Multi-dimensional dynamic coupling water conservancy project toughness evaluation method

Abstract

The invention discloses a multi-dimensional dynamic coupling water conservancy project toughness evaluation method. The method comprises the following steps: S1, data acquisition; s2, disaster chain dynamic simulation: training a GAN model by using the data collected in the step S1, the trained GAN model being used for generating virtual scenes of different disaster combinations, constructing a "structural mechanics-hydrodynamics-ecological model" coupling framework, simulating dynamic evolution after disasters occur, and generating multiple groups of disaster chain evolution processes; s3, calculating a comprehensive toughness index, and evaluating the engineering toughness of each group of disaster chain evolution process generated in the step S2; according to the method, a multi-dimensional dynamic coupling technology system is constructed, the precision and timeliness of toughness evaluation of the water conservancy project are improved, compared with a traditional static evaluation method, the technology realizes dynamic simulation of a disaster chain cascade process and quantitative analysis of multi-system interaction influence, and a more scientific solution is provided for safety management of the water conservancy project.

Description

Technical Field

[0001] This invention belongs to the field of intelligent decision-making technology for water conservancy projects, specifically involving a dynamic assessment method for the resilience of water conservancy projects that integrates digital twins, disaster chain simulation, and multi-objective optimization. Background Technology

[0002] Water engineering resilience refers to the ability of a water system to maintain its basic functions, recover quickly, and adapt to new environments when faced with uncertainties such as natural disasters, climate change, and human interference. It emphasizes the system's disaster resistance, adaptability, and sustainability, and is one of the important concepts in modern water engineering. The core elements of water engineering resilience include: 1. Disaster resistance: The ability to withstand extreme events such as floods, droughts, and earthquakes, such as the structural strength of facilities like dams and reservoirs.

[0003] 2. Rapid Recovery: After a disaster, the system can be quickly repaired and restored to normal operation, such as emergency drainage systems and backup water sources.

[0004] 3. Adaptability: It can adjust its operation mode according to climate change or social needs, such as flexibly allocating water resources and dynamically adjusting flood control standards.

[0005] 4. Sustainability: Maintaining a long-term balance between water supply and demand, and reducing negative impacts on the ecological environment, such as ecological water conservancy and sponge cities.

[0006] The intensification of global climate change has led to frequent extreme rainstorms, floods, earthquakes and other disasters. The safety risks of water conservancy projects are showing new characteristics of multi-hazard coupling and chain transmission. In order to cope with these challenges, resilience assessment technology has gradually become a research hotspot.

[0007] However, most existing resilience assessment patents are static assessments, which are difficult to match the dynamic evolution of disaster chains. This contradiction is particularly prominent in the field of multi-hazard chain simulation: for example, the HAZUS-MH model developed by the Federal Emergency Management Agency (FEMA) can simulate single disasters such as earthquakes and floods, but it cannot simulate the disaster chain transmission and coupling process of "flood-induced dam damage → seepage exacerbating ecological damage → secondary social risks"; existing evolution models of "glacial till - landslide - debris flow" constructed for landslide dam failure have made breakthroughs in local mechanisms, but still lack the ability to dynamically characterize the time-varying characteristics of multi-system coupling.

[0008] It is evident that existing methods are unable to quantify the real-time interactive impacts of structural safety, ecological environment, and socio-economic systems, nor can they adapt to the dynamic path of the disaster chain of "trigger-transmission-amplification," resulting in assessment results that lag behind the rapid decision-making needs of complex engineering scenarios.

[0009] Therefore, there is an urgent need to propose a multi-dimensional, dynamically coupled resilience assessment method for water conservancy projects. By constructing a collaborative response mechanism of the structure-ecology-social system, focusing on the entire life cycle evolution of the disaster chain, dynamic risk quantification and decision optimization can be achieved, providing technical support for the full life cycle management of water conservancy projects. Summary of the Invention

[0010] The purpose of this invention is to provide a multi-dimensional dynamic coupling method for assessing the resilience of water conservancy projects. By establishing a multi-dimensional dynamic coupling mechanism, this method addresses three major technical challenges in assessing the resilience of water conservancy projects: the difficulty in quantifying the real-time interactive impact of structural-ecological-social systems; the lack of simulation of the dynamic evolution of the disaster chain throughout its entire cycle of "triggering → transmission → amplification"; and the serious disconnect between assessment results and emergency decision-making. The aim is to improve the accuracy of risk prediction under extreme scenarios.

[0011] To achieve the above objectives, the technical solution of this invention is as follows: A multidimensional, dynamically coupled method for assessing the resilience of hydraulic engineering projects, the method comprising: S1: Data acquisition, which is completed through multi-source data fusion; S2: Disaster chain dynamic simulation. Using the data collected in step S1, a GAN model is trained. The trained GAN model is used to generate virtual scenarios with different disaster combinations. The dynamic evolution of each virtual scenario after the disaster occurs is simulated, generating multiple sets of disaster chain evolution processes. S3: Calculate the comprehensive resilience index for each disaster chain evolution process generated in step S2, and evaluate the engineering resilience of the complex disaster scenario.

[0012] Furthermore, the data acquisition in step S1 includes deploying testing equipment to collect engineering and ecological data, and extracting industry standards and decision-making rules.

[0013] Furthermore, the S2 step disaster chain dynamic simulation includes: S21: Disaster scenario generation. The historical data collected in step S1 is used to train a generative adversarial network (GAN). Engineering parameters and ecological parameters are input into the trained GAN model to form a risk database for multiple scenarios. S22: Coupled structural mechanics, hydrodynamics, and ecological models, using the initial scenario generated in step S21 as boundary conditions, to achieve dynamic simulation of the cascading disaster chain process, including: A finite element model was used to simulate the transverse cracks that occur in the middle of the dam body due to flooding. An ecological degradation module coupled with a hydrodynamic model was used to predict that river flow interruption would lead to the destruction of fish habitats and an increase in COD concentration. Based on the simulation platform, the study simulates how congestion on evacuation routes leads to residents having to detour and an increase in average evacuation time.

[0014] Furthermore, the S3 step engineering toughness assessment includes: S31: Construction of a three-level indicator system: Establish a three-level indicator system for structural resilience, ecological resilience, and social resilience, and set thresholds for each indicator based on measured data and industry standards. 32: Dynamic weight adjustment, calculating the initial weights of each indicator. When the monitored value exceeds the threshold, an automatic correction is triggered, and the corrected weight is calculated. ; (1) Where r ij Here, n represents the number of standardized metrics, and m represents the number of scenarios. (2) Where mi is the basic probability assignment function of index i, The evidence theory conflict factor for index i; S33: Calculate the comprehensive resilience index The comprehensive resilience index RI is calculated based on a weighted average. (3) (4) Where i represents the primary indicator and j represents the secondary indicator. Let be the weight of the j-th secondary indicator under the i-th primary indicator. The standardized value of the primary indicator; in formulas (3) and (4), and The initial weights, or the modified weights. and When the monitored value does not exceed the threshold, the initial weight is used. and When the monitored value exceeds the threshold, the corrected weight is used. and .

[0015] An apparatus includes at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a multidimensional dynamically coupled hydraulic engineering resilience assessment method.

[0016] A computer-readable storage medium storing a computer program that, when executed by a processor, implements a multi-dimensional dynamically coupled method for assessing the resilience of hydraulic engineering projects.

[0017] The technical solution of this invention has the following technical effects: 1. This invention significantly improves the accuracy and timeliness of resilience assessment for water conservancy projects by constructing a multi-dimensional dynamic coupling technology system. Compared with traditional static assessment methods, this technology realizes dynamic simulation of the cascading disaster chain process and quantitative analysis of the interactive effects of multiple systems, effectively overcoming the problems of dimensional fragmentation and decision-making lag.

[0018] 2. At the engineering application level, the technical solution optimizes the emergency response process, shortens the decision-making cycle, and enhances the comprehensive risk resistance capacity of water conservancy projects under extreme disasters through the synergistic optimization of multi-dimensional indicators such as structural safety, ecological environment, and socio-economic factors. This invention provides a more scientific and efficient solution for the safety management of water conservancy projects and has significant application value for water resource management, flood control, and ecological protection in the context of global climate change. Attached Figure Description

[0019] Figure 1 This is a flowchart of the technical solution of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only some, not all, of the embodiments of this invention, and are not intended to limit the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0021] This invention provides a multi-dimensional resilience assessment method for water conservancy projects. Through the coordinated operation of four modules—data fusion, disaster simulation, dynamic assessment, and decision optimization—it achieves real-time and accurate assessment of the resilience of water conservancy projects. The assessment method includes: S1: Data acquisition, which is completed through multi-source data fusion sensing. This invention collects data by deploying monitoring equipment and by using knowledge graphs, specifically including: S11: Monitoring Equipment Deployment and Data Acquisition Displacement gauges and piezometers are deployed to monitor engineering deformation and seepage pressure in real time. Multispectral water quality sensors are used to acquire ecological parameters such as COD (chemical oxygen demand) and ammonia nitrogen, and fish acoustic telemetry equipment is used to track biological activity. Simultaneously, anonymous mobile phone signaling and emergency supplies databases are accessed to dynamically capture population distribution and resource status. COD is an important indicator for measuring the relative content of organic pollutants in water, representing the amount of oxygen consumed (in mg / L) by oxidizable organic matter and reducing inorganic matter in water under the action of strong oxidants (such as potassium dichromate or potassium permanganate). It indirectly reflects the degree of organic pollution in water bodies and is one of the key parameters for water quality monitoring. S12: Knowledge Graph Construction Knowledge graph technology is a semantic knowledge base that stores and represents entities and their relationships through a graph structure. It can transform complex data and information into a structured form that is easy to understand and process, supporting various applications such as intelligent search, recommendation systems, and natural language processing, thereby improving the accuracy and efficiency of information retrieval. This invention constructs an "engineering-environment-society" causal relationship network and extracts decision rules from industry standards based on natural language processing technology.

[0022] In this invention, step S12 serves as the knowledge hub of the entire resilience assessment method. Its role is to provide structured causal logic and standardized industry rules for subsequent disaster simulation, resilience assessment, and decision optimization. This is specifically reflected in two aspects: Firstly, a semantic association network of "engineering-environment-society" is established. By storing scattered entities and their relationships in a graph structure through knowledge graphs, the problem of "semantic fragmentation" of multi-source data is solved, enabling computers to understand the chain-like association of "disaster factors-engineering response-ecological impact-social consequences".

[0023] Secondly, it solidifies the decision-making logic in industry standards by analyzing industry standards using natural language processing technology, extracting computable decision rules, and transforming vague textual standards into machine-executable logical rules, thus providing a "compliance basis" for subsequent evaluation and decision-making.

[0024] S2: Disaster Chain Dynamic Simulation Using the data collected in step S1, a GAN model is trained. The trained GAN model is then used to generate virtual scenarios with different disaster combinations, constructing a coupled framework of "structural mechanics-hydraulics-ecological model" to simulate the dynamic evolution after a disaster occurs. This helps researchers analyze the cumulative effects of disasters. By generating a large number of possible disaster combinations, the destructive extent of different complex disasters can be assessed. In this invention, scenarios are generated through a GAN model, and then dynamic simulation of the disaster chain is achieved through coupled structural mechanics, hydraulics, and ecological models. Specifically, this includes: S21: Disaster Scenario Generation Historical flood and earthquake data collected in step S1 are used to train a generative adversarial network (GAN), which outputs boundary conditions for complex disaster scenarios. Uncertainty parameters are then injected using the Monte Carlo method to form a risk database for multiple scenarios.

[0025] S22: Coupled Calculation of Cascaded Processes By coupling structural mechanics, hydrodynamics, and ecological models, dynamic simulation of the cascading disaster chain process of "engineering damage → ecological degradation → social risk" can be achieved.

[0026] In this invention, step S21 trains a generative adversarial network (GAN) based on the preprocessed data from step S1. By inputting the probability distributions of engineering and ecological parameters, multiple initial scenarios containing single or combined disasters such as floods, earthquakes, and droughts are generated. Step S22 constructs a coupled framework of "structural mechanics-hydrodynamics-ecological model," using the initial scenarios generated in step S21 as boundary conditions to simulate the cascading disaster chain process.

[0027] Among them, the structural damage module is based on the dynamic damage evolution simulation of the finite element method.

[0028] Modeling method: The three-dimensional finite element method is used to construct the mechanical model of the engineering structure. The material properties of the structure are characterized by the division of solid elements, and the material degradation criteria are defined based on the damage mechanics theory.

[0029] Load processing: The disaster scenario parameters generated by S21 are converted into mechanical loads, and the superposition effect of water flow impact load, structural self-weight, and seismic inertial force is calculated by fluid-structure interaction algorithm.

[0030] Damage calculation logic: Real-time monitoring of stress and strain status of key parts of the structure. When the local stress exceeds the yield strength of the material, crack propagation simulation is initiated. Based on fracture mechanics theory, the length, width and depth of the crack are dynamically updated and converted into performance degradation parameters such as structural permeability coefficient and bearing capacity, providing input for the hydrodynamic module.

[0031] The ecological degradation module is used for coupled simulation of hydrodynamic and ecological processes.

[0032] Hydrodynamic driving method: A hydrodynamic model is constructed using numerical methods of fluid mechanics to simulate water level changes, velocity field distribution and water exchange process under disaster scenarios, and output hydrodynamic parameters such as flow velocity, water depth and hydraulic residence time.

[0033] An ecological response simulation method is coupled with a species suitability index model. Using hydrodynamic parameters as input, the suitability of key habitat factors is quantified, and the usable area and destruction rate of habitat are calculated. Based on the law of conservation of mass, a water quality migration and transformation model is constructed to simulate the diffusion and degradation process of pollutants under the action of water flow. Combined with the water reoxygenation rate, the dynamic changes of water quality indicators are calculated.

[0034] The ecological chain transmission links parameters such as water quality deterioration and habitat destruction with species survival rate and population recovery cycle, quantifying the degree of ecosystem degradation and its chain reaction.

[0035] The social risk module is used for coupled simulation of multi-agent and network analysis. Population dynamic modeling is based on mobile phone signaling and census data, and uses spatial interpolation to construct a population distribution model, updating the population size and spatial location of affected areas in real time.

[0036] The evacuation process simulation method uses a multi-agent model to simulate individual evacuation behavior and sets decision rules for the agents; it couples the traffic network model to convert the flooding range output by the hydrodynamic module into road resistance parameters of the road network, and calculates the congestion index and travel time of the evacuation route; based on the system dynamics method, it correlates evacuation efficiency with social resources and simulates secondary social risks such as evacuation delays and material shortages.

[0037] The module coupling logic and methods are as follows: The data transfer mechanism establishes cross-module data interfaces and uses spatiotemporal synchronous interpolation to achieve dynamic mapping of parameters between different models.

[0038] The coupling logic achieves a chain transmission of "engineering performance degradation → hydrological process change → ecological and social response" through the unidirectional drive of "structural damage module → hydrodynamic module → ecological / social module". When the ecological module outputs water quality exceeding the standard or the social module outputs critical states such as evacuation and congestion, it triggers the parameter correction of the hydrodynamic model in the reverse direction, forming a closed-loop optimization coupling system.

[0039] S3: Dynamic Quantitative Assessment of Toughness The engineering resilience of the composite disaster scenario is assessed under the evolution process of each disaster chain generated in step S2.

[0040] S31: Construction of a Three-Level Indicator System A three-tiered indicator system of structural resilience, ecological resilience, and social resilience was established. Thresholds for each indicator were set based on measured data and industry standards, and dimensionless processing was performed on each indicator.

[0041] Table 1 Three-level index system for resilience assessment

[0042] S32: Dynamic Weight Adjustment Initial weights of each indicator were calculated based on the improved TOPSIS method. It also introduces an evidence theory conflict factor β, which automatically triggers correction when the monitored value exceeds a threshold, and calculates the corrected weight. : (1) Where r ij Here, n represents the number of standardized metrics, and m represents the number of scenarios. Regarding the standardized index value r in the resilience assessment method for water conservancy projects ij The results are obtained based on measured data and indicator thresholds, and the specific processing method is as follows: Positive indicators, such as disaster resilience and water self-purification capacity, indicate better resilience, with higher values ​​representing greater resilience. The processing method involves direct normalization to the [0,1] interval, calculated using the ratio of measured values ​​to a threshold. The formula can be understood as: Standardized value = Measured value / Threshold (when measured value ≤ threshold). If the measured value exceeds the threshold, the standardized value remains 1.

[0043] Negative indicators, such as damage evolution rate and species recovery cycle, indicate better resilience, with smaller values. The standardization method uses the formula "standardized value = 1 - measured value / threshold" (when the measured value ≤ threshold). If the measured value exceeds the threshold, the standardized value is set to 0.

[0044] (2) Where, m i Let i be the basic probability assignment function. Let be the evidence theory conflict factor for index i.

[0045] In evidence theory, A represents propositions such as "the system is resilient" and ¬A represents its complement, namely "the system is not resilient". It is the level of confidence in the support for "good resilience" as indicated by indicator i. (¬A) represents the level of confidence in the support for "poor resilience," reflecting the evidentiary information conveyed by indicator i. There is no unified formula for calculating it; it needs to be determined based on the data source or prior knowledge of the specific problem. Common methods include: based on data statistics or frequency, based on expert experience or subjective judgment, based on derivation from other models, or based on distance or similarity.

[0046] S33: Calculate the comprehensive resilience index The overall resilience index RI is calculated based on a weighted average.

[0047] (3) (4) Where i represents the primary indicator and j represents the secondary indicator. Let be the weight of the j-th secondary indicator under the i-th primary indicator. This is the standardized value of the primary indicator.

[0048] In formulas (3) and (4), and It can be either the initial weights or the modified weights. and The specific choice depends on the computing scenario, and the selection criteria are as follows: This paper clarifies that dynamic weight adjustment is a core component of resilience assessment; when the monitored value does not exceed the threshold, the initial weight is used. and When the monitored value exceeds the threshold, the conflict factor of the evidence theory is used. Adjust the weights according to formula (2) and use the adjusted weights and sums. Formulas (3) and (4) are the general calculation framework for the comprehensive resilience index RI. They do not limit the use of only the initial weights. In actual calculation, the weights need to be dynamically input according to the stage of disaster evolution. If it is in the early stage of disaster, the indicators are all normal, so the initial weights are substituted in. If the indicators have exceeded the threshold, such as the crack propagation rate exceeding the standard, then the corrected weights are substituted in. and .

[0049] In this invention, the core function of the S2 step disaster chain dynamic simulation is to provide "dynamic data during the disaster evolution process" for the S3 step resilience dynamic quantitative assessment. The S2 step connects "S1 basic data acquisition" with the S3 "resilience quantitative assessment," specifically in the following two aspects: The first aspect: transforming static basic data into dynamic disaster scenario data. The measured data in step S11, such as the initial crack width monitored by sensors and the COD baseline value, and the industry standard in step S12, such as the crack propagation rate threshold of 0.5 mm / h, are "baseline state data" that can only reflect the static state before the disaster occurs.

[0050] Step S2 simulates the dynamic evolution after a disaster occurs through "disaster scenario generation in step S21" and "cascaded process coupling calculation in step S22": For example, based on the initial crack data from step S11, step S22 simulates flooding causing cracks to expand from 0.8 mm to 5 mm at a rate of 1.2 mm / h, and outputs the dynamic change value of "damage evolution rate". Based on the COD baseline value of 10 mg / L in step S11, step S22 simulates the rise of COD concentration to 25 mg / L after the river is cut off, and outputs the dynamic change value of "COD concentration decay rate".

[0051] These dynamic data are the source of the "measured values" for the three-level indicators in S31, such as crack propagation rate and COD decay rate, providing real-time input for the quantification of indicators in the course of disaster evolution.

[0052] The second aspect: supporting the dynamic and comprehensive nature of resilience assessment. The S2 step generates 1,000 sets of composite disaster scenarios (including uncertainty parameters), covering disaster situations of different intensities and combinations, such as "rainstorm + earthquake" and "small flood + landslide". This makes the assessment in the S3 step no longer limited to a single static scenario, but can reflect the resilience changes under multiple risks.

[0053] For example, step S2 simulates the rapid collapse of the dam under the scenario of "extreme flood", and step S3 can calculate the situation where the structural toughness drops sharply and the social evacuation time increases dramatically under this scenario; when simulating the minor damage to the dam under the scenario of "mild flood", step S3 can calculate the situation where the toughness decreases slowly.

[0054] This dynamic scenario data ensures that the "dynamic weighting" and "comprehensive resilience index (RI) calculation" in S3 can adapt to the real-time evolution of disasters, avoiding the assessment lag caused by relying solely on static data from S1.

[0055] Therefore, step S2 is the "dynamic data source" of step S3. By simulating the cascading process of the disaster chain, the basic data of step S1 is transformed into the "real-time measured values ​​of indicators" and "multi-scenario risk inputs" required by step S3, ultimately supporting the upgrade of resilience assessment from "static threshold comparison" to "dynamic evolution quantification".

[0056] S4: Adaptive Decision Optimization This invention also includes the application of a multi-dimensional dynamically coupled hydraulic engineering resilience assessment method. When a disaster scenario occurs, constraints are set based on the core safety requirements of structural, ecological, and social systems. Structural safety constraints: Disaster redundancy shall not be lower than the industry's conventional design standards, and the damage evolution rate shall not exceed the tolerance limit of the engineering materials; Ecological protection constraints: the species recovery cycle shall not exceed the critical time for the self-repair of the regional ecosystem, and the water quality self-purification capacity shall not be lower than the basic requirements of the surface water functional zone; Social emergency constraints: Population evacuation efficiency meets the bottom line requirements of emergency management, and the elasticity of economic losses does not exceed the regional economic tolerance threshold.

[0057] With the goal of maximizing the resilience index RI and minimizing the costs of ecological restoration and social evacuation, a multi-objective optimization model is constructed. An evolutionary algorithm is used to generate an initial set of schemes, and reinforcement learning is used to iteratively optimize the model in a simulation environment to output a Pareto optimal policy set. The scheme with the maximum resilience index RI and the minimum costs of ecological restoration and social evacuation is taken as the optimal water conservancy project scheme.

[0058] The present invention also discloses an apparatus comprising at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the multidimensional dynamically coupled hydraulic engineering resilience assessment method.

[0059] The present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the multidimensional dynamically coupled hydraulic engineering resilience assessment method.

[0060] The specific implementation of this invention is described in detail using a reservoir in the Qingjiang River Basin of Hubei Province as an example.

[0061] S1: Multi-source data fusion sensing S11: Monitoring Equipment Deployment and Data Acquisition Two GNSS displacement gauges (model: TrimbleR10) are installed on the upstream and downstream slopes of the reservoir dam, with a sampling frequency of 10Hz, to monitor the horizontal / vertical displacement of the dam in real time; five piezometers (model: vibrating wire piezometer VWP-400) are installed in the dam foundation to monitor changes in seepage pressure.

[0062] Three multispectral water quality sensors (model: YSIEXO2) are deployed at the center of the reservoir and the outlet to collect parameters such as COD, ammonia nitrogen, and dissolved oxygen every 30 minutes; an acoustic telemetry array for fish (8 receiving stations) is installed in the downstream river channel to track the activity trajectories of species such as carp and crucian carp.

[0063] By accessing anonymous mobile signaling data from telecom operators and anonymizing it using blockchain technology, the real-time distribution of 500,000 downstream residents (spatial resolution 100m×100m) is obtained, and the emergency supplies database (containing the supplies reserves of 30 shelters) is accessed simultaneously.

[0064] S12: Knowledge Graph Construction Based on the Neo4j knowledge graph engine, we analyze regulatory texts such as the "Regulations on the Safety Management of Reservoirs and Dams" to construct causal networks. For example: dam cracks → increased soil moisture content → water accumulation in residential areas → congestion of evacuation routes.

[0065] S2: Disaster Chain Dynamic Simulation S21: Disaster Scenario Generation GAN (Generative Adversarial Network) is a deep learning model consisting of a generator and a discriminator. GAN can generate synthetic data for training disaster prediction models. GAN can generate virtual scenarios of different disaster combinations to help researchers analyze the superposition effects of disasters. By generating a large number of possible disaster combinations, the destructive extent of different compound disasters can be assessed.

[0066] In this invention, flood and earthquake data from the Qingjiang River Basin over 73 years (1950-2023) are input to train a GAN model. The GAN model then generates composite disaster scenarios, injecting uncertainties (soil permeability coefficient variation range ±10%, rainfall forecast error ±10%) using Monte Carlo methods, generating 1000 risk scenarios. Typical scenario parameters are as follows: Scenario 1: Actual rainfall of 550 mm in 24 hours, peak flow of 8800 m3 / s, and soil permeability coefficient of 1.1 times the baseline value; Scenario 2: Actual rainfall of 450 mm in 24 hours, peak flow of 7200 m3 / s, and soil permeability coefficient of 0.9 times the baseline value.

[0067] S22: Coupled Calculation of Cascaded Processes Structural damage module: Using a finite element model, it simulates the transverse cracks that occur in the middle of the dam body due to flooding. Ecological degradation module: Coupled with a hydrodynamic model, it predicts that river flow interruption will lead to the destruction of fish habitats and an increase in COD concentration; Social Risk Module: Based on a simulation platform, this module simulates how congestion along evacuation routes can lead to residents having to detour, increasing the average evacuation time.

[0068] S3: Dynamic Quantitative Assessment of Toughness S31: Construction of a Three-Level Indicator System A three-tiered indicator system for structural resilience, ecological resilience, and social resilience is established, with thresholds set based on measured data and industry standards, as detailed below.

[0069] Table 2 Thresholds for each indicator in the three-level indicator system for resilience assessment

[0070] Each indicator is standardized based on the measured value and the threshold. Positive indicators (such as disaster resistance redundancy) are directly normalized to [0,1]. Negative indicators (such as damage evolution rate) are standardized to R=1-measured value / threshold. When the threshold is exceeded, the standardized value is 0.

[0071] Corresponding to Tables 1 and 2: the first primary indicator is structural toughness, disaster redundancy is the first secondary indicator under the first primary indicator, and damage evolution rate is the second secondary indicator under the first primary indicator.

[0072] The second primary indicator is ecological resilience, the species recovery cycle is the first secondary indicator under the second primary indicator, and the water quality self-purification capacity is the second secondary indicator under the second primary indicator.

[0073] The third primary indicator is social resilience, with population dispersal efficiency being the first secondary indicator under the third primary indicator, and economic loss elasticity being the second secondary indicator under the third primary indicator.

[0074] S32: Dynamic Weight Adjustment An improved TOPSIS method was adopted, which combined entropy weighting and analytic hierarchy process to calculate the initial weights of each indicator. When the monitored value exceeded the threshold, the weights were adjusted by introducing the conflict factor β of evidence theory. For example, 12 hours after the flood, the crack expansion rate exceeded the threshold. The conflict factor β was calculated to be 0.4 according to evidence theory. The weight of the structural resilience indicator was adjusted to 0.26×(1-0.4)=0.16, and the weight of the ecological resilience indicator was increased to 0.36.

[0075] S33: Calculate the comprehensive resilience index The comprehensive resilience index RI is calculated based on the weighted average. The resilience calculation results for different time periods are shown in the table below.

[0076] Table 3. Real-time Calculation Results of Comprehensive Resilience Index

[0077] In the S3 step of the dynamic quantitative assessment of resilience, the core role of the GAN model in the S2 step is to provide dynamic input data for the calculation of the tertiary indicators, rather than directly calculating the values ​​of the tertiary indicators. The specific explanation is as follows.

[0078] The 1000 risk scenarios generated by the GAN model, such as different rainfall amounts, peak flood flows, and soil permeability coefficients, are prerequisites for calculating the tertiary indicators. In structural toughness, the calculation of damage evolution rate, i.e. crack propagation speed, depends on parameters such as flood inundation pressure and earthquake magnitude in the scenario.

[0079] In ecological resilience, the calculation of COD concentration decay rate depends on parameters such as the length of river flow interruption and the amount of ecological water replenishment in the scenario.

[0080] In terms of social resilience, the calculation of population evacuation rate depends on parameters such as the extent of road flooding caused by rainfall in the scenario.

[0081] The aforementioned scenario parameters are transformed into measured values ​​of tertiary indicators, such as crack propagation rate and evacuation time, through the disaster chain dynamic simulation in step S2. These values ​​are then standardized by combining the threshold values ​​in step S31 to finally obtain the quantified values ​​of the tertiary indicators.

[0082] Uncertainties introduced by Monte Carlo methods, such as variations in soil permeability coefficient (±10%) and rainfall forecast errors (±10%), ensure that the 1000 scenarios cover disaster situations ranging from the "most likely" to the "extreme." This provides multi-scenario validation for the dynamic weight correction in step S32. When an indicator exceeds a threshold in a certain scenario, such as a crack propagation rate greater than 0.5 mm / h, a conflict factor can be calculated based on evidence theory. This ensures that the weighting adjustments align with the changes in the importance of indicators under different disaster intensities.

[0083] S4: Adaptive Decision Optimization With the objectives of maximizing the resilience index and minimizing ecological restoration and social evacuation costs, and constrained by low water level ≤ flood limit level, COD ≤ 15 mg / L, and evacuation time ≤ 4 hours, a multi-objective optimization model was constructed. This model is the NSGA-II model, a classic multi-objective optimization algorithm primarily used to solve complex problems with multiple optimization objectives. It can find Pareto optimal solutions among multiple conflicting objectives. After 80 iterations using the NSGA-II algorithm, three non-dominated solutions were output: Solution A, with the highest resilience index (RI) and lowest ecological restoration and social evacuation costs, was prioritized by the decision-making system.

[0084] Table 4 Comparison of core parameters of toughness optimization schemes

[0085] Based on the measured data in this example, the dynamic coupling evaluation results of this invention are compared with the traditional static evaluation method. At the same time, the actual ecological and social response after the implementation of the decision-making scheme is verified. Finally, the disaster simulation error is 7.2%, which is 81% higher than the traditional method; the decision-making time is 12 minutes, which meets the requirements of the golden disposal period; the ecological flow guarantee rate is increased from 58% to 92%, and the downstream fish habitat retention rate reaches 90%.

[0086] The above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A multi-dimensional dynamically coupled method for assessing the resilience of hydraulic engineering projects, characterized in that, The method includes: S1: Data acquisition, which is completed through multi-source data fusion; S2: Disaster chain dynamic simulation. Using the data collected in step S1, a GAN model is trained. The trained GAN model is used to generate virtual scenarios with different disaster combinations, simulate the dynamic evolution of each virtual scenario after the disaster occurs, and generate multiple disaster chain evolution processes. S3: Calculate the comprehensive resilience index for each disaster chain evolution process generated in step S2, and evaluate the engineering resilience of the complex disaster scenario.

2. The multidimensional dynamic coupling method for assessing the resilience of hydraulic engineering projects according to claim 1, characterized in that... The S1 step of data acquisition includes deploying testing equipment to collect engineering and ecological data, and extracting industry standards and decision-making rules.

3. The multidimensional dynamic coupling method for assessing the resilience of hydraulic engineering projects according to claim 1, characterized in that, The S2 step disaster chain dynamic simulation includes: S21: Disaster scenario generation. The historical data collected in step S1 is used to train a generative adversarial network (GAN). Engineering parameters and ecological parameters are input into the trained GAN model to form a risk database for multiple scenarios. S22: Coupled structural mechanics, hydrodynamics, and ecological models, using the initial scenario generated in step S21 as boundary conditions, to achieve dynamic simulation of the cascading disaster chain process, including: A finite element model was used to simulate the transverse cracks that occur in the middle of the dam body due to flooding. An ecological degradation module coupled with a hydrodynamic model was used to predict that river flow interruption would lead to the destruction of fish habitats and an increase in COD concentration. Based on the simulation platform, the study simulates how congestion on evacuation routes leads to residents having to detour and an increase in average evacuation time.

4. The multidimensional dynamic coupling method for assessing the resilience of hydraulic engineering projects according to claim 1, characterized in that, The S3 step engineering toughness assessment includes: S31: Construction of a three-level indicator system: Establish a three-level indicator system for structural resilience, ecological resilience, and social resilience, and set thresholds for each indicator based on measured data and industry standards. S32: Dynamic weight adjustment, calculating the initial weights of each indicator. When the monitored value exceeds the threshold, an automatic correction is triggered, and the corrected weight is calculated. ; （1） Where r ij Here, n represents the number of standardized metrics, and m represents the number of scenarios. （2） Where mi is the basic probability assignment function of index i, The evidence theory conflict factor for index i; S33: Calculate the comprehensive resilience index The comprehensive resilience index RI is calculated based on a weighted average. （3） （4） Where i represents the primary indicator and j represents the secondary indicator. Let be the weight of the j-th secondary indicator under the i-th primary indicator. The standardized value of the primary indicator; in formulas (3) and (4), and The initial weights, or the modified weights. and When the monitored value does not exceed the threshold, the initial weight is used. and When the monitored value exceeds the threshold, the corrected weight is used. and .

5. A device, characterized in that, include: At least one processor; And a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1 to 4.

6. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 4.

Images