Digital twinborn-based risk-benefit dynamic game regulation and control method for flood storage and detention area

Abstract

The invention relates to the technical field of flood storage and detention area regulation and control, in particular to a flood storage and detention area risk-benefit dynamic game regulation and control method based on digital twinborn. Performing dynamic simulation on the real-time flood event based on a digital twinborn model to obtain probability distribution data; performing multi-target iterative optimization solution on the historical flood diversion regulation and control strategy based on the probability distribution data to generate a multi-target optimal balance set; generating a real-time controllable gate regulation and control instruction set on the basis of the digital twinborn model and the optimal tradeoff set; and in the process of real-time gate regulation and control through the gate regulation and control instruction set, carrying out quantitative evaluation on the regulation and control risk and benefit of the flood storage and detention area. According to the method, high-precision, dynamic and intelligent regulation and control of the flood storage and detention area are realized, accurate quantitative evaluation of risks and benefits of the flood storage and detention area is realized, a scientific basis is provided for flood control decision, and flood disaster loss and ecological influence are effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of flood storage and detention area regulation, and particularly to a dynamic game regulation method for flood storage and detention area risk-benefit based on digital twin. Background Art

[0002] Flood storage and detention areas are important components of the river flood control system and effective measures to ensure key flood control safety and reduce disasters. To ensure the flood control safety of key areas, it is necessary to develop areas with conditions into flood storage and detention areas and store and detain floods in a planned manner. This is the practical and economically reasonable need of the basin or regional flood control plan, and also a global consideration that has to sacrifice local interests for the overall situation. The start and regulation of flood storage and detention areas usually need to take into account the safety of people in the flood storage and detention areas, the minimization of inundation losses, and ecological impacts, etc.

[0003] However, traditional regulation methods face the following problems: ① Data fragmentation and model distortion: The monitoring system is separated from the decision-making model, and the parameter solidification leads to the accumulation of prediction errors; ② Dependence on static decision-making: Based on real-time data and fixed plans, it is difficult to respond to changes in dynamic hydrological conditions, especially when dealing with sudden disasters (such as sudden heavy rain, upstream dam break), it is even more ineffective; ③ There are limitations in multi-objective optimization: First, most decision-making models focus on flood control safety and ignore the coordination of inundation losses and ecology. There are no effective means to reduce the inundation losses in the flood storage and detention areas as much as possible while ensuring the flood control safety of the downstream. Second, a small number of models consider flood control safety, inundation losses, ecological impacts, etc. at the same time, but the quantification of uncertain risks is insufficient, which may lead to the situation that although the expected loss of the final decision is the smallest, the risk of getting out of control is relatively high.

[0004] Therefore, there is an urgent need for a dynamic game regulation method for flood storage and detention area risk-benefit that integrates digital twin. Combining with the construction of water conservancy digital twin, on the basis of ensuring the flood control safety of the downstream, it can realize the rapid quantitative assessment of the safety of people in the flood storage and detention areas, inundation losses, environmental impacts and potential risks, and propose the optimal regulation strategy for the flood storage and detention area. At the same time, it takes into account the advantages of dynamic prediction, rapid prediction, accurate prediction, etc. Summary of the Invention

[0005] The present invention overcomes the deficiencies of the prior art and provides a dynamic game regulation method for flood storage and detention area risk-benefit based on digital twin.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows: The first aspect of the present invention discloses a dynamic game regulation method for flood storage and detention area risk-benefit based on digital twin, including the following steps: Construct a digital twin model of the flood storage and detention area based on multi-dimensional data sources, and dynamically simulate real-time flood events based on the digital twin model to obtain data on flood evolution, inundation range and probability distribution of risks; Obtain the historical flood diversion regulation strategies for the flood storage and detention area, and perform multi-objective iterative optimization on the historical flood diversion regulation strategies based on the probability distribution data to generate an optimal trade-off set with multiple objectives; Based on the digital twin model and the optimal trade-off set, generate a real-time controllable gate regulation instruction set; During the process of real-time regulating the gates through the gate regulation instruction set, quantitatively evaluate the risks and benefits of the flood storage and detention area regulation.

[0007] Preferably, construct a digital twin model of the flood storage and detention area based on multi-dimensional data sources, specifically: Real-time collect hydrological, meteorological, topographical and engineering facility data through the deployed sensor network, and at the same time integrate the relevant data sets of historical flood events to form multi-dimensional data sources; Based on the multi-dimensional data sources, use spatial interpolation algorithms to refine and reconstruct the topographical data to generate a digital elevation model, and combine with a hydrological model to simulate the runoff and confluence processes of the basin; Dynamically calibrate the parameters of the hydrological model. By comparing the real-time monitoring data with the model simulation results, judge whether the model accuracy meets the preset accuracy requirements. If not, adjust the model parameters through an iterative optimization algorithm until the accuracy reaches the standard; Combine the calibrated hydrological model with the digital elevation model to construct a flood propagation simulation module, and use the finite element analysis method to dynamically simulate and predict the flood inundation range and water depth; At the same time, introduce ecological factor data to construct an ecological impact assessment module to quantitatively analyze the ecological risks in the flood inundation area; Form a digital twin model with multi-dimensional dynamic mapping by integrating the hydrological model, digital elevation model, flood propagation simulation module and ecological impact assessment module.

[0008] Preferably, dynamically simulate real-time flood events based on the digital twin model to obtain the probability distribution data of flood propagation, inundation range and risk, specifically: Input the real-time monitoring data into the digital twin model, start the flood propagation simulation module, and use the finite element analysis method to simulate the dynamic propagation process of the flood in the flood storage and detention area to generate real-time flood water level, flow velocity and flow direction data; Compare and analyze the simulation results with the data of historical flood events to judge the similarity between the current flood event and historical events. If the similarity reaches the preset similarity threshold, directly call the probability distribution data of historical flood events, otherwise go to the next step; Adopt the Monte Carlo simulation method, combine with the uncertainty characteristics of historical flood data, generate a large number of random flood scenarios, and use the digital twin model to simulate each scenario to calculate the flood propagation path, inundation range and water depth distribution; Extract the flood evolution path, inundation range, and water depth distribution generated by simulation as the input sample data for kernel density estimation; use the kernel density estimation algorithm to calculate the probability density of the sample data and generate the probability density functions of flood evolution, inundation range, and risk; Perform integral operations on the probability density functions to obtain the probability distribution curves of flood evolution, inundation range, and risk; convert the probability distribution curves into intuitive distribution maps through visualization techniques, and combine statistical characteristic values to obtain probability distribution data.

[0009] Preferably, perform multi-objective iterative optimization and solution on the historical flood diversion regulation strategy based on the probability distribution data to generate a multi-objective optimal trade-off set, specifically: Based on the probability distribution data of flood evolution, inundation range, and risk, use the Monte Carlo sampling method to generate an initial solution set, where each solution represents a historical flood diversion regulation strategy, and ensure that the solution set covers the multi-objective parameter space of flood control safety, inundation loss, ecological impact, and risk control; Define the flood control safety target as the downstream water level control threshold, the inundation loss target as the correlation function between the economic value model and the inundation area, the ecological impact target as the inundation index of the ecological sensitive area, and the risk control target as the potential dam break probability and the risk propagation model, and convert the four into quantifiable objective functions; According to the value range of the objective function, use the uniform distribution or the adaptive density method to generate a reference point set as the diversity distribution in the multi-objective parameter space; randomly select a reference point from the reference point set as the associated reference point; Perform non-dominated sorting on the initial solution set based on the associated reference point, divide the front rank of the initial solution set, and screen the initial solution set through the front rank to obtain the screened solution set; determine whether the screened solution set meets the diversity requirement, if not, select a new associated reference point in the reference point set to re-screen the initial solution set; Perform simulated binary crossover and polynomial mutation operations on the screened solution set to generate a new solution set and calculate its multi-objective fitness value, and combine the elitist retention strategy to merge the parent and offspring, and iteratively update the solution set; Set the preset hypervolume index change rate as the termination condition. If the hypervolume index change rate of the solution set is lower than the preset hypervolume index change rate, terminate the optimization and output the current optimal trade-off set.

[0010] Preferably, based on the digital twin model and the optimal trade-off set, generate a real-time controllable gate regulation instruction set, specifically: Based on the digital twin model, perform water flow simulation on real-time flood events, extract flood water levels, flow velocities, and inundation ranges, and combine the multi-objective strategy parameters in the optimal trade-off set to construct a state space to represent the global characteristics of the current flood scenario; Taking the gate opening, opening and closing sequence, and flood discharge flow as the core variables of the state space, defining a discretized or continuous state set; Pre-training the state space based on a deep neural network and mapping it into the digital twin model; In the flood environment simulated by the digital twin model, execute gate regulation actions and observe state transitions and reward values, and store the interaction data in the experience replay pool; Sample data from the experience replay pool, update the policy network and value network, generate the optimal gate regulation action in the current state, and convert the gate regulation action into an executable gate regulation instruction set.

[0011] Preferably, during the real-time regulation of the gate through the gate regulation instruction set, quantitatively evaluate the risks and benefits of the regulation of the flood detention and retarding area, specifically: Define the flood evolution process and the gate regulation strategy as the two sides of the game respectively. The flood evolution is the "opponent", and its uncertainty is simulated by the digital twin model. The gate regulation is the "decision maker"; Based on the non-cooperative game framework in game theory, construct a game model for flood evolution and gate regulation, define the objective functions of both sides. The flood evolution aims to maximize the inundation loss, and the gate regulation aims to minimize the risks and losses; According to the real-time simulation results of the digital twin model, generate a possible strategy set for flood evolution; Combine the flood evolution strategy set and the gate regulation instruction set, obtain the risk and benefit indicators under each strategy combination, and construct a game matrix; Extract the Nash equilibrium point of the game model, judge whether there is a state where neither side can obtain a better result by unilaterally changing the strategy. If it exists, take this state as the optimal game solution; Based on the Nash equilibrium solution, quantitatively evaluate the risks and benefits of the current gate regulation strategy, and generate a comprehensive evaluation result; According to the comprehensive evaluation result, judge whether the current gate regulation strategy meets the preset goal. If it does not meet, regenerate the gate regulation instruction set.

[0012] The multi-objectives include flood control safety, minimization of inundation loss, ecological impact, and potential risk control.

[0013] The second aspect of the present invention discloses a risk-benefit dynamic game regulation system for a flood detention and retarding area based on digital twin. The risk-benefit dynamic game regulation system for the flood detention and retarding area includes a memory and a processor. The memory stores a program for the risk-benefit dynamic game regulation method for the flood detention and retarding area. When the program for the risk-benefit dynamic game regulation method for the flood detention and retarding area is executed by the processor, the steps of any one of the risk-benefit dynamic game regulation methods for the flood detention and retarding area are implemented.

[0014] The present invention solves the technical defects existing in the background art and has the following beneficial effects: constructing a digital twin model of a flood detention and retention area based on multi-dimensional data sources, dynamically simulating real-time flood events based on the digital twin model, and obtaining data on flood evolution, inundation range, and probability distribution of risks; obtaining historical flood diversion control strategies for the flood detention and retention area, and performing multi-objective iterative optimization and solution on the historical flood diversion control strategies based on the probability distribution data to generate an optimal trade-off set with multiple objectives; generating a real-time controllable gate control instruction set based on the digital twin model and the optimal trade-off set; and quantitatively evaluating the risks and benefits of flood detention and retention area control during the process of real-time controlling the gates through the gate control instruction set. The present invention realizes high-precision, dynamic, and intelligent flood detention and retention area control, realizes accurate quantitative evaluation of the risks and benefits of the flood detention and retention area, provides a scientific basis for flood control decision-making, and effectively reduces flood disaster losses and ecological impacts. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can also obtain the drawings of other embodiments without creative efforts.

[0016] Figure 1 It is the overall method flow chart of the risk-benefit dynamic game control method for this flood detention and retention area; Figure 2 It is a partial method flow chart of the risk-benefit dynamic game control method for this flood detention and retention area; Figure 3 It is the system block diagram of the risk-benefit dynamic game control system for this flood detention and retention area. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] In order to be able to more clearly understand the above-mentioned objects, features, and advantages of the present invention, the following will further describe the present invention in detail with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0018] Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0019] As Figure 1 shown, the first aspect of the present invention discloses a risk-benefit dynamic game control method for a flood detention and retention area based on digital twin, including the following steps: S102. Build a digital twin model of the flood detention area based on multi-dimensional data sources, dynamically simulate real-time flood events based on the digital twin model, and obtain data on flood evolution, inundation range, and probability distribution of risks; S104. Obtain the historical flood diversion control strategies of the flood detention area, perform multi-objective iterative optimization and solution on the historical flood diversion control strategies based on the probability distribution data, and generate an optimal trade-off set of multiple objectives; the multiple objectives include flood control safety, minimization of inundation losses, ecological impacts, and potential risk control; S106. Based on the digital twin model and the optimal trade-off set, generate a real-time controllable gate control instruction set; S108. During the process of real-time controlling the gates through the gate control instruction set, quantitatively evaluate the risks and benefits of the regulation of the flood detention area.

[0020] Preferably, building a digital twin model of the flood detention area based on multi-dimensional data sources specifically includes: Real-time collect hydrological, meteorological, topographic, and engineering facility data through the deployed sensor network, and at the same time integrate the relevant data sets of historical flood events to form multi-dimensional data sources; Based on the multi-dimensional data sources, use spatial interpolation algorithms to refine and reconstruct the topographic data, generate a digital elevation model, and combine with a hydrological model to simulate the runoff and confluence processes of the basin; Dynamically calibrate the parameters of the hydrological model. By comparing the real-time monitoring data with the model simulation results, judge whether the model accuracy meets the preset accuracy requirements. If not, adjust the model parameters through an iterative optimization algorithm until the accuracy reaches the standard; It should be noted that topographic data is collected through the sensor network and remote sensing technology, including elevation points, contour lines, and topographic feature points, to form an initial topographic data set. Secondly, the collected data is preprocessed, including data cleaning, outlier removal, and missing value filling, to ensure data quality. Then, spatial interpolation is performed on the topographic data to generate a high-resolution digital elevation model (DEM), and terrain analysis tools are used to extract basin boundaries, river networks, and slope information. Then, a hydrological model is built based on the DEM, and a distributed hydrological simulation method is used to divide the basin into multiple sub-basins, and calculate the runoff and confluence processes of each sub-basin. Further, combined with rainfall data and soil characteristic parameters, simulate the runoff generation, confluence, and river channel evolution processes of the basin to generate dynamic data on runoff and flood propagation. Finally, by comparing the simulation results with the measured data, verify the accuracy of the hydrological model. If the preset threshold is not met, adjust the model parameters or re-interpolate the topographic data until the accuracy reaches the standard.

[0021] Combine the calibrated hydrological model with the digital elevation model to construct a flood propagation simulation module, and use the finite element analysis method to dynamically simulate and predict the flood inundation area and water depth; Meanwhile, introduce ecological factor data to construct an ecological impact assessment module to quantitatively analyze the ecological risks in the flood inundation area; It should be noted that ecological factor data is collected, including vegetation cover types, species distribution, wetland ranges, and locations of ecological sensitive areas, to form an ecological basic database. Secondly, based on the digital elevation model (DEM) and the flood propagation simulation results, data on the flood inundation area and water depth distribution are extracted. Using the spatial overlay analysis method, the inundation area is overlaid with the ecological factor data to identify the affected ecological sensitive areas and key species habitats. Then, an ecological risk assessment model is constructed. By quantifying the inundation duration, water depth changes, and ecological restoration ability, the risk index of the ecological sensitive areas is determined. Combining the ecological value assessment method, the loss of ecological service functions in the affected area is quantitatively analyzed, including water conservation, biodiversity, and carbon sequestration functions.

[0022] By integrating the hydrological model, digital elevation model, flood propagation simulation module, and ecological impact assessment module, a digital twin model with multi-dimensional dynamic mapping is formed.

[0023] The finally constructed digital twin model with multi-dimensional dynamic mapping in the present invention provides a scientific basis for the regulation of flood storage and detention areas, enhances the real-time nature and adaptability of flood control decisions, and effectively reduces flood risks and ecological impacts.

[0024] Preferably, based on the digital twin model, dynamic simulation of real-time flood events is carried out to obtain probability distribution data of flood propagation, inundation area, and risks, as Figure 2 shown, specifically: S202. Input the real-time monitoring data into the digital twin model, start the flood propagation simulation module, and use the finite element analysis method to simulate the dynamic propagation process of the flood in the flood storage and detention area to generate real-time flood water level, flow velocity, and flow direction data; Among them, the real-time monitoring data refers to relevant data such as hydrology, meteorology, topography, and engineering facilities that are collected in real time through a sensor network, remote sensing equipment, and other monitoring means deployed in the flood storage and detention area and its surrounding basins. These data include, but are not limited to, real-time water level, rainfall, flow velocity, flow rate, soil moisture, meteorological conditions (such as temperature, wind speed), and gate status, etc.

[0025] S204. Compare and analyze the simulation results with the data of historical flood events to judge the similarity between the current flood event and historical events. If the similarity reaches the preset similarity threshold, directly call the probability distribution data of the historical flood event; otherwise, proceed to the next step; S206. Use the Monte Carlo simulation method, combine with the uncertainty characteristics of historical flood data, generate a large number of random flood scenarios, and use the digital twin model to simulate each scenario, calculate the flood evolution path, inundation range and water depth distribution; S208. Extract the simulated flood evolution path, inundation range and water depth distribution as the input sample data for kernel density estimation; use the kernel density estimation algorithm to calculate the probability density of the sample data, and generate the probability density functions of flood evolution, inundation range and risk; S210. Perform integral operations on the probability density functions to obtain the probability distribution curves of flood evolution, inundation range and risk; convert the probability distribution curves into intuitive distribution maps through visualization technology, and combine with statistical characteristic values to obtain probability distribution data.

[0026] Among them, the probability distribution data includes the peak flow, inundation depth, duration, propagation path and influence range of historical floods, etc.

[0027] It should be noted that by comparing and analyzing the simulation results with the data of historical flood events to judge the similarity between the current flood event and historical events, if the similarity reaches the preset similarity threshold, directly call the probability distribution data of historical flood events, which can effectively reduce the model operation volume, avoid repeated calculations, maximize the efficiency, and have high reliability and stability. If the similarity does not reach the preset similarity threshold, obtain the probability distribution data through random scenario generation and simulation calculation, which is more dynamic and adaptable.

[0028] The present invention realizes the dynamic and accurate simulation of the flood propagation process, provides real-time water level, flow velocity and flow direction data; combines historical flood data with random scenario generation, enhances the quantitative analysis ability of uncertain factors; through probability density calculation and visualization technology, intuitively presents the flood risk distribution, provides a scientific basis for flood control decision-making, effectively improves the real-time performance and adaptability of flood storage and detention area regulation, and reduces the flood disaster risk.

[0029] Preferably, perform multi-objective iterative optimization solution on the historical flood diversion regulation strategy based on the probability distribution data to generate an optimal trade-off set of multiple objectives, specifically: Based on the probability distribution data of flood evolution, inundation range and risk, use the Monte Carlo sampling method to generate an initial solution set, each solution represents a historical flood diversion regulation strategy, and ensure that the solution set covers the multi-objective parameter space of flood control safety, inundation loss, ecological impact and risk control; Define the flood control safety target as the downstream water level control threshold, the inundation loss target as the correlation function between the economic value model and the inundated area, the ecological impact target as the inundation index of the ecological sensitive area, and the risk control target as the potential dam failure probability and the risk propagation model. Transform these four into quantifiable objective functions, including the flood control safety objective function, the inundation loss objective function, the ecological impact objective function, and the risk control objective function; Among them, the formula for the flood control safety objective function is:

[0030] In the formula, is the flood control safety index; is the actual water level at the th monitoring point; is the control threshold at the th monitoring point; is the total number of downstream water level monitoring points.

[0031] The formula for the inundation loss objective function is:

[0032] In the formula, is the inundation loss value; is the value coefficient of the kth inundated area; is the land inundated area of the kth; is the correction factor for the influence of inundation depth; is the total number of inundated areas.

[0033] The formula for the ecological impact objective function is:

[0034] In the formula, is the comprehensive ecological impact index; is the sensitivity coefficient of the mth ecological zone; is the inundated area of the mth ecological zone; is the inundation duration of the mth ecological zone; is the critical inundation time of the mth ecological zone; is the number of ecological sensitive area categories.

[0035] The formula for the risk control objective function is:

[0036] In the formula, is the comprehensive risk expectation value; is the dam failure probability; is the structural risk propagation coefficient; is the vulnerability index of the jth risk receptor; is the spatial propagation attenuation weight; is the direct / indirect risk weight factor; is the number of nodes in the risk propagation path.

[0037] According to the value range of the objective function, a reference point set is generated using the uniform distribution or the adaptive density method as the diversity distribution in the multi-objective parameter space; a reference point is randomly selected from the reference point set as the associated reference point; Among them, the reference point set is a set of preset points used to guide the search direction in the multi-objective optimization process. These points are evenly distributed in the multi-objective parameter space to ensure the diversity and global distribution balance of the solution set.

[0038] Based on the associated reference point, non-dominated sorting is performed on the initial solution set, the front rank of the initial solution set is divided, and the initial solution set is screened through the front rank to obtain the screened solution set; it is judged whether the screened solution set meets the diversity requirement. If not, a new associated reference point is selected from the reference point set to re-screen the initial solution set; It should be noted that based on the value ranges of the four objective functions of flood control safety, inundation loss, ecological impact, and risk control, the minimum and maximum values of each objective are determined; secondly, the uniform distribution method is used to generate a uniformly distributed reference point set in the objective space, or the adaptive density method is used to dynamically adjust the reference point density according to the weights and priorities of the objective functions; then, the generated reference point set is normalized to ensure the weight balance of each objective function. A reference point is randomly selected from the reference point set as the initial associated reference point to guide the screening and optimization of the solution set. By calculating the Euclidean distance between the solution and the reference point, the solution set is associated with the associated reference point to ensure the diversity distribution of the solution set in the objective space; finally, it is judged whether the distribution of the solution set meets the preset diversity requirement. If not, a new associated reference point is selected until the solution set distribution is balanced.

[0039] Perform simulated binary crossover and polynomial mutation operations on the screened solution set to generate a new solution set and calculate its multi-objective fitness value. Combine the elitist retention strategy to merge the parent generation and the offspring generation, and iteratively update the solution set; Set the preset hypervolume index change rate as the termination condition. If the hypervolume index change rate of the solution set is lower than the preset hypervolume index change rate, terminate the optimization and output the current optimal trade-off set.

[0040] It should be noted that parent individuals are randomly selected from the filtered solution set, and simulated binary crossover operations are used to generate offspring individuals. The position and number of crossover points are controlled by the crossover probability to ensure that the offspring inherit the excellent characteristics of the parents. Polynomial mutation operations are performed on the offspring individuals, and the gene values of the individuals are adjusted by the mutation probability to enhance the diversity of the solution set. The multi-objective fitness values of the offspring individuals are obtained, including flood control safety, inundation loss, ecological impact, and risk control. The parent and offspring individuals are combined, and non-dominated sorting and crowding degree calculation are used to screen the combined solution set, and individuals with higher fitness values are retained. Through the elitist retention strategy, the excellent individuals in the parent generation are directly retained to the next generation to ensure the global optimality of the solution. Finally, it is judged whether the solution set meets the iteration termination condition. If not, the crossover and mutation operations are returned and the iteration is continued for optimization until the optimal trade-off set is generated.

[0041] Based on probability distribution data, the present invention performs multi-objective iterative optimization and solution on historical flood diversion control strategies to generate a multi-objective optimal trade-off set, significantly improving the scientificity and balance of flood detention and retarding area control strategies. The finally generated optimal trade-off set can provide an efficient and balanced global strategy reference for flood detention and retarding area control, taking into account multiple objectives such as flood control safety, economy, ecological protection, and risk control.

[0042] Preferably, based on the digital twin model and the optimal trade-off set, a real-time controllable gate control instruction set is generated, specifically as follows: Based on the digital twin model, a water flow simulation of real-time flood events is carried out, the flood water level, flow velocity, and inundation range are extracted, and combined with the multi-objective strategy parameters in the optimal trade-off set, a state space is constructed to characterize the global characteristics of the current flood scenario. Among them, the multi-objective strategy parameters in the optimal trade-off set refer to the key parameters corresponding to the optimal solution set that can balance flood control safety, minimization of inundation loss, control of ecological impact, and risk control generated during the multi-objective optimization process. These parameters include specific operation variables such as gate opening, flood discharge flow, and regulation timing sequence, as well as quantitative indicators such as downstream water level control thresholds, economic value loss models, ecological sensitive area inundation indices, and dam-break risk propagation models. By integrating these parameters into the state space and the reward function, it can provide a scientific basis for real-time gate control, ensuring that the control strategy is efficient and operable while taking into account multiple objectives.

[0043] It should be noted that based on the extracted flood water level, flow velocity, inundation range, and risk distribution data, combined with the multi-objective strategy parameters in the optimal trade-off set, including the downstream water level control threshold, economic value loss model, ecological sensitive area inundation index, and dam-break risk propagation model, the core variables of the state space are defined. The extracted flood data is normalized to ensure that all variables are in the same dimension. The flood data and the multi-objective strategy parameters are integrated to construct a multi-dimensional state vector, which characterizes the global features of the current flood scenario. Through feature engineering methods, the state vector is dimensionally reduced or feature extracted to remove redundant information and improve the effectiveness of the state space.

[0044] The gate opening, opening and closing sequence, and flood discharge flow rate are used as the core variables of the state space, and a discrete or continuous state set is defined; Based on a deep neural network, the state space is pre-trained and mapped into the digital twin model; It should be noted that a deep neural network (DNN) architecture is constructed, including an input layer, hidden layers, and an output layer. The input layer receives the multi-dimensional vector of the state space (such as flood water level, flow velocity, inundation range, and multi-objective strategy parameters), and the output layer generates the predicted value of the gate regulation action. Based on the multi-objective strategy parameters in the optimal trade-off set, a pre-training data set is generated, including the state vector and the corresponding gate regulation action. The DNN is trained using backpropagation and gradient descent methods, and the network weights are optimized by minimizing the error between the predicted value and the actual value; then, the trained DNN is mapped into the digital twin model as a prediction module for real-time gate regulation.

[0045] In the flood environment simulated by the digital twin model, the gate regulation action is executed and the state transition and reward value are observed, and the interaction data is stored in the experience replay pool; Data is sampled from the experience replay pool to update the policy network and the value network, generate the optimal gate regulation action in the current state, and convert the gate regulation action into an executable gate regulation instruction set, including gate opening, opening and closing sequence, and flood discharge flow rate parameters.

[0046] It should be noted that in the digital twin model, predefined gate control actions (such as adjusting the gate opening, opening and closing sequence, and flood discharge flow) are executed, the changes in flood water level, flow velocity, and inundation range are observed, and the state transitions and corresponding reward values (including flood control safety, inundation losses, ecological impacts, and risk control) are obtained; the state transitions, reward values, and action information are stored in the experience replay pool as interaction data to ensure data diversity and historical record integrity; a batch of interaction data is randomly sampled from the experience replay pool as training samples; then, using the Deep Q-Network (DQN) or Proximal Policy Optimization (PPO) algorithm, the weights of the policy network and value network are updated based on the sampled data, and the prediction accuracy of the network is improved by minimizing the loss function. The updated policy network is used to generate the optimal gate control action in the current state to ensure the efficiency of the action while taking into account multiple objectives. The generated action is converted into an executable gate control instruction set, including specific gate opening, opening and closing sequence, and flood discharge flow parameters. Additionally, the execution effect of the instruction can be verified through the digital twin model to ensure the scientificity and operability of the control strategy.

[0047] It should be noted that observing state transitions and reward values refers to the process in which the system transfers from the current state to the next state and the corresponding immediate reward value after executing the gate control action in the flood environment simulated by the digital twin model, reflecting the effect of the action; interaction data is a collection of state transitions and reward values, stored in the experience replay pool, and used to record the interaction history between the algorithm and the environment; sampled data is a subset of interaction data randomly selected from the experience replay pool, used to update the policy network and value network to ensure the diversity and randomness of the training data.

[0048] Based on the digital twin model and the optimal trade-off set, the present invention generates a real-time controllable gate control instruction set, improving the intelligence and accuracy of flood storage and detention area regulation; the finally generated gate control instruction set can respond to the dynamic changes of floods in real time, providing a scientific and efficient control strategy that takes into account multiple objectives such as flood control safety, economy, ecological protection, and risk control.

[0049] Preferably, during the real-time regulation of the gate through the gate control instruction set, the risks and benefits of flood storage and detention area regulation are quantitatively evaluated, specifically as follows: The flood evolution process and the gate control strategy are respectively defined as the two sides of the game. The flood evolution is the "opponent", and its uncertainty is simulated by the digital twin model. The gate control is the "decision maker". Based on the non-cooperative game framework in game theory, a game model of flood evolution and gate control is constructed, and the objective functions of both sides are defined, including the flood evolution objective function and the gate control objective function. The flood evolution aims to maximize the inundation loss, and the gate control aims to minimize the risks and losses. Among them, the flood routing objective function:

[0050] In the formula, is the flood routing objective function; is the inundation loss coefficient; is the inundated area of the k-th area; is the inundation duration of the k-th area; is the flood energy consumption coefficient; is the peak flood discharge; is the total number of inundated areas.

[0051] Among them, the gate regulation objective function:

[0052] In the formula, is the gate regulation objective function; is the safety benefit coefficient; is the safety threshold of the i-th monitoring point; is the real-time water level of the i-th monitoring point; is the risk control coefficient; is the risk threshold of the m-th type; is the real-time risk value of the m-th type; is the regulation cost coefficient; is the change amount of the opening of the g-th gate.

[0053] According to the real-time simulation results of the digital twin model, generate a set of possible flood routing strategies, including different peak flood discharges, rainfall patterns, and propagation paths; Combine the flood routing strategy set with the gate regulation instruction set to obtain the risk and benefit indicators under each strategy combination, and construct a game matrix; the risk and benefit indicators include downstream flood control safety, inundation loss, ecological impact, and dam-break risk; It should be noted that based on the real-time simulation results of the digital twin model, a set of flood routing strategies is generated, including different peak flood discharges, rainfall patterns, and propagation paths. At the same time, a set of gate regulation instructions is extracted from the optimal trade-off set, including gate opening, opening and closing timing, and flood discharge. Combine the flood routing strategy set with the gate regulation instruction set in pairs to form strategy combination pairs, which serve as the rows and columns of the game matrix. For each strategy combination, simulate the interaction process of flood routing and gate regulation through the digital simulation model to obtain risk and benefit indicators such as downstream flood control safety, inundation loss, ecological impact, and dam-break risk. Then, fill the obtained index values into the game matrix to form a complete game matrix.

[0054] Extract the Nash equilibrium point of the game model, and determine whether there is a state where neither party can obtain a better result by unilaterally changing the strategy. If it exists, take this state as the optimal game solution; It should be noted that based on the constructed game matrix, all combinations of the flood evolution strategy set and the gate regulation instruction set are traversed, and the payoff values of both parties under each combination are calculated. For each flood evolution strategy, find the optimal response strategy in the gate regulation instruction set that maximizes the payoff, and record the corresponding payoff value. For each gate regulation instruction, find the optimal response strategy in the flood evolution strategy set that maximizes the payoff, and record the corresponding payoff value; then, compare the payoff values of both parties to determine whether there is a strategy combination such that neither party can obtain a higher payoff by unilaterally changing the strategy. If it exists, mark this strategy combination as the Nash equilibrium point; if there are multiple Nash equilibrium points, select the optimal solution according to the preset priority rules (such as flood control safety first, risk control first, etc.).

[0055] Based on the Nash equilibrium solution, quantitatively evaluate the risks and benefits of the current gate regulation strategy, including flood control safety benefits, reduced economic losses, ecological impact control degree, and risk reduction rate, and generate a comprehensive evaluation result; According to the comprehensive evaluation result, determine whether the current gate regulation strategy meets the preset goals. If not, regenerate the gate regulation instruction set.

[0056] It should be noted that extract the current gate regulation strategy and its corresponding flood evolution strategy from the Nash equilibrium solution as the basic data for evaluation. Based on the digital twin model, simulate the flood evolution process under this strategy combination, and obtain the actual values of indicators such as downstream flood control safety level, inundation loss, ecological impact, and dam-break risk; then, compare the actual values with the preset target values to obtain flood control safety benefits (such as the compliance rate of downstream water level control), reduced economic losses (such as the reduction in inundated area), ecological impact control degree (such as the protection rate of ecological sensitive areas), and risk reduction rate (such as the reduction amplitude of dam-break probability); then, use the weighted summation method to normalize each indicator to generate a comprehensive evaluation score, reflecting the overall effect of the strategy; further, determine whether the strategy meets the preset goals (including flood control safety goals, loss minimization goals, ecological impact control goals, risk control goals) according to the evaluation score. If not, adjust the strategy parameters or re-optimize; finally, generate a comprehensive evaluation report, including the quantitative values of each indicator, the comprehensive score, and optimization suggestions.

[0057] The present invention can improve the scientificity and balance of the regulation strategy of the flood detention area; dynamically optimize the gate regulation instruction set based on the comprehensive evaluation result, improve the real-time performance and operability of the regulation strategy, effectively reduce flood risks and losses, and provide scientific decision-making support for the management of the flood detention area.

[0058] In this embodiment, the risk-benefit dynamic game regulation method for flood detention areas may further include the following steps: Based on the digital twin model, obtain the real-time status data of each reservoir, gate and downstream river channel in the basin, including reservoir water storage, gate opening, river channel flow and carrying capacity, and construct a joint operation state space; Use a distributed optimization algorithm (such as the ADMM algorithm) to preliminarily allocate the flood discharge tasks, aiming to minimize the regional inundation loss and maximize flood control safety, and generate the recommended flood discharge values for each reservoir / gate; Simulate the downstream river channel state after the execution of the flood discharge task through the digital twin model, and judge whether there are local area resource conflicts (such as downstream river channel overloading or insufficient gate flood discharge capacity); If there are conflicts, enter the game model. Take each reservoir / gate as a game participant, define its benefit function as the weighted sum of the flood discharge task completion degree and downstream safety degree, and solve the optimal flood discharge volume allocation scheme through the Nash bargaining solution or Shapley value; Feed back the flood discharge volume allocation scheme generated by the game model to the digital twin model to verify its global resource optimality. If the verification passes, execute the scheme; otherwise, return to the game model to solve again; Convert the final flood discharge task allocation scheme into a regulation instruction set for each reservoir / gate, including flood discharge volume, opening degree and time sequence parameters, and verify the instruction execution effect through real-time monitoring data; Feed back the execution effect to the distributed optimization algorithm and the game model, dynamically adjust the parameters and weights, and ensure the continuous optimization and adaptability of the scheduling strategy.

[0059] It should be noted that through this method, the global resource optimal allocation of multi-reservoir / gate joint operation is realized, which has the characteristics of high efficiency, balance and operability.

[0060] In this embodiment, the risk-benefit dynamic game regulation method for flood detention areas may further include the following steps: Based on the digital twin model, obtain the vegetation type, soil organic matter content and hydrological condition data of the inundated area, and construct a carbon emission prediction model; Simulate the inundation range and duration under different regulation strategies, and calculate the carbon emissions generated by the decomposition of organic matter in the inundated area, including the emissions of carbon dioxide and methane; Combined with the carbon sink capacity of the regional ecological sensitive area (such as the carbon absorption of wetlands and forests), design a carbon sink compensation model, and calculate the net carbon emission value of the regulation strategy; Compare the net carbon emission value with the regional ecological carrying capacity threshold to judge whether it exceeds the standard. If it exceeds the standard, enter the low-carbon emission strategy optimization stage, and use a multi-objective optimization algorithm (such as NSGA-III) to regenerate the gate regulation instruction set, and give priority to strategies with a small inundation range and short duration; Input the optimized control instruction set into the digital twin model to verify its flood control effect and carbon emission value. If the dual objectives are met, it will be executed, otherwise it will return to the optimization stage. The execution results will be fed back to the carbon emission prediction model and carbon sink compensation model, and the parameters and weights will be adjusted dynamically to ensure the continuous optimization and adaptability of the strategy.

[0061] It should be noted that the above steps achieve the dual goals of flood control and carbon neutrality through carbon emission forecasting, carbon sink compensation, threshold judgment and strategy optimization, and are scientific, efficient and eco-friendly.

[0062] like Figure 3 As shown, the second aspect of the present invention discloses a risk-benefit dynamic game control system 6 for flood storage and detention areas based on digital twins, wherein the risk-benefit dynamic game control system for flood storage and detention areas comprises a memory 41 and a processor 52, wherein the memory 41 stores a risk-benefit dynamic game control method program for flood storage and detention areas, and when the risk-benefit dynamic game control method program for flood storage and detention areas is executed by the processor 52, any step of the risk-benefit dynamic game control method for flood storage and detention areas is implemented.

[0063] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A risk-benefit dynamic game control method for flood storage and detention areas based on digital twins, characterized in that: The following steps are involved: Build a digital twin model of the flood storage area based on multi-dimensional data sources, dynamically simulate real-time flood events based on the digital twin model, and obtain probability distribution data of flood evolution, inundation range and risk; Obtaining historical flood diversion control strategies of the flood storage and detention area, performing multi-objective iterative optimization and solving of the historical flood diversion control strategies based on the probability distribution data, and generating an optimal trade-off set of multiple objectives; Based on the digital twin model and the optimal trade-off set, a real-time controllable gate regulation instruction set is generated; In the process of real-time gate control through the gate control instruction set, the risks and benefits of flood storage and detention area control are quantitatively evaluated.

2. The risk-benefit dynamic game control method for flood storage and detention areas based on digital twins according to claim 1 is characterized in that: The digital twin model of the flood storage area is constructed based on multidimensional data sources, specifically: The sensor network deployed collects hydrological, meteorological, topographical and engineering facility data in real time, while integrating relevant data sets of historical flood events to form a multi-dimensional data source; Based on multi-dimensional data sources, spatial interpolation algorithms are used to reconstruct terrain data in a refined manner to generate digital elevation models, and combined with hydrological models to simulate the runoff and confluence processes of the basin; Dynamically calibrate the parameters of the hydrological model, and determine whether the model accuracy meets the preset accuracy requirements by comparing the real-time monitoring data with the model simulation results. If not, adjust the model parameters through an iterative optimization algorithm until the accuracy meets the requirements; The calibrated hydrological model is combined with the digital elevation model to construct a flood evolution simulation module, and the finite element analysis method is used to dynamically simulate and predict the flood inundation range and water depth; At the same time, ecological factor data is introduced to build an ecological impact assessment module to conduct quantitative analysis of the ecological risks of flood-inundated areas; By integrating hydrological models, digital elevation models, flood evolution simulation modules and ecological impact assessment modules, a multi-dimensional dynamic mapping digital twin model is formed.

3. The risk-benefit dynamic game control method for flood storage and detention areas based on digital twins according to claim 1 is characterized in that: Based on the digital twin model, real-time flood events are dynamically simulated to obtain probability distribution data of flood evolution, inundation range and risk, specifically: Input the real-time monitoring data into the digital twin model, start the flood evolution simulation module, use the finite element analysis method to simulate the dynamic propagation process of the flood in the flood storage area, and generate real-time flood water level, flow velocity and flow direction data; Compare and analyze the simulation results with the data of historical flood events to determine the similarity between the current flood event and the historical event. If the similarity reaches the preset similarity threshold, the probability distribution data of the historical flood event is directly called, otherwise proceed to the next step; Monte Carlo simulation method is used to combine the uncertainty characteristics of historical flood data to generate a large number of random flood scenarios, and the digital twin model is used to simulate each scenario to calculate the flood evolution path, inundation range and water depth distribution; Extract the simulated flood evolution path, inundation range and water depth distribution as the input sample data for kernel density estimation; use the kernel density estimation algorithm to calculate the probability density of the sample data and generate the probability density function of flood evolution, inundation range and risk; By integrating the probability density function, we can obtain the probability distribution curve of flood evolution, inundation range and risk. Through visualization technology, we can transform the probability distribution curve into an intuitive distribution map, and combine it with the statistical characteristic value to obtain the probability distribution data.

4. The risk-benefit dynamic game control method for flood storage and detention areas based on digital twins according to claim 1 is characterized in that: Based on the probability distribution data, the multi-objective iterative optimization solution of the historical flood diversion control strategy is performed to generate the optimal trade-off set of multiple objectives, specifically: Based on the probability distribution data of flood evolution, inundation range and risk, the Monte Carlo sampling method is used to generate an initial solution set. Each solution represents a historical flood diversion and regulation strategy, and ensures that the solution set covers the multi-objective parameter space of flood control safety, inundation loss, ecological impact and risk control. The flood control safety target is defined as the downstream water level control threshold, the flooding loss target is the correlation function between the economic value model and the flooded area, the ecological impact target is the flooding index of the ecologically sensitive area, and the risk control target is the potential dam failure probability and risk propagation model. The four are transformed into quantifiable objective functions. According to the value range of the objective function, a reference point set is generated by using a uniform distribution or an adaptive density method as a diversity distribution in the multi-objective parameter space; a reference point is randomly selected from the reference point set as an associated reference point; Based on the associated reference points, the initial solution set is non-dominatedly sorted, the frontier level of the initial solution set is divided, and the initial solution set is screened by the frontier level to obtain a screened solution set; whether the screened solution set meets the diversity requirement is determined, and if not, a new associated reference point is selected in the reference point set to re-screen the initial solution set; The filtered solution set is simulated with binary crossover and polynomial mutation operations to generate a new solution set and calculate its multi-objective fitness value. The parent generation and the child generation are merged with the elite retention strategy to iteratively update the solution set. The preset hypervolume index change rate is set as the termination condition. If the hypervolume index change rate of the solution set is lower than the preset hypervolume index change rate, the optimization is terminated and the current optimal trade-off set is output.

5. The risk-benefit dynamic game control method for flood storage and detention areas based on digital twins according to claim 1 is characterized in that: Based on the digital twin model and the optimal trade-off set, a real-time controllable gate regulation instruction set is generated, specifically: Based on the digital twin model, the water flow of real-time flood events is simulated to extract the flood water level, flow velocity and inundation range. The state space is constructed by combining the multi-objective strategy parameters in the optimal trade-off set to characterize the global characteristics of the current flood scene. The gate opening, opening and closing sequence and flood discharge are used as the core variables of the state space to define a discrete or continuous state set; Pre-training the state space based on a deep neural network and mapping it to the digital twin model; In the flood environment simulated by the digital twin model, gate control actions are executed and state transitions and reward values ​​are observed, and the interaction data is stored in the experience replay pool; Sample data from the experience replay pool, update the policy network and value network, generate the optimal gate control action under the current state, and convert the gate control action into an executable gate control instruction set.

6. The risk-benefit dynamic game control method for flood storage and detention areas based on digital twins according to claim 1 is characterized in that: In the process of real-time gate control through the gate control instruction set, the risks and benefits of flood storage and detention area control are quantitatively evaluated, specifically: The flood evolution process and the gate control strategy are defined as the two sides of the game, with the flood evolution as the "opponent" and its uncertainty simulated through the digital twin model, and the gate control as the "decision maker"; Based on the non-cooperative game framework in game theory, a game model of flood evolution and gate regulation is constructed, and the objective functions of both parties are defined. The goal of flood evolution is to maximize the inundation loss, and the goal of gate regulation is to minimize the risk and loss. Generate a set of possible strategies for flood evolution based on the real-time simulation results of the digital twin model; Combine the flood evolution strategy set with the gate control instruction set to obtain the risk and benefit indicators under each strategy combination and construct a game matrix; Extract the Nash equilibrium point of the game model and determine whether there is a state where neither party can obtain a better result by unilaterally changing their strategies. If so, take this state as the optimal game solution. Based on the Nash equilibrium solution, the risks and benefits of the current gate control strategy are quantitatively evaluated to generate comprehensive evaluation results; Based on the comprehensive evaluation results, determine whether the current gate control strategy meets the preset goals. If not, regenerate the gate control instruction set.

7. The risk-benefit dynamic game control method for flood storage and detention areas based on digital twins according to claim 1 is characterized by: The multiple objectives include flood control safety, minimization of flooding losses, ecological impact and potential risk control.

8. The risk-benefit dynamic game control system of flood storage and detention areas based on digital twins is characterized by: The flood storage and detention area risk-benefit dynamic game control system includes a memory and a processor. The memory stores a flood storage and detention area risk-benefit dynamic game control method program. When the flood storage and detention area risk-benefit dynamic game control method program is executed by the processor, the steps of the flood storage and detention area risk-benefit dynamic game control method as described in any one of claims 1 to 7 are implemented.

Images