intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas. Background Art

[0002] The patent with patent publication number CN120235457A discloses an intelligent early warning and disposal decision-making system and method for leakage risk of embankments in flood storage and detention areas, which relates to the field of water conservancy projects and intelligent monitoring technology. The system includes a data acquisition module for collecting multi-source perception data of the embankment and performing real-time pre-processing on the data to obtain monitoring data, an AI analysis module for combining monitoring data with historical data to analyze the current working condition of the embankment and the initial leakage characteristics through a pre-trained AI model, a risk prediction module for constructing a dynamic prediction model for leakage risk based on the analysis results and generating risk prediction reports under different working conditions, an emergency response module for guiding on-site emergency response based on the risk prediction report, and implementing a dynamic early warning mechanism to adjust the response strategy in a timely manner; the intelligent early warning and disposal decision-making system and method for leakage risk of embankments in flood storage and detention areas dynamically predicts the development trend of leakage risk under different working conditions to solve the problem of slow emergency response caused by untimely early warning.

[0003] The existing intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas has the following main problems:

[0004] Existing flood storage and detention area early warning methods often rely on static thresholds, historical water level data, or manual experience. These methods struggle to reflect the dynamic evolution of flood tension as it propagates along diverse pathways within complex geographic structures, and lack the ability to quantitatively perceive risk propagation trends within interconnected instability pathways. Existing methods focus primarily on the occurrence or total amount of risk, while ignoring key characteristics such as its rate of change and acceleration. This delays the identification of critical phase transition points and hinders early multi-stage early warning. The spatial grids within flood zones are complex in topology, and pressure changes in different regions affect each other non-uniformly. Existing models lack a path connectivity weighting mechanism, resulting in crude tension estimation and inaccurate identification of instability chains.

[0005] Traditional flood storage and detention area early warning methods are mostly based on fixed thresholds or static indicators. They are unable to dynamically respond to the spatial evolution characteristics and mutation trends of risk expansion, resulting in delayed warnings. Existing systems often use linear models or stage judgment rules, which make it difficult to identify mutation points or phase transition nodes in the development of disasters and lack a mechanism to capture the critical state of disaster spread. Traditional early warning level classification methods often rely on manually set static rules and lack data-driven real-time dynamic adjustment mechanisms. They are difficult to adapt to sudden disaster scenarios under complex hydrogeological conditions. In existing solutions, risk level classification is often based on empirical rules or subjective judgments. There is a lack of a real-time dynamic indicator system coupled with the actual evolution of disasters, which is prone to false alarms or omissions.

[0006] In view of this, the present invention proposes an intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas to solve the above problems. Summary of the Invention

[0007] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned objectives, the present invention provides the following technical solution: a flood storage and detention area risk intelligent perception and multi-stage progressive early warning system, comprising:

[0008] The flood storage and detention area perception module collects multi-source heterogeneous data covering the flood storage and detention area, builds a unified spatiotemporal coding framework, maps the multi-source heterogeneous data to a standardized geographic information grid, and adopts a confidence control mechanism to form a dynamic storage and detention pressure map;

[0009] The linked instability propagation module constructs the flood diversion driven flow field based on the dynamic storage pressure map, continuously tracks the topological disturbance characteristics in the flood diversion driven flow field, and identifies the linked instability propagation chain and instability propagation path;

[0010] The risk phase change identification module dynamically tracks the time derivative of the instability propagation tension gradient based on the instability propagation chain and path. Combining the center of gravity migration trend and edge activation frequency in the flood diversion driven flow field, it establishes a time-varying risk phase change criterion, generates a phase change trigger flag, and divides the risk warning level into multiple stages.

[0011] After identifying the phase change trigger flag, the intent-guided response module retrieves matching instruction combinations from the preset response strategy library based on the preset response target corresponding to the current stage risk warning level, and outputs dynamic executable control instructions;

[0012] The adaptive feedback calibration module continuously obtains disaster development information and external environmental monitoring data after the implementation of executable control instructions, calculates the deviation between the predicted evolution trajectory and the actual evolution trajectory of the flood storage and detention area, and if it is greater than the preset deviation threshold, triggers the response tuning mechanism, corrects the pressure factor distribution in the dynamic storage and detention pressure map, and automatically enters the next stage of risk warning.

[0013] Preferably, the method for mapping multi-source heterogeneous data to a standardized geographic information grid comprises:

[0014] Collect multi-source heterogeneous data covering the flood storage area, including hydrological monitoring data, geological structure monitoring data, meteorological remote sensing data, and group behavior dynamics data; build a unified spatiotemporal coding framework to perform time alignment and spatial standardization on the multi-source heterogeneous data;

[0015] Through the rasterized data mapping method, multi-source heterogeneous data in different structural forms such as point type, trajectory type and regional type are filled into the preset target grid area to form a multi-factor spatial data structure covering the flood storage and detention area.

[0016] Preferably, the method for forming a dynamic stagnation pressure map comprises:

[0017] The sources of multi-source heterogeneous data falling into each geographic information grid cell are identified, and corresponding confidence weights are preset according to the collection method and update frequency of the multi-source heterogeneous data. When the same type of data from different sources exists in the same geographic information grid cell, weighted fusion is performed according to their respective confidence weights to obtain the fused grid pressure value.

[0018] A continuity analysis is performed on the changing trend of grid pressure values ​​between adjacent geographic information grids. When a discontinuous geographic information grid is detected, the multi-source heterogeneous data corresponding to the discontinuous geographic information grid is corrected based on the changing trend of the grid pressure values ​​after fusion of the surrounding geographic information grids. Different confidence weights are dynamically adjusted, and the spatial distribution of the fused grid pressure values ​​is visualized to form a dynamic stagnant pressure map.

[0019] Preferably, the method for constructing the flood diversion driving situation flow field includes:

[0020] The pressure values ​​of each geographic information grid cell in the dynamic stagnant pressure map are extracted, and the pressure gradient information in each direction is calculated based on the spatial adjacency relationship between geographic information grid cells. Based on the geographical constraints between the geographic information grid cells, a coupling relationship network between the geographic information grid cells is established to form a structural map reflecting the pressure transmission path.

[0021] Combining pressure gradient information with geographic constraints, a local flood diversion potential energy model is constructed for each geographic information grid unit. Based on the flood diversion potential energy model, a dynamic potential vector field containing directionality and intensity information is generated to form a flood diversion driving potential flow field covering the flood storage area.

[0022] Preferably, the method for identifying the linked instability propagation chain and the instability propagation path includes:

[0023] A graph structure based on the dynamic stagnant pressure map is constructed, with geographic information grid cells as nodes and the direction and intensity of stagnant pressure transmission between grids as edges, forming a potential graph structure. The temporal changes in stagnant pressure at each node in the potential graph structure are collected and analyzed, and the response intensity of each node under external disturbances is calculated to form a disturbance sensitivity index.

[0024] Based on the time sliding window method, the paths between nodes whose disturbance sensitivity rise rate is greater than the preset disturbance sensitivity rise rate threshold are extracted from the potential graph structure and regarded as disturbance priority conduction paths. All disturbance priority conduction paths are identified and aggregated according to the time evolution trend. The path weight clustering method is used to divide the linkage instability propagation chains and connect them to form a disturbance propagation chain map.

[0025] Dynamically track the disturbance propagation chain map, record the priority transmission paths of new disturbances, path change trends and trigger frequency changes of nodes, construct a structural set of linked instability propagation paths, and indicate the systemic instability risk sections through the structural set.

[0026] Preferably, the method for dynamically tracking the time derivative of the instability propagation tension gradient includes:

[0027] Based on the constructed linked instability propagation path structure set, the stored pressure change value of the key node in each disturbance propagation chain is obtained. Combined with the instability propagation path direction and the stored pressure difference of the adjacent nodes, the instability propagation tension value of the instability propagation path in the current time window is estimated.

[0028] Based on the sliding time window, the unstable propagation tension value change sequence on each unstable propagation path is continuously sampled, and the tension change rate of each unstable propagation path at different time points is calculated to form the unstable propagation tension gradient time series; the first-order derivative and second-order derivative of the unstable propagation tension gradient time series are dynamically calculated.

[0029] Preferably, the method of generating a phase change trigger flag and dividing the multi-stage risk warning levels includes:

[0030] Continuously monitor the geographic information grid cells covering the flood storage area and dynamically calculate the pressure center of gravity of the area based on the changes in the storage pressure in each geographic information grid cell. The pressure center of gravity is determined by the weighted centroid of the storage pressure values ​​of each grid cell.

[0031] Using the sliding time window technique, we obtain the trajectory of the pressure center of gravity at different moments, and calculate the moving speed and acceleration of the pressure center of gravity. In the edge areas of the flood diversion driving flow field, we perform statistical analysis of the activation frequency changes per unit time for the geographic information grid cells in the edge areas.

[0032] Based on the activation frequency changes per unit time, the activation frequency change trends of edge grid cells are counted; if the activation frequency change trends of different edge grid cells are greater than the preset activation frequency change trend threshold, it is determined that the edge grid cell has spillover risks or signs of diffusive instability;

[0033] Combined with the analysis of the instability propagation tension of the instability propagation path, a time-varying risk trigger criterion function is constructed to determine whether the current instability propagation path has entered a risk phase change state. If it has entered a risk phase change state, a phase change trigger flag is generated and multi-stage risk warning levels are divided.

[0034] Preferably, the method for outputting dynamically executable control instructions includes:

[0035] Analyze the current stage risk warning level and the corresponding preset response target, which includes target type, target area, response time and expected control range; based on the preset response target, retrieve the response strategy template that matches the current spatial location, risk type and risk warning level in the preset response strategy library;

[0036] Evaluate the matching degree of different response strategy templates and select the instruction combination that best meets the current response requirements; convert the instruction combination into an instruction format that can be recognized and executed by the system to form executable control instructions, and push the executable control instructions to the corresponding response terminal or control platform.

[0037] Preferably, the method for acquiring the disaster development information and external environment monitoring data includes:

[0038] Collect and control disaster development information and external environmental monitoring data related to flood storage areas. Disaster development information includes water level information, flow rate information, rainfall information, soil moisture content information, structural deformation information, dangerous situation alarm information, and manual inspection records. External environmental monitoring data includes upstream water volume data, downstream drainage capacity data, surrounding terrain change data, regional traffic status data, and water depth data in surrounding villages.

[0039] Disaster development information and external environment monitoring data are spatially aggregated according to a unified geographic information grid, and time-aligned based on a unified time scale. A sliding time window mechanism is used to continuously sample and dynamically update disaster development information and external environment monitoring data.

[0040] Preferably, the method for correcting the pressure factor distribution in the dynamic stagnation pressure map includes:

[0041] The spatial mean square error is used to calculate the spatial distribution deviation between the predicted evolution trajectory of the flood storage and detention area and the actual evolution trajectory. When the deviation is greater than the preset deviation threshold, the response tuning mechanism is automatically triggered to determine whether there is a risk misjudgment or a mismatch in the response strategy.

[0042] The pressure factor distribution in the dynamic storage pressure map is corrected. After the correction operation is completed, the risk phase change state of each instability propagation path is re-evaluated according to the updated dynamic storage pressure map. If the higher level risk warning criteria are met, it will automatically switch to the next stage of risk warning level and issue a new executable control instruction.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention constructs a potential graph network based on a geographic information grid structure, comprehensively estimates the path tension based on the accumulated pressure difference and spatial distance between nodes, and introduces a topological conductivity factor to correct the tension contribution, thereby establishing a physically reasonable instability propagation tension calculation model; dynamically samples the tension value sequence of each unstable path through a sliding time window, obtains the first-order derivative and the second-order derivative, and can sensitively identify whether the risk in the system is in a stage of continuous enhancement and accelerated evolution, providing theoretical support for triggering graded warnings. By constructing a disturbance propagation chain map and an unstable path structure set, this method not only identifies risks on a single path, but also characterizes chain coupling areas, achieving early positioning of linked large-scale instability risks, which is superior to traditional independent single-point analysis methods. By calculating the first-order and second-order derivatives of the propagation tension in real time and quantitatively analyzing its changing trends, it can automatically determine the multi-stage evolution process of the risk from normal state, metastable state, critical point, unstable state to phase transition state, thereby supporting the system to achieve step-by-step intelligent warning.

[0045] A dynamic calculation method for the location, velocity, and acceleration of the pressure center of gravity is proposed, enabling continuous, multi-time series monitoring of changes in stagnant pressure within geographic information grid cells. This method can proactively identify trends in pressure concentration and rapid movement within stagnant areas, providing a macroscopic understanding of the accumulation and migration of stagnant pressure and global awareness. By statistically analyzing the activation frequency trends of edge geographic cells, it dynamically identifies whether a region is at risk of diffusive instability. This method addresses the inability of traditional methods to accurately depict areas of boundary expansion and instability, demonstrating edge-sensitivity. A sliding time window mechanism is combined with real-time center of gravity analysis to enhance the ability to continuously monitor and perceive trends in the evolution of stagnant areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a schematic diagram of the structure of the flood storage and detention area risk intelligent perception and multi-stage progressive early warning system of the present invention;

[0047] Figure 2 A flow chart of the method for mapping multi-source heterogeneous data to a standardized geographic information grid provided by the present invention;

[0048] Figure 3 This is a flow chart of the intelligent perception and multi-stage progressive early warning method for flood storage and detention area risks of the present invention. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0050] Example 1

[0051] See also Figure 1 and Figure 2 As shown, Example 1 further illustrates the intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas proposed by the present invention, including:

[0052] Flood storage and detention areas, as important functional areas for flood regulation and diversion, play a critical buffering role during extreme rainfall or basin-wide flooding events. However, with the intensification of climate extremes and the increasing complexity of the surface environment, the risk transmission mechanisms within flood storage and detention areas are becoming increasingly complex, and traditional early warning methods face significant challenges in terms of timeliness, accuracy, and dynamic adaptability.

[0053] Existing flood storage and detention area risk warning systems mostly rely on static threshold judgments, water level overlimit monitoring, or historical experience deduction. They lack the ability to dynamically model the risk diffusion process within the storage and detention area. Especially in areas with complex spatial structures and significant terrain undulations, traditional methods have difficulty capturing the true dynamic behavior of flood propagation in different paths and signs of local instability, resulting in blind spots in risk identification.

[0054] Current approaches generally ignore dynamic indicators such as rate and acceleration of risk evolution, making it difficult to promptly identify sudden changes in critical risk states, known as phase transition points or disaster inflection points. This directly limits the ability to build a multi-stage progressive early warning mechanism, resulting in risk responses remaining at the level of a single threshold trigger, lagging early warning effectiveness, and the coexistence of redundant or insufficient responses.

[0055] Existing systems often use static indicators or empirical judgments to categorize warning levels, lacking a real-time dynamic indicator system that is coupled to the actual evolution of disasters. Differences between grid cells in pressure response and disturbance transmission are not systematically modeled, and changes in risk across different regions are not quantified as transmission path weights. This results in low accuracy in identifying unstable chains and inaccurate prediction of path evolution trends.

[0056] Traditional early warning models are mostly based on linear deduction or rule engine models, and lack an understanding of nonlinear evolution characteristics and the coupled behavior of complex systems. In particular, they are unable to cope with characteristics such as sudden changes, discontinuous evolution or chain reactions that may occur in the system during the evolution of the disaster situation. Therefore, they cannot support the needs of accurate, efficient, dynamic and visual intelligent early warning.

[0057] In order to effectively solve the above problems, the present invention proposes an intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas, including:

[0058] The flood storage and detention area perception module collects multi-source heterogeneous data covering the flood storage and detention area, builds a unified spatiotemporal coding framework, maps the multi-source heterogeneous data to a standardized geographic information grid, and adopts a confidence control mechanism to form a dynamic storage and detention pressure map;

[0059] The linked instability propagation module constructs the flood diversion driven flow field based on the dynamic storage pressure map, continuously tracks the topological disturbance characteristics in the flood diversion driven flow field, and identifies the linked instability propagation chain and instability propagation path;

[0060] The risk phase change identification module dynamically tracks the time derivative of the instability propagation tension gradient based on the instability propagation chain and path. Combining the center of gravity migration trend and edge activation frequency in the flood diversion driven flow field, it establishes a time-varying risk phase change criterion, generates a phase change trigger flag, and divides the risk warning level into multiple stages.

[0061] After identifying the phase change trigger flag, the intent-guided response module retrieves matching instruction combinations from the preset response strategy library based on the preset response target corresponding to the current stage risk warning level, and outputs dynamic executable control instructions;

[0062] The adaptive feedback calibration module continuously obtains disaster development information and external environmental monitoring data after the implementation of executable control instructions, calculates the deviation between the predicted evolution trajectory and the actual evolution trajectory of the flood storage and detention area, and if it is greater than the preset deviation threshold, triggers the response tuning mechanism, corrects the pressure factor distribution in the dynamic storage and detention pressure map, and automatically enters the next stage of risk warning.

[0063] Methods for mapping multi-source heterogeneous data to standardized geographic information grids include:

[0064] Collect multi-source heterogeneous data covering the flood storage area, including hydrological monitoring data, geological structure monitoring data, meteorological remote sensing data, and group behavior dynamics data; build a unified spatiotemporal coding framework to perform time alignment and spatial standardization on the multi-source heterogeneous data. Time alignment involves reconstructing data with different sampling frequencies to a unified time scale, and spatial standardization involves projecting the multi-source heterogeneous data onto standardized geographic information grid units divided under a preset geographic coordinate system.

[0065] Through the rasterized data mapping method, multi-source heterogeneous data in different structural forms such as point type, trajectory type and regional type are filled into the preset target grid area to form a multi-factor spatial data structure covering the flood storage and detention area.

[0066] The method of forming a dynamic stagnation pressure map includes:

[0067] The sources of multi-source heterogeneous data falling into each geographic information grid cell are identified, and corresponding confidence weights are preset according to the collection method and update frequency of the multi-source heterogeneous data. When the same type of data from different sources exists in the same geographic information grid cell, weighted fusion is performed according to their respective confidence weights to obtain the fused grid pressure value.

[0068] A continuity analysis is performed on the changing trend of grid pressure values ​​between adjacent geographic information grids. When a discontinuous geographic information grid is detected, the multi-source heterogeneous data corresponding to the discontinuous geographic information grid is corrected based on the changing trend of the grid pressure values ​​after fusion of the surrounding geographic information grids. Different confidence weights are dynamically adjusted, and the spatial distribution of the fused grid pressure values ​​is visualized to form a dynamic stagnant pressure map.

[0069] The construction methods of flood diversion driven flow field include:

[0070] The pressure values ​​of each geographic information grid cell in the dynamic storage pressure map are extracted, and the pressure gradient information in each direction is calculated based on the spatial adjacency relationship between geographic information grid cells. Based on the geographic constraints between the geographic information grid cells, a coupling relationship network between the geographic information grid cells is established to form a structural map reflecting the pressure transmission path. The geographic constraints include spatial connectivity, terrain elevation characteristics, water connectivity, and soil permeability.

[0071] Combining pressure gradient information with geographic constraints, a local flood diversion potential energy model is constructed for each geographic information grid unit. Based on the flood diversion potential energy model, a dynamic potential vector field containing directionality and intensity information is generated to form a flood diversion driving potential flow field covering the flood storage area.

[0072] Methods for identifying linked instability transmission chains and instability transmission paths include:

[0073] A graph structure based on the dynamic stagnant pressure map is constructed, with geographic information grid cells as nodes and the direction and intensity of stagnant pressure transmission between grids as edges, forming a potential graph structure. The temporal changes in stagnant pressure at each node in the potential graph structure are collected and analyzed, and the response intensity of each node under external disturbances is calculated to form a disturbance sensitivity index.

[0074] Based on the time sliding window method, the paths between nodes whose disturbance sensitivity rise rate is greater than the preset disturbance sensitivity rise rate threshold are extracted from the potential graph structure and regarded as disturbance priority conduction paths. All disturbance priority conduction paths are identified and aggregated according to the time evolution trend. The path weight clustering method is used to divide the linkage instability propagation chains and connect them to form a disturbance propagation chain map.

[0075] Dynamically track the disturbance propagation chain map, record the priority transmission paths of new disturbances, path change trends and trigger frequency changes of nodes, construct a structural set of linked instability propagation paths, and indicate the systemic instability risk sections through the structural set.

[0076] Methods for dynamically tracking the time derivative of the instability propagation tension gradient include:

[0077] Based on the constructed linked instability propagation path structure set, the stored pressure change value of the key node in each disturbance propagation chain is obtained. The instability propagation tension value of the instability propagation path in the current time window is estimated by combining the instability propagation path direction and the stored pressure difference of the adjacent nodes.

[0078] The propagation tension values ​​are: ;in, Indicates at a point in time The instability propagation tension value of the path in the current time window; Indicates at a point in time Geographic Information Grid The stagnation pressure; Indicates at a point in time Geographic Information Grid The stagnation pressure; Represents the spatial distance between two geographic information grids; Represents the topological conductivity of the unstable propagation path. According to expert experience, The value range is between 0 and 1; and Represents the index of the geographic information grid; An index representing a time point;

[0079] Based on the sliding time window, the unstable propagation tension value change sequence on each unstable propagation path is continuously sampled, and the tension change rate (i.e., tension gradient) of each unstable propagation path at different time points is calculated to form an unstable propagation tension gradient time series; the first-order derivative (i.e., change rate) and second-order derivative (i.e., change acceleration) of the unstable propagation tension gradient time series are dynamically calculated.

[0080] Defining the instability propagation path At the time point The instability propagation tension value is , then the first-order derivative of the instability propagation tension gradient time series is: ;in, It represents the first derivative of the time series of the instability propagation tension gradient, reflecting the rate of change of tension; Indicates that at the next time point The instability propagation tension value; Indicates a time interval;

[0081] The second-order derivative of the instability propagation tension gradient time series is: ;in, The second derivative of the time series of the tension gradient representing the propagation of instability, the change in the rate of change of tension (i.e., acceleration); Indicates that at the next time point The instability propagation tension value;

[0082] This approach addresses the following issues with existing technologies: Existing flood storage and detention area warning methods often rely on static thresholds, historical water level data, or manual experience, making it difficult to reflect the dynamic evolution of tension as floods propagate along different paths within complex geographical structures. They also lack the ability to quantitatively perceive the risk propagation trends within interconnected instability paths. Existing methods focus primarily on the occurrence or total amount of risk, while ignoring key characteristics such as its rate of change and acceleration. This delays the identification of critical phase transition points and prevents early multi-stage warnings. The spatial grids within the flood zone are complex in topology, and pressure changes in different regions have a non-uniform impact on each other. Existing models lack a path connectivity weighting mechanism, resulting in crude tension estimation and inaccurate identification of instability chains.

[0083] Compared to existing technologies, this method offers several advantages: It constructs a potential graph network based on a geographic information grid structure, comprehensively estimates path tension using the stagnant pressure difference and spatial distance between nodes, and introduces a topological conductivity factor to modify the tension contribution, thereby establishing a physically plausible instability propagation tension calculation model. By dynamically sampling the tension value sequence of each instability path through a sliding time window and obtaining the first-order derivative (rate) and second-order derivative (acceleration), it can sensitively identify whether the risk in the system is in a stage of continuous intensification or accelerated evolution, providing theoretical support for triggering graded warnings. By constructing a disturbance propagation chain map and a set of instability path structures, this method not only identifies risks on a single path but also characterizes chain-coupling regions, enabling early identification of linked, large-scale instability risks, surpassing traditional independent single-point analysis methods. By calculating the first- and second-order derivatives of the propagation tension in real time and quantitatively analyzing their changing trends, it can automatically determine the multi-stage evolution of risk from normal, metastable, critical, unstable, to phase transition states, thereby supporting the system's implementation of progressive intelligent warnings.

[0084] The method of generating a phase change trigger flag and dividing the multi-stage risk warning level includes:

[0085] Continuously monitor the geographic information grid cells covering the flood storage area and dynamically calculate the pressure center of gravity of the area based on the changes in the storage pressure in each geographic information grid cell. The pressure center of gravity is determined by the weighted centroid of the storage pressure values ​​of each grid cell.

[0086] Using the sliding time window technique, we obtain the trajectory of the pressure center of gravity at different moments, and calculate the moving speed and acceleration of the pressure center of gravity. In the edge areas of the flood diversion driving flow field, we perform statistical analysis of the activation frequency changes per unit time for the geographic information grid cells in the edge areas.

[0087] Based on the activation frequency changes per unit time, the activation frequency change trends of edge grid cells are counted; if the activation frequency change trends of different edge grid cells are greater than the preset activation frequency change trend threshold, it is determined that the edge grid cell has spillover risks or signs of diffusive instability;

[0088] Combined with the instability propagation tension analysis of the instability propagation path, a time-varying risk trigger criterion function is constructed to determine whether the current instability propagation path has entered a risk phase transition state.

[0089] The time-varying risk trigger criterion function is: ;in, Indicates the The instability propagation path at time point The phase change trigger flag indicates whether the path enters the risk phase change state; Indicates the preset tension change rate threshold; Indicates the preset tension acceleration threshold;

[0090] when , indicating the path The propagation state reaches or exceeds the preset non-steady evolution critical condition; all The number of paths and their spatial distribution can be used as the basis for judging the overall risk level of the system. The path proportion and spatial concentration are used to divide the risk warning levels into multiple stages.

[0091] If the risk phase change state is entered, a phase change trigger flag is generated and multi-stage risk warning levels are divided. Multi-stage risk warning levels include ;in, Indicates at a point in time Risk warning level at that time; Indicates a low risk warning level; Indicates a medium risk warning level; Indicates a high-risk warning level; Indicates at a point in time The proportion of unstable propagation paths in the risk phase transition state when ; Indicates the first level threshold of risk warning; Indicates the first level threshold of risk warning;

[0092] The following problems existing in existing technologies are solved: Traditional flood storage and detention area warning methods are mostly based on fixed thresholds or static indicators, which cannot dynamically respond to the spatial evolution characteristics and mutation trends of risk expansion, resulting in delayed warnings. Existing systems mostly use linear models or stage judgment rules, which make it difficult to identify mutation points or phase change nodes in the development of disasters, and lack a mechanism to capture the critical state of disaster spread. Traditional warning level classification methods often rely on manually set static rules and lack data-driven real-time dynamic adjustment mechanisms, making it difficult to adapt to sudden disaster scenarios under complex hydrogeological conditions. In existing solutions, risk level classification is often based on empirical rules or subjective judgments, lacking a real-time dynamic indicator system coupled with actual disaster evolution, and is prone to false alarms or omissions.

[0093] Compared with the existing technology, the beneficial effects are as follows: a dynamic calculation method for the position, velocity and acceleration of the pressure center of gravity is proposed, and the changes in the stagnant pressure in the geographic information grid unit are continuously monitored in multiple time series; the trend of pressure concentration and rapid movement within the stagnant area can be identified in advance, and the accumulation and migration of stagnant pressure can be perceived from a macro perspective, with global perception. By statistically analyzing the activation frequency change trend of the edge geographic unit, it is possible to dynamically identify whether the region has the risk of diffusive instability; it makes up for the problem that traditional methods cannot accurately depict the boundary expansion instability area, and has edge perception sensitivity. By analyzing the first-order and second-order derivatives of the propagation tension gradient, a time-varying risk trigger criterion function is introduced to automatically judge whether the unstable propagation path has entered a phase change state, effectively solving the problem that the existing technology cannot automatically identify the risk mutation point, and realizing the identification and early warning from quantitative change to qualitative change; the sliding time window mechanism is combined with real-time center of gravity analysis to improve the continuous monitoring and trend perception capabilities of the evolution process of the stagnant area.

[0094] The method of outputting dynamic executable control instructions includes:

[0095] Analyze the current stage risk warning level and the corresponding preset response target. The preset response target includes the target type, target area, response time and expected control range. Based on the preset response target, retrieve the response strategy template that matches the current spatial location, risk type and risk warning level from the preset response strategy library. The response strategy template includes the specific response behavior, execution unit, control object, trigger condition and execution feedback requirements.

[0096] Evaluate the matching degree of different response strategy templates and select the instruction combination that best meets the current response requirements; convert the instruction combination into an instruction format that can be recognized and executed by the system to form executable control instructions, and push the executable control instructions to the corresponding response terminal or control platform.

[0097] Methods for obtaining disaster development information and external environment monitoring data include:

[0098] Collect and control disaster development information and external environmental monitoring data related to flood storage areas. Disaster development information includes water level information, flow rate information, rainfall information, soil moisture content information, structural deformation information, dangerous situation alarm information, and manual inspection records. External environmental monitoring data includes upstream water volume data, downstream drainage capacity data, surrounding terrain change data, regional traffic status data, and water depth data in surrounding villages.

[0099] Disaster development information and external environment monitoring data are spatially aggregated according to a unified geographic information grid, and time-aligned based on a unified time scale. A sliding time window mechanism is used to continuously sample and dynamically update disaster development information and external environment monitoring data.

[0100] Methods for correcting the pressure factor distribution in the dynamic stagnation pressure map include:

[0101] The spatial mean square error is used to calculate the spatial distribution deviation between the predicted evolution trajectory of the flood storage and detention area and the actual evolution trajectory. When the deviation is greater than the preset deviation threshold, the response tuning mechanism is automatically triggered to determine whether there is a risk misjudgment or a mismatch in the response strategy.

[0102] The pressure factor distribution in the dynamic stagnant pressure map is corrected. The correction includes recalculating or adjusting the confidence weights of different types of monitoring data, adjusting the stagnant pressure propagation conductance factors in key areas, updating the tension gain parameters of abnormal grid cells in the map, and adjusting the activation threshold of edge activation areas.

[0103] After completing the correction operation, the risk phase change state of each instability propagation path is re-evaluated according to the updated dynamic accumulation pressure map. If the higher level risk warning criteria are met, it will automatically switch to the next stage of risk warning level and issue a new executable control instruction.

[0104] The preset semantic centrality score threshold is set by the staff. By collecting different semantic centrality scores, the average of multiple semantic centrality scores is taken as the preset semantic centrality score threshold; similarly, the preset switching threshold and the preset cosine similarity threshold are set.

[0105] This embodiment constructs a potential graph network based on a geographic information grid structure, comprehensively estimating path tension using the stagnant pressure difference and spatial distance between nodes. A topological conductivity factor is introduced to modify the tension contribution, thereby establishing a physically plausible instability propagation tension calculation model. By dynamically sampling the tension value sequence of each instability path through a sliding time window and obtaining the first- and second-order derivatives, this method can sensitively identify whether the risk in the system is in a stage of continuous strengthening and accelerated evolution, providing theoretical support for triggering graded warnings. By constructing a disturbance propagation chain map and a set of instability path structures, this method not only identifies risks on a single path but also characterizes chain-coupled regions, enabling early identification of linked, large-scale instability risks, surpassing traditional independent single-point analysis methods. By calculating the first- and second-order derivatives of the propagation tension in real time and quantitatively analyzing its changing trends, it can automatically determine the multi-stage evolution of risk from normal, metastable, critical, unstable, to phase transition states, thereby supporting the system's implementation of progressive intelligent warnings.

[0106] A dynamic calculation method for the location, velocity, and acceleration of the pressure center of gravity is proposed, enabling continuous, multi-time series monitoring of changes in stagnant pressure within geographic information grid cells. This method can proactively identify trends in pressure concentration and rapid movement within stagnant areas, providing a macroscopic understanding of the accumulation and migration of stagnant pressure and global awareness. By statistically analyzing the activation frequency trends of edge geographic cells, it dynamically identifies whether a region is at risk of diffusive instability. This method addresses the inability of traditional methods to accurately depict areas of boundary expansion and instability, demonstrating edge-sensitivity. A sliding time window mechanism is combined with real-time center of gravity analysis to enhance the ability to continuously monitor and perceive trends in the evolution of stagnant areas.

[0107] Example 2

[0108] See also Figure 3 As shown, for the parts not described in detail in this embodiment, please refer to the description of Example 1, which provides a method for intelligent perception of flood storage and detention area risks and multi-stage progressive early warning, including:

[0109] S1. Collect multi-source heterogeneous data covering the flood storage and detention area, build a unified spatiotemporal coding framework, map the multi-source heterogeneous data to a standardized geographic information grid, and adopt a confidence control mechanism to form a dynamic storage and detention pressure map;

[0110] S2. Construct a flood diversion driven flow field based on the dynamic storage pressure map, continuously track the topological disturbance characteristics in the flood diversion driven flow field, and identify the linked instability propagation chain and instability propagation path;

[0111] S3. Based on the instability propagation chain and instability propagation path, the time derivative of the instability propagation tension gradient is dynamically tracked. Combined with the center of gravity migration trend and edge activation frequency in the flood diversion driven flow field, a time-varying risk phase change criterion is established, a phase change trigger flag is generated, and a multi-stage risk warning level is divided;

[0112] S4. After identifying the phase change trigger flag, according to the preset response target corresponding to the current stage risk warning level, a matching instruction combination is retrieved from the preset response strategy library and a dynamic executable control instruction is output;

[0113] S5. Continuously obtain disaster development information and external environmental monitoring data after the implementation of executable control instructions, calculate the deviation between the predicted evolution trajectory and the actual evolution trajectory of the flood storage and detention area, and if it is greater than the preset deviation threshold, trigger the response tuning mechanism, correct the pressure factor distribution in the dynamic storage and detention pressure map, and automatically enter the next stage of risk warning.

[0114] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters and thresholds in the formulas are set by technicians in this field according to actual conditions.

[0115] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiment. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for users of ordinary skill in the art, various improvements and modifications that do not depart from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. Intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas, characterized by: include: The flood storage and detention area perception module collects multi-source heterogeneous data covering the flood storage and detention area, builds a unified spatiotemporal coding framework, maps the multi-source heterogeneous data to a standardized geographic information grid, and adopts a confidence control mechanism to form a dynamic storage and detention pressure map; The method for forming a dynamic stagnation pressure map includes: The sources of multi-source heterogeneous data falling into each geographic information grid cell are identified, and corresponding confidence weights are preset according to the collection method and update frequency of the multi-source heterogeneous data. When the same type of data from different sources exists in the same geographic information grid cell, weighted fusion is performed according to their respective confidence weights to obtain the fused grid pressure value. Conduct continuity analysis on the changing trends of grid pressure values ​​between adjacent geographic information grids. When a discontinuous geographic information grid is detected, the multi-source heterogeneous data corresponding to the discontinuous geographic information grid is corrected based on the changing trends of the grid pressure values ​​after fusion of the surrounding geographic information grids. Dynamically adjust different confidence weights, and visualize the spatial distribution of the fused grid pressure values ​​to form a dynamic stagnation pressure map. The linked instability propagation module constructs the flood diversion driven flow field based on the dynamic storage pressure map, continuously tracks the topological disturbance characteristics in the flood diversion driven flow field, and identifies the linked instability propagation chain and instability propagation path; The method for constructing the flood diversion driving situation flow field includes: The pressure values ​​of each geographic information grid cell in the dynamic stagnant pressure map are extracted, and the pressure gradient information in each direction is calculated based on the spatial adjacency relationship between geographic information grid cells. Based on the geographical constraints between the geographic information grid cells, a coupling relationship network between the geographic information grid cells is established to form a structural map reflecting the pressure transmission path. Combining pressure gradient information with geographic constraints, a local flood diversion potential energy model is constructed for each geographic information grid cell. Based on the flood diversion potential energy model, a dynamic potential vector field containing directionality and intensity information is generated to form a flood diversion driving potential flow field covering the flood storage area. The method for identifying the linked instability propagation chain and the instability propagation path includes: A graph structure based on the dynamic stagnant pressure map is constructed, with geographic information grid cells as nodes and the direction and intensity of stagnant pressure transmission between grids as edges, forming a potential graph structure. The temporal changes in stagnant pressure at each node in the potential graph structure are collected and analyzed, and the response intensity of each node under external disturbances is calculated to form a disturbance sensitivity index. Based on the time sliding window method, the paths between nodes whose disturbance sensitivity rise rate is greater than the preset disturbance sensitivity rise rate threshold are extracted from the potential graph structure and regarded as disturbance priority conduction paths. All disturbance priority conduction paths are identified and aggregated according to the time evolution trend. The path weight clustering method is used to divide the linkage instability propagation chains and connect them to form a disturbance propagation chain map. Dynamically track the disturbance propagation chain map, record the priority transmission paths of new disturbances, path change trends, and changes in node triggering frequencies, construct a structure set of linked instability propagation paths, and use the structure set to indicate systemic instability risk sections; The risk phase change identification module dynamically tracks the time derivative of the instability propagation tension gradient based on the instability propagation chain and path. Combining the center of gravity migration trend and edge activation frequency in the flood diversion driven flow field, it establishes a time-varying risk phase change criterion, generates a phase change trigger flag, and divides the risk warning level into multiple stages. The method for dynamically tracking the time derivative of the instability propagation tension gradient includes: Based on the constructed linked instability propagation path structure set, the stored pressure change value of the key node in each disturbance propagation chain is obtained. The instability propagation tension value of the instability propagation path in the current time window is estimated by combining the instability propagation path direction and the stored pressure difference of the adjacent nodes. Based on a sliding time window, the instability propagation tension value change sequence on each instability propagation path is continuously sampled, and the tension change rate of each instability propagation path at different time points is calculated to form an instability propagation tension gradient time series; the first-order derivative and second-order derivative of the instability propagation tension gradient time series are dynamically calculated; The method of generating a phase change trigger flag and dividing the multi-stage risk warning levels includes: Continuously monitor the geographic information grid cells covering the flood storage and detention area, and dynamically calculate the pressure center of gravity of the flood storage and detention area based on the changes in the storage pressure in each geographic information grid cell. The pressure center of gravity is determined by the weighted centroid of the storage pressure value of each grid cell. Using the sliding time window technique, we obtain the trajectory of the pressure center of gravity at different moments, and calculate the moving speed and acceleration of the pressure center of gravity. In the edge areas of the flood diversion driving flow field, we perform statistical analysis of the activation frequency changes per unit time for the geographic information grid cells in the edge areas. Based on the activation frequency changes per unit time, the activation frequency change trends of edge grid cells are counted; if the activation frequency change trends of different edge grid cells are greater than the preset activation frequency change trend threshold, it is determined that the edge grid cell has spillover risks or signs of diffusive instability; Combined with the instability propagation tension analysis of the instability propagation path, a time-varying risk trigger criterion function is constructed to determine whether the current instability propagation path has entered a risk phase transition state. If it has entered a risk phase transition state, a phase transition trigger flag is generated and multi-stage risk warning levels are divided. After identifying the phase change trigger flag, the intent-guided response module retrieves matching instruction combinations from the preset response strategy library based on the preset response target corresponding to the current stage risk warning level, and outputs dynamic executable control instructions; The adaptive feedback calibration module continuously obtains disaster development information and external environmental monitoring data after the implementation of executable control instructions, calculates the deviation between the predicted evolution trajectory and the actual evolution trajectory of the flood storage and detention area, and if it is greater than the preset deviation threshold, triggers the response tuning mechanism, corrects the pressure factor distribution in the dynamic storage and detention pressure map, and automatically enters the next stage of risk warning.

2. The intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas according to claim 1 is characterized by: The method for mapping multi-source heterogeneous data to a standardized geographic information grid includes: Collect multi-source heterogeneous data covering the flood storage area, including hydrological monitoring data, geological structure monitoring data, meteorological remote sensing data, and group behavior dynamics data; build a unified spatiotemporal coding framework to perform time alignment and spatial standardization on the multi-source heterogeneous data; Through the rasterized data mapping method, multi-source heterogeneous data in different structural forms such as point type, trajectory type and regional type are filled into the preset target grid area to form a multi-factor spatial data structure covering the flood storage and detention area.

3. The intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas according to claim 2 is characterized by: The method for outputting a dynamically executable control instruction includes: Analyze the current stage risk warning level and the corresponding preset response target, which includes target type, target area, response time and expected control range; based on the preset response target, retrieve the response strategy template that matches the current spatial location, risk type and risk warning level in the preset response strategy library; Evaluate the matching degree of different response strategy templates and select the instruction combination that best meets the current response requirements; convert the instruction combination into an instruction format that can be recognized and executed by the system to form executable control instructions, and push the executable control instructions to the corresponding response terminal or control platform.

4. The intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas according to claim 3 is characterized by: The method for obtaining disaster development information and external environment monitoring data includes: Collect and control disaster development information and external environmental monitoring data related to flood storage areas. Disaster development information includes water level information, flow rate information, rainfall information, soil moisture content information, structural deformation information, dangerous situation alarm information, and manual inspection records. External environmental monitoring data includes upstream water volume data, downstream drainage capacity data, surrounding terrain change data, regional traffic status data, and water depth data in surrounding villages. Disaster development information and external environment monitoring data are spatially aggregated according to a unified geographic information grid, and time-aligned based on a unified time scale. A sliding time window mechanism is used to continuously sample and dynamically update disaster development information and external environment monitoring data.

5. The intelligent risk perception and multi-stage progressive early warning system for flood storage and detention areas according to claim 4 is characterized in that: The method for correcting the pressure factor distribution in the dynamic stagnation pressure map includes: The spatial mean square error is used to calculate the spatial distribution deviation between the predicted evolution trajectory of the flood storage and detention area and the actual evolution trajectory. When the deviation is greater than the preset deviation threshold, the response tuning mechanism is automatically triggered to determine whether there is a risk misjudgment or a mismatch in the response strategy. The pressure factor distribution in the dynamic storage pressure map is corrected. After the correction operation is completed, the risk phase change state of each instability propagation path is re-evaluated according to the updated dynamic storage pressure map. If the higher level risk warning criteria are met, it will automatically switch to the next stage of risk warning level and issue a new executable control instruction.

Images