Water conservancy project intelligent investigation design and simulation method and system based on multi-source data fusion

Abstract

The invention relates to the technical field of water conservancy project investigation design, and discloses a water conservancy project intelligent investigation design and simulation method and system based on multi-source data fusion, and the method comprises the steps: obtaining a basin reservoir group basic data set; generating a geological parameter field set by using a sequential Gaussian simulation method, and configuring a multi-agent system; training the dam body mechanical response agent model set, and generating a candidate robust scheme set; calculating a risk conduction sensitivity index in the upstream and downstream geological parameter field combined space; generating a coordination reinforcement scheme through an iterative negotiation mechanism; and outputting the integral robustness evaluation result of the watershed by using combined Monte Carlo simulation. According to the method, the technical problems of geological uncertainty and risk conduction assessment distortion in a basin reservoir group reinforcement decision are solved.

Description

Technical Field

[0001] This invention relates to the field of water conservancy engineering survey and design technology, and more specifically, to a method and system for intelligent survey, design and simulation of water conservancy engineering based on multi-source data fusion. Background Technology

[0002] In watershed reservoir reinforcement projects, multiple reservoirs are typically operated by different management entities, each independently selecting reinforcement schemes based on its own interests and safety objectives. Due to limitations in the number and density of exploration boreholes, geological parameters in the areas between boreholes at different reservoirs rely on spatial interpolation estimation, resulting in significant spatial uncertainty. Simultaneously, the reinforcement effect of upstream reservoirs directly impacts the flood risk boundary conditions faced by downstream reservoirs, and the risk status of downstream reservoirs may trigger coordination requirements for upstream reinforcement standards, creating an interactive decision-making relationship.

[0003] Existing technologies typically involve each management entity independently optimizing reinforcement schemes based on deterministic geological parameter fields and conducting upstream and downstream chain impact assessments according to fixed rules.

[0004] This approach has the following technical problems: When each management entity optimizes independently, it fails to consider the impact of its own geological parameter spatial uncertainty on the reinforcement effect, nor does it consider the superposition effect of uncertainty in the risk transmission relationship between upstream and downstream reservoirs; when the upstream reservoir chooses a reinforcement scheme under its own geological uncertainty, the risk transmission effect of the reinforcement scheme on the downstream may be significantly worse under certain geological parameters, leading to a distortion of the overall risk assessment of the basin and an underestimation of the reinforcement needs of downstream reservoirs; at the same time, ignoring the decision-making interaction between management entities leads to a systematic deviation between the assessment results and the actual coordination behavior. Summary of the Invention

[0005] This invention provides a method and system for intelligent surveying, design and simulation of water conservancy projects based on multi-source data fusion, which solves the technical problems in related technologies such as failure to effectively handle the uncertainty of geological parameters, lack of multi-subject coordination mechanism at the watershed level, and difficulty in achieving overall robust equilibrium of the watershed.

[0006] This invention provides an intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion, comprising: Geological exploration borehole data and spatial coordinates of each reservoir in the basin, upstream and downstream topological relationships between reservoirs, reinforcement objective function parameters of each management entity, and description of information sharing mechanisms among management entities are obtained to generate a basic dataset of the basin reservoir group. For each reservoir, the sequential Gaussian simulation method is used to generate multiple equally probable geological parameter fields based on geological exploration borehole data to form a set of geological parameter fields for each reservoir. An agent decision-making model is configured for each reservoir and the interaction protocol between the agent decision-making models is defined to generate a multi-agent system configuration with geological uncertainty description. Each reservoir's intelligent agent decision-making model trains a dam mechanical response proxy model for each sample in its own geological parameter field set, generating a dam mechanical response proxy model set. The dam mechanical response proxy model set is then used to perform batch predictions on candidate schemes and calculate local robustness indices, generating a candidate robust scheme set for each reservoir. The optimal solution is passed on according to the upstream and downstream topological order. The downstream intelligent agent decision model calculates the risk transmission sensitivity index in the joint space of the upstream geological parameter field set and its own geological parameter field set. The downstream intelligent agent decision model generates a coordination request based on the risk transmission sensitivity index and sends it back to the upstream intelligent agent decision model. The upstream intelligent agent decision model incorporates the coordination constraints into the objective function and then re-selects alternative solutions, iteratively negotiates until the overall robust equilibrium of the watershed is achieved, and generates a coordinated reinforcement solution. Using joint Monte Carlo simulation, a risk distribution assessment of the watershed system is conducted within the combined space of geological parameter fields of each reservoir, and a coordinated reinforcement scheme for each reservoir and an overall robustness assessment report of the watershed are output.

[0007] Furthermore, the execution process of the sequential Gaussian simulation method includes: Normal transformation is performed on borehole measured data to establish a spatial correlation structure of geological parameters using spatial variability functions. Define the three-dimensional mesh nodes to be simulated, visit each three-dimensional mesh node in sequence according to the preset path, and for each node to be simulated, perform Kriging estimation using known borehole data and simulated node data to obtain the conditional mean and conditional variance. Randomly select a value from the corresponding conditional Gaussian distribution as the simulated value of the node to be simulated. After traversing all three-dimensional mesh nodes, a complete geological parameter field sample is generated. The above simulation process is repeated a preset number of times, each time using a different random number seed and node access path to generate multiple geological parameter field samples with equal probability.

[0008] Furthermore, the training process of the dam mechanical response surrogate model includes: For each geological parameter field realization sample in the geological parameter field set, the dam safety factor corresponding to multiple sets of reinforcement scheme parameters is calculated using the finite element method to generate a training sample set. The training sample set is input into the neural network for training, and the corresponding geological parameter field is generated to realize the dam mechanical response proxy model of the sample. The above process is repeated for all samples in the geological parameter field set to generate a set of dam mechanical response proxy models.

[0009] Furthermore, the execution process of the finite element method includes: The dam body and foundation are divided into a finite number of units. Material parameters are assigned to each unit based on the geological parameter field. Boundary conditions and loading conditions are set according to the reinforcement scheme parameters. A set of mechanical equilibrium equations is established and solved to obtain the stress-strain field. The reduction coefficient when the dam body reaches the limit equilibrium state is calculated iteratively using the strength reduction method as the dam body safety factor.

[0010] Furthermore, the local robustness index includes the mean safety factor, the minimum safety factor, and the probability of falling below a threshold. The average safety factor is the arithmetic mean of the safety factors of the candidate scheme under all geological parameter field implementation samples; The minimum safety factor is the minimum safety factor of the candidate scheme under all geological parameter field implementation samples; The probability below the threshold is the ratio of the number of samples in which the safety factor of the candidate scheme is lower than the safety factor threshold under all geological parameter field implementation samples to the total number of samples.

[0011] Furthermore, the calculation process of the risk transmission sensitivity index includes: The upstream intelligent agent decision model transmits the flood discharge capacity parameters corresponding to the optimal solution to the downstream intelligent agent decision model. The flood discharge capacity parameters are calculated by the hydrological and hydraulic model based on the upstream reinforcement scheme parameters and the upstream geological parameter field. The downstream intelligent agent decision model injects the flood discharge capacity parameters into its own dam mechanical response proxy model set, traverses the joint space of the upstream geological parameter field set and its own geological parameter field set, and calculates the downstream safety factor under each joint scenario; The ratio of the number of scenarios with downstream security coefficients lower than the downstream security coefficient threshold to the total number of scenarios in all joint scenarios is used as a risk transmission sensitivity indicator.

[0012] Furthermore, the iterative negotiation process includes: The downstream intelligent agent decision model determines whether the risk transmission sensitivity index exceeds the acceptable threshold. If it does, it generates a coordination request containing the expected flood discharge capacity constraint range and sends it back to the upstream intelligent agent decision model. The upstream agent decision-making model searches for a solution that meets the flood discharge capacity constraint in the candidate robust solution set. If such a solution exists, the one with the best local robustness index is selected as the alternative solution. If such a solution does not exist, a new candidate solution is generated in the constrained parameter space and the best solution is selected as the alternative solution. The alternative solution is passed to the downstream intelligent agent decision model to recalculate the risk transmission sensitivity index. If it still exceeds the acceptable threshold, the flood discharge capacity constraint range is adjusted and the coordination request is sent again until the overall robust equilibrium of the basin is achieved or the upper limit of the negotiation rounds is exceeded.

[0013] Furthermore, the adjustment strategy for the flood discharge capacity constraint range is as follows: if the upstream alternative still causes the risk transmission sensitivity index to exceed the acceptable threshold, the downstream intelligent agent decision model will further reduce the upper limit of the flood discharge capacity constraint, and the reduction range will be dynamically determined based on the degree of exceeding the limit and the remaining negotiation rounds.

[0014] Furthermore, the execution process of the joint Monte Carlo simulation includes: A geological parameter field is extracted from the geological parameter field set of each reservoir to form a joint scenario; For the joint scenario, the flood discharge capacity parameters and dam safety factor of each reservoir are calculated sequentially according to the upstream and downstream topology, and the safety status of each reservoir and whether the watershed system fails are recorded under the joint scenario. Repeat the above sampling and calculation process until all combinations in the joint space are traversed or the preset number of samplings is reached, and statistically analyze the joint distribution characteristics of the failure probability of the watershed system and the safety factor of each reservoir. Among them, the failure of a watershed system is defined as the state in which the safety factor of at least one reservoir is lower than its corresponding safety factor threshold.

[0015] This invention provides an intelligent survey, design, and simulation system for water conservancy projects based on multi-source data fusion, comprising: The data acquisition module is used to acquire geological exploration borehole data and spatial coordinates of each reservoir in the basin, the upstream and downstream topological relationships between reservoirs, the reinforcement objective function parameters of each management entity, and the description of the information sharing mechanism between management entities, and generate a basic dataset of the basin reservoir group. The multi-agent configuration module is used to generate a set of geological parameter fields for each reservoir using the sequential Gaussian simulation method, configure an agent decision model for each reservoir and define the interaction protocol, and generate a multi-agent system configuration with geological uncertainty description. The robust solution generation module is used to train a set of dam mechanical response proxy models for the geological parameter field set of each reservoir, calculate local robustness indices, and generate a set of candidate robust solutions for each reservoir. The risk transmission assessment module is used to transmit the optimal solution in the order of upstream and downstream topology and calculate the risk transmission sensitivity index in the joint space of the upstream and downstream geological parameter field sets. The iterative negotiation module is used to generate coordination requests based on risk transmission sensitivity indicators, incorporate coordination constraints into the upstream objective function, re-select alternative solutions, iterate and negotiate until the overall robust equilibrium of the basin is achieved, and generate coordinated reinforcement solutions. The risk assessment output module is used to assess the risk distribution of the watershed system using joint Monte Carlo simulation, and outputs coordinated reinforcement schemes for each reservoir and an overall robustness assessment report for the watershed.

[0016] The beneficial effects of this invention are as follows: This invention generates a set of geological parameter fields using a sequential Gaussian simulation method to explicitly characterize spatial uncertainty. It achieves batch assessment under geological condition assumptions through a set of dam mechanical response proxy models. It quantifies the combined effect of uncertainty by calculating risk transmission sensitivity indicators within the joint space of upstream and downstream geological parameter fields. It simulates decision-making interactions between management entities through an iterative negotiation mechanism. Finally, it assesses the risk distribution of the watershed system through joint Monte Carlo simulation. This invention addresses the technical problems of neglecting the superposition effect of uncertainty in the spatial uncertainty of geological parameters and the relationship between upstream and downstream risk transmission in watershed reservoir group reinforcement decisions, distorting the overall risk assessment of the watershed, underestimating the reinforcement needs of downstream reservoirs, and causing systematic deviations between assessment results and actual coordination behavior. The invention achieves the technical effects of reinforcement schemes possessing acceptable safety performance under various geological conditions, avoiding the omission of extreme scenarios, reducing systematic deviations between assessment results and actual coordination behavior, and ensuring robust performance of coordinated reinforcement schemes at the watershed level. Attached Figure Description

[0017] Figure 1 This is a flowchart of the intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion, as proposed in this invention. Detailed Implementation

[0018] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed only to enable those skilled in the art to better understand and implement the subject matter described herein, and changes may be made to the function and arrangement of the elements discussed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the examples. Furthermore, some features described in the examples may be combined in other examples.

[0019] At least one embodiment of the present invention discloses an intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion, such as... Figure 1 As shown, it includes the following steps: Step 1: Obtain the basic dataset of the watershed reservoir group.

[0020] Geological exploration borehole data and spatial coordinates of each reservoir in the basin, upstream and downstream topological relationships between reservoirs, reinforcement objective function parameters of each management entity, and description of information sharing mechanisms among management entities are obtained to generate a basic dataset of the basin reservoir group.

[0021] Furthermore, geological exploration borehole data includes measured values ​​of geological parameters such as permeability coefficient and shear strength, along with their corresponding three-dimensional spatial coordinates.

[0022] Furthermore, the parameters of the objective function are reinforced to include flood control safety weights, cost constraints, downstream impact constraints, and other parameters that reflect the decision-making preferences of various management entities.

[0023] Furthermore, the description of the information sharing mechanism among management entities includes the types of information that can be exchanged between the management entities, the timing of information transmission, and confidentiality constraints.

[0024] Step 2: Generate a multi-agent system configuration with geological uncertainty description.

[0025] For each reservoir in the basic dataset of the watershed reservoir group, the sequential Gaussian simulation method is used to generate multiple equally probable geological parameter fields based on geological exploration borehole data to form a set of geological parameter fields for each reservoir. At the same time, an intelligent agent decision-making model reflecting the decision-making logic of its management entity is configured for each reservoir, the interaction protocol between intelligent agent decision-making models is defined, and a multi-agent system configuration with geological uncertainty description and interaction protocol encoding is generated.

[0026] Furthermore, the input to the sequential Gaussian simulation method is the borehole data point set. ,in For the first Spatial coordinates of a borehole These are the corresponding measured values ​​of geological parameters. For the number of holes, The output is a set of geological parameter fields, which serves as the borehole index. ,Include There are 1 equally probable realized samples, among which To achieve the sample size for the geological parameter field.

[0027] Furthermore, the execution process of the sequential Gaussian simulation method includes the following sub-steps: First, the borehole measured data is normally transformed to make the data conform to a standard normal distribution; then, a spatial variogram is established to describe the spatial correlation structure of geological parameters; next, the three-dimensional grid nodes to be simulated are defined, and each three-dimensional grid node is visited sequentially according to a preset path; for each node to be simulated, the conditional mean and conditional variance are obtained by Kriging estimation using known borehole data and simulated node data, and a value is randomly selected from the corresponding conditional Gaussian distribution as the simulated value of the node to be simulated; after traversing all three-dimensional grid nodes, a complete geological parameter field is generated to realize sample generation; the above simulation process is repeated. Each time, a different random number seed and node access path are used to generate... A set of geological parameter fields is formed by realizing samples with equal probability.

[0028] Furthermore, spatial variability functions describe the correlation decay of geological parameters at different spatial distances. Commonly used forms include spherical models, exponential models, and Gaussian models. By fitting the spatial variability characteristics of borehole data, parameters such as range, sill value, and nugget value can be determined.

[0029] Furthermore, the agent decision-making model refers to the decision-making logic unit configured for each reservoir. The input of the agent decision-making model is the geological parameter field, reinforcement scheme parameters, and interactive information from other agent decision-making models. The output is the reinforcement scheme selection and the interactive information sent to other agent decision-making models.

[0030] Furthermore, the interaction protocol between intelligent agent decision-making models includes rules for transmitting reinforcement scheme information from upstream to downstream, triggering conditions for downstream to send coordination requests to upstream, and response mechanisms.

[0031] Furthermore, the definition of the interaction protocol between intelligent agent decision-making models includes the following components: message type definition, including scheme notification messages and coordination request messages; message content specification, where scheme notification messages contain reinforcement scheme parameters and corresponding flood discharge capacity parameters, and coordination request messages contain the expected constraint range; transmission rules, where scheme notification messages are transmitted from upstream to downstream in the order of upstream and downstream topology, and coordination request messages are transmitted back from downstream to upstream; and response mechanism, where the upstream intelligent agent decision-making model receiving the coordination request incorporates the constraints into its own optimization objective to adjust the reinforcement scheme.

[0032] Step 3: Generate a set of candidate robust solutions for each reservoir.

[0033] Each reservoir's intelligent agent decision-making model trains a dam mechanical response proxy model for each sample in its own geological parameter field set, generating a dam mechanical response proxy model set. A genetic algorithm is used to generate a candidate scheme population in the reinforcement scheme parameter space. For each candidate scheme, the dam mechanical response proxy model set is called for batch prediction to obtain the safety factor distribution of the candidate scheme under all geological condition assumptions. The local robustness index of each candidate scheme is calculated to generate a candidate robust scheme set for each reservoir.

[0034] Furthermore, the dam mechanical response surrogate model refers to a machine learning model trained based on finite element analysis results. The inputs to the dam mechanical response surrogate model are reinforcement scheme parameters and a geological parameter field, and the output is the dam safety factor. For the geological parameter field set... The first in Each geological parameter field realizes a sample Train the corresponding dam mechanical response surrogate model This forms a set of surrogate models for the mechanical response of the dam. ,in To implement sample indexing for geological parameter fields, To achieve the sample size for the geological parameter field.

[0035] Furthermore, the training process of the dam mechanical response surrogate model includes the following sub-steps: Step 301: For each geological parameter field sample in the geological parameter field set, calculate the dam safety factor corresponding to multiple sets of reinforcement scheme parameters using the finite element method, and generate a training sample set.

[0036] Furthermore, the execution process of the finite element method includes: dividing the dam body and foundation into a finite number of elements; assigning material parameters to each element based on the geological parameter field sample; setting boundary conditions and loading conditions based on the reinforcement scheme parameters; establishing a set of mechanical equilibrium equations and solving them to obtain the stress-strain field; and using the strength reduction method to iteratively calculate the reduction factor when the dam body reaches the limit equilibrium state as the dam body safety factor.

[0037] Step 302: Input the training sample set into the neural network for training, and generate the corresponding geological parameter field to realize the dam mechanical response proxy model of the sample.

[0038] Furthermore, the input layer of the neural network receives a parameter vector of the reinforcement scheme, the hidden layer contains multiple fully connected layers, and the output layer is a single numerical value output by a fully connected layer representing the predicted safety factor of the dam. The training process employs a supervised learning model, and the loss function is the mean square error between the predicted value and the finite element calculation value.

[0039] Step 303: Repeat steps 301 to 302 for all geological parameter fields in the geological parameter field set to generate a set of dam mechanical response proxy models.

[0040] Furthermore, local robustness indicators include the average safety factor. Minimum safety factor Probability below the threshold For candidate solutions The safety factor distribution of the candidate solutions is as follows: ,in To implement sample indexing for geological parameter fields, To achieve the sample size for the geological parameter field, then:

[0041] in, For the safety factor threshold, This is an indicator function that takes the value 1 when the condition inside the parentheses is met, and 0 otherwise.

[0042] Furthermore, the safety factor threshold According to the technical specifications for the safety of reservoir dams, the value is usually set between 1.3 and 1.5, and the specific value is determined according to the reservoir level and the type of working conditions.

[0043] Furthermore, the genetic algorithm takes as input the range of values ​​for the reinforcement scheme parameters and the definition of the fitness function, and outputs a population of candidate schemes. The fitness function is defined based on a local robustness index, and candidate schemes that meet the preset conditions for the local robustness index are selected to enter the candidate robust scheme set.

[0044] Furthermore, the fitness function takes the form of a weighted combination that comprehensively considers the mean safety factor, the minimum safety factor, and the probability of falling below the threshold. The weighting coefficients are set according to the risk preferences of the management entity, and candidate solutions with higher minimum safety factors and lower probabilities of falling below the threshold are given priority.

[0045] Step 4: Calculate the risk transmission sensitivity index.

[0046] Following the upstream and downstream topological order, the upstream reservoir intelligent agent decision-making model transmits the preferred solution from the candidate robust solution set to the downstream intelligent agent decision-making model through the interaction protocol between intelligent agent decision-making models. The downstream intelligent agent decision-making model injects the flood discharge capacity parameters corresponding to the preferred solution into its own dam mechanical response proxy model set, evaluates the impact distribution of the preferred solution on its own risk boundary conditions in the joint space of the upstream geological parameter field set and its own geological parameter field set, and calculates the risk transmission sensitivity index.

[0047] Furthermore, the preferred solution refers to the solution with the best local robustness index in the candidate robust solution set, or the solution that satisfies the management subject's preference constraints.

[0048] Furthermore, flood discharge capacity parameters The flood discharge capacity parameter is calculated using a hydrological and hydraulic model based on upstream reinforcement scheme parameters and upstream geological parameter fields. This parameter characterizes the flood discharge flow of the upstream reservoir under specific reinforcement schemes and geological conditions. For the upstream reservoir in the first Each geological parameter field realizes the flood discharge capacity parameters under the sample. For the parameters of the upstream reinforcement scheme, To implement sample indexing for upstream geological parameter fields.

[0049] Furthermore, the calculation process of the hydrological and hydraulic model includes: determining the spillway size and discharge capacity based on the reinforcement scheme parameters; determining the dam seepage characteristics and stability constraints by combining geological parameter fields with samples; performing reservoir flood regulation calculations based on the design flood hydrograph; calculating the inflow, outflow and reservoir water level changes at different times; and extracting the maximum discharge flow as the discharge capacity parameter.

[0050] Furthermore, the design flood hydrograph is obtained from the statistical data of the basin hydrology, describing the variation of flood flow over time under a specific return period, and the flood regulation calculation is based on the water balance equation and the discharge capacity equation for iterative calculation.

[0051] Furthermore, the risk transmission sensitivity index is used to quantify the impact of upstream reinforcement schemes on downstream risk boundary conditions under the combined effects of geological uncertainties. Let the geological parameter field set of the upstream reservoir be... The geological parameter field set of the downstream reservoir is as follows The upstream preferred solution is Then the downstream safety factor distribution within the joint space is as follows: ,in For the downstream reservoir at the first A proxy model of the dam's mechanical response under a sample is realized using a geological parameter field. The current downstream solution, For joint scenarios The downstream safety factor is below To implement sample indexing for upstream geological parameter fields, To implement sample indexing for downstream geological parameter fields, For the upstream reservoir Each geological parameter field realizes a sample. For the downstream reservoir Each geological parameter field realizes a sample To achieve the required number of samples for the upstream geological parameter field, To determine the sample size for downstream geological parameter fields. The risk transmission sensitivity index is defined as:

[0052] in, The safety factor threshold for downstream reservoirs, This is an indicator function.

[0053] Furthermore, the process includes the following steps: calculating the sensitivity of the downstream safety factor to the downstream geological parameters when the upstream geological parameters are fixed, the sensitivity of the downstream safety factor to the upstream geological parameters when the downstream geological parameters are fixed, and the sensitivity contribution of the interaction between the upstream and downstream geological parameters, thereby generating a risk transmission sensitivity decomposition result.

[0054] Step 5: Iteratively negotiate to generate a coordinated reinforcement plan.

[0055] If the downstream intelligent agent decision model identifies that the risk transmission sensitivity index exceeds the acceptable threshold, it generates a coordination request and sends it back to the upstream intelligent agent decision model through the interaction protocol between intelligent agent decision models. The coordination request includes the expected upstream flood discharge capacity constraint range. The upstream intelligent agent decision model incorporates the coordination constraint into its own objective function, and re-selects or generates alternative solutions that meet the downstream coordination constraint based on the dam mechanical response proxy model set in the candidate robust solution set. The alternative solutions are then sent back to the downstream intelligent agent decision model for evaluation. Iterative negotiation continues until the overall robustness equilibrium of the basin is achieved or the upper limit of the negotiation round is exceeded, and a coordinated reinforcement solution is generated.

[0056] Furthermore, the acceptable threshold refers to the upper limit of risk transmission sensitivity set by the downstream management entity based on its own risk tolerance. It is usually determined according to the safety level and flood control standards of the downstream reservoir, and the value ranges from 0.05 to 0.15.

[0057] Furthermore, the format of a coordination request includes a request identifier, an initiator identifier, a recipient identifier, expected constraints, and priority.

[0058] Furthermore, the upstream agent decision-making model incorporates coordination constraints into the objective function by adding constraint terms to the original objective function. ,in The flood discharge capacity constraint range specified in the downstream coordination request. To constrain the lower limit of flood discharge capacity, The upper limit of flood discharge capacity is constrained. For upstream solutions The corresponding flood discharge capacity parameters. Under the premise of satisfying coordination constraints, the upstream agent decision-making model re-selects the scheme with the best local robustness index from the candidate robust scheme set as the alternative.

[0059] Furthermore, the overall robustness equilibrium of the watershed refers to the state in which the risk transmission sensitivity index of all downstream agent decision-making models does not exceed their respective acceptable thresholds, and the alternative solutions of all upstream agent decision-making models meet their own local robustness requirements.

[0060] Furthermore, the upper limit of the negotiation rounds is a preset threshold for the number of iterations, used to prevent the negotiation process from looping indefinitely, and is usually set to 5 to 10 rounds.

[0061] Furthermore, based on step 5, the following steps are also included: Step 501: After receiving the coordination request, the upstream agent decision model first searches the candidate robust solution set for existing solutions that meet the flood discharge capacity constraint. If there is a solution that meets the flood discharge capacity constraint, the one with the best local robustness index is directly selected as the alternative solution.

[0062] Step 502: If there is no solution in the candidate robust solution set that meets the flood discharge capacity constraint, use a genetic algorithm to regenerate the candidate solution population in the constrained parameter space, call the dam mechanical response proxy model set to calculate the local robustness index, and select the optimal solution as the alternative solution.

[0063] Furthermore, the process of regenerating the candidate scheme population within the constrained parameter space includes: converting the flood discharge capacity constraint into the constraint conditions of the reinforcement scheme parameter space; generating only individuals that meet the constraint conditions during the population initialization phase of the genetic algorithm; performing constraint checks on the newly generated individuals after crossover and mutation operations; eliminating individuals that do not meet the constraint conditions and regenerating them; and ensuring that all candidate schemes meet the downstream coordination constraints throughout the entire evolution process.

[0064] Step 503: The alternative solution is passed to the downstream intelligent agent decision model. The downstream intelligent agent decision model recalculates the risk transmission sensitivity index. If the risk transmission sensitivity index still exceeds the acceptable threshold, the flood discharge capacity constraint range is adjusted and the coordination request is sent again until the overall robust equilibrium of the basin is achieved or the upper limit of the negotiation rounds is exceeded.

[0065] Furthermore, the strategy for adjusting the flood discharge capacity constraint range is as follows: if upstream alternative solutions still lead to excessive risk transmission sensitivity, the downstream intelligent agent decision-making model will adjust the upper limit of the flood discharge capacity constraint. Further reductions will be made, with the reduction amount dynamically determined based on the degree of exceeding the limit and the remaining negotiation rounds, to ensure convergence to an acceptable state within a limited number of rounds.

[0066] Step 6: Output the watershed coordinated reinforcement plan and risk assessment results.

[0067] The final reinforcement schemes for each reservoir under the overall robust equilibrium state of the watershed are collected. The risk distribution of the watershed system is assessed by using joint Monte Carlo simulation in the combined space of the geological parameter fields of each reservoir. The results output coordinated reinforcement schemes for each reservoir, an overall robust assessment report of the watershed, and recommendations for the division of risk transmission responsibilities among management entities.

[0068] Furthermore, the input to the joint Monte Carlo simulation is the Cartesian product space of the set of geological parameter fields of all reservoirs, and the output is the risk distribution characteristics of the watershed system. Assume that there are a total of [missing information - likely related to watershed geological parameters]. The geological parameter field sets for each of the reservoirs are as follows: ,in For the first Geological parameter field set of the reservoir, For reservoir index and Then the joint space is For each sample combination in the joint space ,in For the first The first reservoir Each geological parameter field realizes a sample. For the first The geological parameter field of the reservoirs was sampled and indexed to calculate the safety factor of each reservoir under the final reinforcement scheme and to statistically analyze the overall risk distribution characteristics of the watershed.

[0069] Furthermore, the execution process of the joint Monte Carlo simulation includes: extracting a geological parameter field from the geological parameter field set of each reservoir to form a joint scenario; calculating the flood discharge capacity parameters and dam safety factors of each reservoir in the order of upstream and downstream topology for the joint scenario; recording the safety status of each reservoir and whether the watershed system has failed under the joint scenario; repeating the above sampling and calculation process until all combinations in the joint space are traversed or the preset number of samplings is reached; and statistically analyzing the watershed system failure probability, the joint distribution characteristics of the safety factors of each reservoir, and the risk transmission path.

[0070] Furthermore, the failure of a watershed system is defined as the state in which the safety factor of at least one reservoir is lower than its corresponding safety factor threshold, and the failure probability of a watershed system is the proportion of the number of failure scenarios in all sampled scenarios to the total number of samples.

[0071] Furthermore, the overall robustness assessment report for the watershed includes the marginal distribution of safety factors for each reservoir, the probability of failure of the watershed system, the results of risk transmission path identification, and the analysis of key sources of uncertainty.

[0072] Furthermore, the proposed division of responsibility for risk transmission among management entities should be based on the risk transmission sensitivity decomposition results, quantifying the contribution ratio of each reservoir to downstream risks.

[0073] This implementation method generates a set of geological parameter fields for each reservoir using a sequential Gaussian simulation method. The spatial uncertainty of the exploration data of each reservoir is explicitly represented in the form of multiple equally probable geological parameter field implementation samples. Therefore, when optimizing reinforcement schemes, the intelligent agent decision-making model of each reservoir can make decisions based on the complete probability distribution of geological parameters rather than a single definite value.

[0074] This implementation method trains dam mechanical response proxy models for samples from different geological parameter fields and forms a set of dam mechanical response proxy models. This allows the performance evaluation of candidate reinforcement schemes to be carried out in batches under all geological condition assumptions. The local robustness index is calculated based on the statistical distribution of the safety factor. Therefore, the screening of reinforcement schemes at the single reservoir level can take into account the impact of the spatial uncertainty of geological parameters on the reinforcement effect, ensuring that the selected reinforcement scheme has acceptable safety performance under various possible geological conditions.

[0075] This implementation defines an interaction protocol between intelligent agent decision-making models and calculates a risk transmission sensitivity index in the joint space of the upstream and downstream geological parameter field sets. This enables the assessment of risk boundary conditions faced by downstream reservoirs to simultaneously consider the combined effects of upstream geological uncertainties and their own geological uncertainties, thus avoiding extreme scenarios that may be missed when assessing risk transmission based on a single deterministic upstream scheme.

[0076] This implementation uses an iterative negotiation mechanism to enable the downstream intelligent agent decision-making model to send a coordination request to the upstream intelligent agent decision-making model based on the risk transmission sensitivity assessment results. The upstream intelligent agent decision-making model adjusts the reinforcement scheme after incorporating the coordination constraints into its own optimization objectives. The iterative negotiation mechanism simulates the game behavior between management entities in actual reinforcement decision-making, thus reducing the systematic deviation between the assessment results and the actual coordination behavior.

[0077] This implementation method ultimately assesses the risk distribution of the watershed system by combining Monte Carlo simulation within the combined space of all reservoir geological parameter fields. This enables the overall risk assessment of the watershed to reflect the combined effects of geological uncertainties of multiple reservoirs. Therefore, the coordinated reinforcement scheme has robust performance at the watershed level, overcoming the problems of risk transmission assessment distortion caused by independent optimization by each management entity based on deterministic geological parameters and the underestimation of downstream reservoir reinforcement needs.

[0078] There are three cascade reservoirs in a certain river basin that are in poor condition and require reinforcement and safety improvement projects. These are located upstream of... Reservoirs, middle reaches Reservoir and downstream Reservoir. The reservoir is operated by the provincial water resources administration bureau, with flood control and safety as its primary objective. The reservoir is managed by the municipal water affairs group, taking into account both flood control and power generation benefits; The reservoir is managed by the county-level water conservancy station, which is mainly responsible for the protection of downstream towns.

[0079] In early January 20XX, geological exploration work was completed for the three reservoirs. Eight boreholes were drilled in the reservoir, and the measured permeability coefficient ranged from [missing value]. ; Six boreholes were drilled in the reservoir, and the measured permeability coefficient ranged from [missing value]. ; Seven boreholes were drilled in the reservoir, and the measured permeability coefficient ranged from [missing value]. Due to funding constraints in exploration, the borehole spacing exceeded 150 meters, and many regional geological parameters relied on spatial interpolation estimation, resulting in significant spatial uncertainty.

[0080] The management entities of the three reservoirs must submit reinforcement plans before March 20XX. The upstream water flow of the reservoir directly affects Peak inflow of the reservoir, The process of releasing water from the reservoir determines The reservoir faces risk boundary conditions. Each management entity needs to consider the robustness of reinforcement plans under the uncertainty of its own geological conditions, and also coordinate the risk transmission relationship between upstream and downstream areas to ensure the overall flood control safety of the basin.

[0081] against Data from eight boreholes obtained from the reservoir were used to extract the measured permeability coefficients and their three-dimensional spatial coordinates, while also recording... The parameters of the reinforcement objective function for the reservoir management entity are as follows: flood control safety weight 0.75, cost constraint upper limit 8.5 million yuan, and downstream impact constraint of flood discharge flow not exceeding 1200 cubic meters / s. and The same operation is performed on the reservoirs, and the data is aggregated to form the basic dataset of the watershed reservoir group.

[0082] Table 1 Geological exploration borehole data for the reservoir:

[0083] Table 2. Decision-making parameter configuration for the management entity of the watershed reservoir group:

[0084] against Data from eight boreholes in the reservoir were used. First, the measured permeability coefficients were normalized to establish a spherical spatial variogram (range 185 meters, sill value 0.82, nugget value 0.15). A three-dimensional grid of 1200 nodes was defined, and each node was accessed sequentially using random paths. Conditional mean and conditional variance were obtained using Kriging estimation, and simulated values ​​were extracted from a conditional Gaussian distribution. This process was repeated 50 times, each time using a different random number seed, to generate a set of geological parameter fields. At the same time, for Reservoir configuration intelligent agent decision-making model It is defined that it can receive coordination request messages from downstream and can send scheme notification messages to downstream.

[0085] Table 3 Reservoir geological parameter field generation configuration:

[0086] Table 4 Statistical characteristics of some geological parameters in the reservoir:

[0087] Statistical characteristics of the geological parameter field samples show that the spatial distribution of permeability coefficients differs under different random realizations, with the mean range being... The standard deviation reflects the degree of spatial variation within the field.

[0088] For geological parameter field ensemble Each geological parameter field in the model is sampled, and 120 sets of training samples (seepage wall depth range 12-28 meters, grouting curtain thickness range 1.5-3.5 meters) are generated in the reinforcement scheme parameter space using Latin hypercube sampling. For each set of parameters, the finite element method is called to calculate the dam safety factor. The training samples are then input into a three-layer fully connected neural network (input layer 2 nodes, hidden layer 32 nodes, output layer 1 node) to train and obtain a set of dam mechanical response surrogate models. .

[0089] A population of candidate schemes was generated using a genetic algorithm (population size 200, number of generations 100, crossover probability 0.8, mutation probability 0.15).

[0090] For candidate solutions = (Anti-seepage wall depth 22.5 meters, grouting curtain thickness 2.8 meters), using the dam body mechanical response proxy model set for batch prediction, the safety factor distribution under 50 geological condition assumptions was obtained. .

[0091] Table 5 Reservoir candidate schemes Safety factor distribution (partial):

[0092] Calculate the local robustness index:

[0093]

[0094] The probability of the candidate solution being below the threshold is 0.24, indicating that the safety factor does not meet the requirements in 12 scenarios out of 50 equally probable geological condition assumptions, and it is necessary to continue screening for candidate solutions with better robustness.

[0095] Table 6 Reservoir candidate robust scheme set (top 5 schemes preferred):

[0096] Select the solution with the lowest probability of falling below the threshold from the set of candidate robust solutions. As the preferred option.

[0097] The preferred solution Transmitted to downstream via intelligent agent interaction protocol .

[0098] Calculation using hydrological and hydraulic models exist The flood discharge capacity parameters of the reservoir under the sample of geological parameter fields in various locations. For The flood discharge capacity parameters were calculated. cubic meters per second; for Calculations yielded cubic meters per second.

[0099] Injecting flood discharge capacity parameters into the dam's own mechanical response proxy model set: In the upstream geological parameter field set With its own geological parameter field set The joint assessment of risk transmission impact within space. (Assume...) The current plan is = (Irreceptive wall depth 18.5 meters, grouting curtain thickness 2.2 meters). For combined scenarios Downstream safety factor For joint scenarios Downstream safety factor .

[0100] Table 7 Joint spatial risk assessment results for reservoirs (partial scenarios):

[0101] Calculate the risk transmission sensitivity index:

[0102] The risk transmission sensitivity index is 0.153, exceeding... The set acceptable threshold of 0.12 indicates that the impact of the upstream scheme on the downstream risk boundary conditions under the combined effects of geological uncertainties exceeds the acceptable range.

[0103] Identify risk transmission sensitivity indicators If the value exceeds the acceptable threshold of 0.12, a coordination request message (request identifier) ​​is generated. Initiator Recipient Expected flood discharge capacity constraints cubic meters per second (high priority) and sent back .

[0104] Upon receiving the coordination request, search the set of candidate robust solutions for solutions that satisfy the flood discharge capacity constraint: Existing schemes for cubic meters per second, discover alternative schemes. and The flood discharge capacities are 1028 cubic meters / s and 1015 cubic meters / s respectively, both of which meet the constraints. Choose the option with a lower probability of falling below the threshold. (The anti-seepage wall is 24.3 meters deep, and the grouting curtain is 3.0 meters thick.) As an alternative solution, this alternative solution is again presented to [the relevant authority / organization]. .

[0105] Table 8: Record of Plan Adjustments During Negotiation:

[0106] Recalculating the risk transmission sensitivity index, in the new joint space assessment, alternative solutions The corresponding range of flood discharge capacity parameters has been narrowed down to cubic meters per second, recalculated to obtain:

[0107] New risk transmission sensitivity index =0.107 is lower than the acceptable threshold of 0.12. The alternative solution has been accepted. Meanwhile, The best solution Passed down to downstream After conducting a risk transmission assessment and going through a similar consultation process, the three reservoirs finally reached an overall robust equilibrium for the basin after the third round of consultations.

[0108] The final reinforcement scheme, which integrates the three reservoirs, achieves a robust equilibrium state for the entire watershed: Reservoir Scheme , Reservoir Scheme , Reservoir Scheme Using joint Monte Carlo simulations, the Cartesian product space of the geological parameter fields of three reservoirs was analyzed. The risk distribution assessment of the watershed system was conducted, and a total of 50×50×50=125,000 joint scenarios were assessed.

[0109] Table 9 Summary of Watershed Coordination and Reinforcement Schemes:

[0110] Table 10. Overall Robustness Assessment Results of the Watershed:

[0111] The overall robustness assessment report for the watershed shows that under the coordinated reinforcement scheme, the probability of watershed system failure is 0.070, a 45.3% reduction compared to the 0.128 probability under the independent optimization scheme for each reservoir. Recommendations for the division of responsibility for risk transmission indicate that... The reservoir contributes the most to the overall risk of the basin, and it is recommended that the monitoring of the reservoir be increased in subsequent operation and maintenance.

[0112] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

Claims

1. A method for intelligent surveying, design, and simulation of water conservancy projects based on multi-source data fusion, characterized in that, Includes the following steps: Geological exploration borehole data and spatial coordinates of each reservoir in the basin, upstream and downstream topological relationships between reservoirs, reinforcement objective function parameters of each management entity, and description of information sharing mechanisms among management entities are obtained to generate a basic dataset of the basin reservoir group. For each reservoir, the sequential Gaussian simulation method is used to generate multiple equally probable geological parameter fields based on geological exploration borehole data to form a set of geological parameter fields for each reservoir. An agent decision-making model is configured for each reservoir and the interaction protocol between the agent decision-making models is defined to generate a multi-agent system configuration with geological uncertainty description. Each reservoir's intelligent agent decision-making model trains a dam mechanical response proxy model for each sample in its own geological parameter field set, generating a dam mechanical response proxy model set. The dam mechanical response proxy model set is then used to perform batch predictions on candidate schemes and calculate local robustness indices, generating a candidate robust scheme set for each reservoir. The optimal solution is passed on according to the upstream and downstream topological order. The downstream intelligent agent decision model calculates the risk transmission sensitivity index in the joint space of the upstream geological parameter field set and its own geological parameter field set. The downstream intelligent agent decision model generates a coordination request based on the risk transmission sensitivity index and sends it back to the upstream intelligent agent decision model. The upstream intelligent agent decision model incorporates the coordination constraints into the objective function and then re-selects alternative solutions, iteratively negotiates until the overall robust equilibrium of the watershed is achieved, and generates a coordinated reinforcement solution. Using joint Monte Carlo simulation, a risk distribution assessment of the watershed system is conducted within the combined space of geological parameter fields of each reservoir, and a coordinated reinforcement scheme for each reservoir and an overall robustness assessment report of the watershed are output.

2. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion as described in claim 1, characterized in that, The execution process of the sequential Gaussian simulation method includes: Normal transformation is performed on borehole measured data to establish a spatial correlation structure of geological parameters using spatial variability functions. Define the three-dimensional mesh nodes to be simulated, visit each three-dimensional mesh node in sequence according to the preset path, and for each node to be simulated, perform Kriging estimation using known borehole data and simulated node data to obtain the conditional mean and conditional variance. Randomly select a value from the corresponding conditional Gaussian distribution as the simulated value of the node to be simulated. After traversing all three-dimensional mesh nodes, a complete geological parameter field sample is generated. The above simulation process is repeated a preset number of times, each time using a different random number seed and node access path to generate multiple geological parameter field samples with equal probability.

3. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion as described in claim 1, characterized in that, The training process of the dam mechanical response surrogate model includes: For each geological parameter field realization sample in the geological parameter field set, the dam safety factor corresponding to multiple sets of reinforcement scheme parameters is calculated using the finite element method to generate a training sample set. The training sample set is input into the neural network for training, and the corresponding geological parameter field is generated to realize the dam mechanical response proxy model of the sample. The above process is repeated for all samples in the geological parameter field set to generate a set of dam mechanical response proxy models.

4. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion according to claim 3, characterized in that, The execution process of the finite element method includes: The dam body and foundation are divided into a finite number of units. Material parameters are assigned to each unit based on the geological parameter field. Boundary conditions and loading conditions are set according to the reinforcement scheme parameters. A set of mechanical equilibrium equations is established and solved to obtain the stress-strain field. The reduction coefficient when the dam body reaches the limit equilibrium state is calculated iteratively using the strength reduction method as the dam body safety factor.

5. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion according to claim 1, characterized in that, The local robustness index includes the mean safety factor, the minimum safety factor, and the probability of falling below the threshold. The average safety factor is the arithmetic mean of the safety factors of the candidate scheme under all geological parameter field implementation samples; The minimum safety factor is the minimum safety factor of the candidate scheme under all geological parameter field implementation samples; The probability below the threshold is the ratio of the number of samples in which the safety factor of the candidate scheme is lower than the safety factor threshold under all geological parameter field implementation samples to the total number of samples.

6. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion according to claim 1, characterized in that, The calculation process of the risk transmission sensitivity index includes: The upstream intelligent agent decision model transmits the flood discharge capacity parameters corresponding to the optimal solution to the downstream intelligent agent decision model. The flood discharge capacity parameters are calculated by the hydrological and hydraulic model based on the upstream reinforcement scheme parameters and the upstream geological parameter field. The downstream intelligent agent decision model injects the flood discharge capacity parameters into its own dam mechanical response proxy model set, traverses the joint space of the upstream geological parameter field set and its own geological parameter field set, and calculates the downstream safety factor under each joint scenario; The ratio of the number of scenarios with downstream security coefficients lower than the downstream security coefficient threshold to the total number of scenarios in all joint scenarios is used as a risk transmission sensitivity indicator.

7. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion according to claim 1, characterized in that, The iterative negotiation process includes: The downstream intelligent agent decision model determines whether the risk transmission sensitivity index exceeds the acceptable threshold. If it does, it generates a coordination request containing the expected flood discharge capacity constraint range and sends it back to the upstream intelligent agent decision model. The upstream agent decision-making model searches for a solution that meets the flood discharge capacity constraint in the candidate robust solution set. If such a solution exists, the one with the best local robustness index is selected as the alternative solution. If such a solution does not exist, a new candidate solution is generated in the constrained parameter space and the best solution is selected as the alternative solution. The alternative solution is passed to the downstream intelligent agent decision model to recalculate the risk transmission sensitivity index. If it still exceeds the acceptable threshold, the flood discharge capacity constraint range is adjusted and the coordination request is sent again until the overall robust equilibrium of the basin is achieved or the upper limit of the negotiation rounds is exceeded.

8. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion according to claim 7, characterized in that, The adjustment strategy for the flood discharge capacity constraint range is as follows: if the upstream alternative still causes the risk transmission sensitivity index to exceed the acceptable threshold, the downstream intelligent agent decision model will further reduce the upper limit of the flood discharge capacity constraint, and the reduction range will be dynamically determined based on the degree of exceeding the limit and the remaining negotiation rounds.

9. The intelligent survey, design, and simulation method for water conservancy projects based on multi-source data fusion according to claim 1, characterized in that, The execution process of the joint Monte Carlo simulation includes: A geological parameter field is extracted from the geological parameter field set of each reservoir to form a joint scenario; For the joint scenario, the flood discharge capacity parameters and dam safety factor of each reservoir are calculated sequentially according to the upstream and downstream topology, and the safety status of each reservoir and whether the watershed system fails are recorded under the joint scenario. Repeat the above sampling and calculation process until all combinations in the joint space are traversed or the preset number of samplings is reached, and statistically analyze the joint distribution characteristics of the failure probability of the watershed system and the safety factor of each reservoir. Among them, the failure of a watershed system is defined as the state in which the safety factor of at least one reservoir is lower than its corresponding safety factor threshold.

10. A smart survey, design, and simulation system for water conservancy projects based on multi-source data fusion, used to execute the smart survey, design, and simulation method for water conservancy projects based on multi-source data fusion as described in any one of claims 1 to 9, characterized in that, include: The data acquisition module is used to acquire geological exploration borehole data and spatial coordinates of each reservoir in the basin, the upstream and downstream topological relationships between reservoirs, the reinforcement objective function parameters of each management entity, and the description of the information sharing mechanism between management entities, and generate a basic dataset of the basin reservoir group. The multi-agent configuration module is used to generate a set of geological parameter fields for each reservoir using the sequential Gaussian simulation method, configure an agent decision model for each reservoir and define the interaction protocol, and generate a multi-agent system configuration with geological uncertainty description. The robust solution generation module is used to train a set of dam mechanical response proxy models for the geological parameter field set of each reservoir, calculate local robustness indices, and generate a set of candidate robust solutions for each reservoir. The risk transmission assessment module is used to transmit the optimal solution in the order of upstream and downstream topology and calculate the risk transmission sensitivity index in the joint space of the upstream and downstream geological parameter field sets. The iterative negotiation module is used to generate coordination requests based on risk transmission sensitivity indicators, incorporate coordination constraints into the upstream objective function, re-select alternative solutions, iterate and negotiate until the overall robust equilibrium of the basin is achieved, and generate coordinated reinforcement solutions. The risk assessment output module is used to assess the risk distribution of the watershed system using joint Monte Carlo simulation, and outputs coordinated reinforcement schemes for each reservoir and an overall robustness assessment report for the watershed.

Images