Hydraulic engineering digital design method based on digital twinning

Abstract

The invention discloses a hydraulic engineering digital design method based on digital twinning, and relates to the technical field of hydraulic engineering design. According to the digital twin-based hydraulic engineering digital design method, design demand information of a to-be-designed diversion tunnel is obtained and input into a pre-trained building recognition model for comprehensive analysis, a spatial layout map of the to-be-designed diversion tunnel is obtained, and a tunnel design digital twin model is constructed; simulation data of the tunnel design digital twinborn model are obtained, simulation analysis is carried out respectively, a tunnel construction risk index and a hydraulic action risk comprehensive index of the tunnel design digital twinborn model are obtained, comprehensive analysis is carried out, and a tunnel feasibility index of the tunnel design digital twinborn model is obtained; according to the method, the tunnel design digital twin model is subjected to design optimization based on the feasibility index, so that a lagging mode of re-evaluation after building is effectively avoided, later frequent reinforcement and reworking are prevented, and the engineering safety coefficient is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of water conservancy project design, and in particular to a digital design method for water conservancy projects based on digital twins. Background Art

[0002] With the continuous expansion of the scale and increase in complexity of water conservancy project construction, traditional design methods have gradually exposed many problems, such as the disconnection between design and actual working conditions, lack of dynamic simulation analysis capabilities, and information separation between design and construction. This design method that relies on static drawings and manual experience is difficult to fully cope with the challenges of extreme climate, hydrological uncertainty and geological complexity, and is not conducive to later construction control and risk prediction. In recent years, the widespread application of information technologies such as the Internet of Things, building information models, geographic information systems, big data and artificial intelligence in water conservancy projects has promoted the gradual upgrading of design methods and provided basic conditions for realizing the digitization and intelligence of the entire design process. Digital twin technology has emerged and gradually become an important development direction in the field of water conservancy projects. Digital twin technology builds a virtual model that is highly consistent with the actual project, realizes real-time mapping and interaction of physical entities in terms of structure, behavior and state, and transforms water conservancy project design from static drawing visualization to dynamic virtual simulation. However, in the design of tunnel water conservancy projects, the application of digital twin technology still faces several key difficulties, such as how to establish a scientific risk assessment mechanism.

[0003] Existing technologies, such as the digital twin-based digital design method for water conservancy projects disclosed in the patent application with publication number CN119442386A, include: collecting three-dimensional point cloud data of water conservancy projects through industrial cameras; constructing building information models based on the three-dimensional point cloud data of water conservancy projects; using the optimized edge folding algorithm to perform model lightweight processing on the building information model of water conservancy projects; using oblique photography technology to collect dense matching point cloud data of water conservancy projects through image sensors carried by drones; three-dimensional modeling of water conservancy projects; filtering processing of digital surface models using cloth simulation filtering algorithms; data fusion integration of building information twin models and digital surface twin models to establish a fusion maturity evaluation index system for digital twin models of water conservancy projects. By fusing and integrating building information model data and digital surface model data, multi-source data fusion management is performed to improve the information level of water conservancy projects.

[0004] Based on the above scheme, it is found that the limitations of the existing technology include at least the following problems: the existing technology lacks a comprehensive evaluation mechanism for the design scheme itself in terms of construction feasibility, hydraulic safety, etc., and fails to provide effective analysis and quantitative support for core issues such as whether there is a high risk in the construction process and whether the hydraulic environment has a non-negligible long-term impact on the structure. Especially in engineering areas with complex geological structures and frequent extreme hydrological conditions, the existing technology has not introduced a simulation module with geological stress modeling and hydrological erosion evolution analysis capabilities, which makes it difficult to make forward-looking judgments based on multi-dimensional risk indicators during the scheme review stage, and thus it is difficult to provide design optimization suggestions based on digital simulation. As a result, when encountering unpredictable risks such as geological dislocation, surrounding rock damage, and groundwater infiltration during the construction process, design changes and on-site reinforcements are forced to occur frequently, which not only increases the construction period and cost, but also brings long-term hidden dangers to engineering safety. Summary of the invention

[0005] In view of the shortcomings of the existing technology, the present invention provides a digital design method for water conservancy projects based on digital twins, which solves the problem that the existing technology is difficult to achieve simulation-based construction and hydraulic risk pre-assessment, which easily leads to design delays, increased costs and safety hazards.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a digital design method for water conservancy projects based on digital twins, comprising the following steps: obtaining design requirement information of a water diversion tunnel to be designed, and inputting it into a pre-trained building recognition model for comprehensive analysis to obtain a spatial layout map of the water diversion tunnel to be designed, and constructing a digital twin model of tunnel design; and obtaining simulation data of the digital twin model of tunnel design, wherein the simulation data includes geological simulation data and hydrological simulation data; performing simulation analysis on the simulation data of the digital twin model of tunnel design respectively to obtain a tunnel construction risk index and a comprehensive hydraulic action risk index of the digital twin model of tunnel design, and performing comprehensive analysis to obtain a tunnel feasibility index of the digital twin model of tunnel design; and optimizing the design of the digital twin model of tunnel design based on the feasibility index.

[0007] Furthermore, the design requirement information includes tunnel functional index requirements, preset spatial path coordinates and longitudinal slope range, and construction process constraints. The building recognition model is specifically a structure generation network, which includes an input layer, a parameter aggregation layer, a component classification reasoning layer, a size and material generation layer, and a spatial layout output layer.

[0008] Furthermore, the specific steps for obtaining the spatial layout map of the water diversion tunnel to be designed are as follows: in the input encoding layer of the structure generation network, the design requirement information of the water diversion tunnel to be designed is encoded to obtain a multi-dimensional design input embedding vector set of the water diversion tunnel to be designed; in the parameter aggregation layer of the structure generation network, the multi-dimensional design input embedding vector set of the water diversion tunnel to be designed is subjected to constraint relationship learning processing to obtain a design context semantic feature vector of the water diversion tunnel to be designed; in the component classification reasoning layer of the structure generation network, the design context semantic feature vector of the water diversion tunnel to be designed is subjected to structural component type classification processing to obtain an adaptation score value of each structural component of the water diversion tunnel to be designed; in the size and material generation layer of the structure generation network, the adaptation score value of each structural component of the water diversion tunnel to be designed is subjected to size reasoning and material parameter prediction processing to obtain a multi-dimensional structural parameter set of the water diversion tunnel to be designed; in the spatial layout output layer of the structure generation network, the multi-dimensional structural parameter set of the water diversion tunnel to be designed is subjected to layout reasoning processing to obtain a spatial layout map of the water diversion tunnel to be designed.

[0009] Furthermore, the geological simulation data includes a stratum heterogeneity index, an underground stress distribution index, and a surrounding rock stability index, and the specific steps for obtaining the tunnel construction risk index of the tunnel design digital twin model are as follows: standardizing the stratum heterogeneity index, underground stress distribution index, and surrounding rock stability index of the tunnel design digital twin model; and comprehensively analyzing the standardized stratum heterogeneity index, underground stress distribution index, and surrounding rock stability index of the tunnel design digital twin model to obtain the tunnel construction risk index of the tunnel design digital twin model.

[0010] Furthermore, the specific formula for calculating the tunnel construction risk index of the tunnel design digital twin model is as follows: ;in, Tunnel construction risk index for digital twin model design of tunnels, The stratigraphic heterogeneity index of the digital twin model for the tunnel design after standardization is is the heterogeneity adjustment coefficient stored in the database, The underground stress distribution index of the digital twin model of the tunnel design after standardization, is the stress distribution adjustment coefficient stored in the database, The surrounding rock stability index of the digital twin model for tunnel design after standardization. is the surrounding rock stability adjustment coefficient stored in the database, is the interaction adjustment coefficient stored in the database.

[0011] Furthermore, the hydrological simulation data includes hydraulic gradient value, water flow corrosion index, water flow kinetic energy value, and hydrological vulnerability index, and the specific steps for obtaining the comprehensive index of hydraulic action risk of the tunnel design digital twin model are as follows: comprehensively analyze the hydrological simulation data of the tunnel design digital twin model respectively to obtain the hydrological simulation index set of the tunnel design digital twin model, including the hydraulic impact effect index and the hydraulic abrasion sensitivity index; and comprehensively analyze the hydrological simulation index set of the tunnel design digital twin model to obtain the comprehensive index of hydraulic action risk of the tunnel design digital twin model.

[0012] Furthermore, the specific steps for obtaining the hydrological simulation index set of the tunnel design digital twin model are as follows: obtain the water flow value of the tunnel design digital twin model, and conduct a comprehensive analysis in combination with the hydraulic gradient value and the kinetic energy value to obtain the hydraulic impact effect index of the tunnel design digital twin model; obtain the water flow shear stress index of the tunnel design digital twin model, and conduct a comprehensive analysis in combination with the water flow corrosion index and the hydrological vulnerability index to obtain the hydraulic abrasion sensitivity index of the tunnel design digital twin model.

[0013] Furthermore, the specific formula for calculating the comprehensive index of hydraulic risk of the tunnel design digital twin model is as follows: ;in, A comprehensive index of hydraulic risk for tunnel design digital twin models, Design of hydraulic impact index for digital twin model of tunnel, is the shock effect adjustment coefficient stored in the database, Design of hydraulic erosion sensitivity index for digital twin model of tunnel, is the abrasion sensitivity adjustment coefficient stored in the database, is the synergy adjustment coefficient stored in the database, is the risk coupling adjustment coefficient stored in the database, is the smoothing coefficient stored in the database.

[0014] Furthermore, the specific formula for calculating the tunnel feasibility index of the tunnel design digital twin model is as follows: ;in, Tunnel feasibility index for designing digital twin models of tunnels, Tunnel construction risk index for digital twin model design of tunnels, is the tunnel construction adjustment coefficient stored in the data, A comprehensive index of hydraulic risk for tunnel design digital twin models, is the hydraulic adjustment coefficient stored in the data, is the overlay adjustment coefficient stored in the data, is the tunnel construction coefficient stored in the data, is the hydraulic action coefficient stored in the data, .

[0015] Furthermore, the specific steps for design optimization of the tunnel design digital twin model based on the feasibility index are as follows: the tunnel feasibility index of the tunnel design digital twin model is judged and analyzed with the preset tunnel feasibility index threshold; if the tunnel feasibility index of the tunnel design digital twin model is lower than or equal to the preset tunnel feasibility index threshold, the first optimization measure is taken; if the tunnel feasibility index of the tunnel design digital twin model is higher than the preset tunnel feasibility index threshold, the second optimization measure is taken.

[0016] The present invention has the following beneficial effects: (1) This digital twin-based water conservancy project digital design method constructs a digital twin model of the water diversion tunnel and introduces geological simulation data and hydrological simulation data for dynamic evaluation. It can conduct multi-dimensional simulation analysis of construction feasibility and hydraulic environmental risks in the early design stage, and integrate them to form a tunnel feasibility index for design decision-making, thereby effectively avoiding the lag mode of re-evaluation after completion. For example, in actual projects, if simulation analysis finds that a certain section of the tunnel has high underground stress and is accompanied by a high risk of water flow impact, support measures can be preset or the path can be adjusted in the design stage, thereby preventing frequent reinforcement and rework in the later stage, thereby fundamentally reducing construction costs and cycles, and improving the project safety factor.

[0017] (2) This digital twin-based water conservancy project digital design method introduces a structure generation network, takes the design requirement information as the model input, and relies on a multi-layer structure such as input coding, parameter aggregation, component classification, size generation and spatial layout to automatically infer the structural component type, size parameters, material configuration and spatial layout map of the target tunnel, and then directly constructs a digital twin model of the tunnel design, which significantly reduces the manual intervention and parameter extraction workload required for modeling, and improves the automation and accuracy of structural modeling. For example, it automatically outputs a component layout map with spatial positioning and connection relationships, and constructs a digital twin model of the tunnel design, thereby providing high-precision digital basic model support for subsequent construction simulation, hydrological analysis and structural evaluation.

[0018] (3) This digital twin-based water conservancy project digital design method introduces a multi-dimensional index after the digital twin model of tunnel design is constructed, and quantifies the geological and hydrological multi-source simulation information into a structured and computable indicator system, which is summarized as a tunnel feasibility index. The feasibility of the design scheme is judged and dynamically optimized with the set threshold, thereby enabling the design process to shift from empirical decision-making to data-driven decision-making. For example, in water diversion projects in mountainous areas with complex geological conditions and drastic hydrological changes, this method can reduce the frequency of construction changes and safety hazards caused by environmental uncertainties, thereby effectively improving the project adaptability, robustness and risk prevention and control capabilities in complex mountainous environments.

[0019] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is a flow chart of the digital design method for water conservancy projects based on digital twins of the present invention.

[0021] Figure 2 It is a flow chart of the steps of obtaining the spatial layout map of the water diversion tunnel to be designed in the digital design method of water conservancy projects based on digital twins of the present invention.

[0022] Figure 3 This is an example diagram of geological simulation data of the digital twin model for tunnel design in the digital twin-based water conservancy project digital design method of the present invention.

[0023] Figure 4 This is a flow chart of the steps for obtaining the comprehensive hydraulic action risk index of the digital twin model of tunnel design in the digital twin-based water conservancy project digital design method of the present invention. DETAILED DESCRIPTION

[0024] See also Figure 1 The embodiment of the present invention provides a technical solution: a digital twin-based water conservancy project digital design method, comprising the following steps: obtaining design requirement information of a water diversion tunnel to be designed, and inputting it into a pre-trained building recognition model for comprehensive analysis to obtain a spatial layout map of the water diversion tunnel to be designed, and constructing a tunnel design digital twin model; and obtaining simulation data of the tunnel design digital twin model, the simulation data including geological simulation data and hydrological simulation data; performing simulation analysis on the simulation data of the tunnel design digital twin model respectively to obtain a tunnel construction risk index and a comprehensive hydraulic action risk index of the tunnel design digital twin model, and performing comprehensive analysis to obtain a tunnel feasibility index of the tunnel design digital twin model; and optimizing the design of the tunnel design digital twin model based on the feasibility index.

[0025] The specific steps for constructing a digital twin model for tunnel design are as follows: read the component type, spatial coordinates, dimensional parameters, material properties and other fields in the spatial layout map; for each component type (such as main tunnel section, ventilation shaft, inspection shaft, etc.), call the component template library in the 3D modeling engine (such as OpenCascade, Revit API, CityGML model library), generate the geometric entity model (B-Rep or Mesh structure) of the component according to the dimensional parameters, and perform spatial posture transformation operations on the generated component entity according to the spatial coordinates and direction vector of each component in the map (translation, moving the origin of the local coordinate system of the component entity to the corresponding layout coordinate point; rotation, adjusting the posture according to the orientation vector so that the component orientation is consistent with the layout path), and perform interface docking processing on adjacent components according to the "adjacent connection node" field in the map, including node alignment (adjusting the connection end face to make it fit accurately in geometry), interface type matching (calling different connection methods according to the component type, such as plug-in , overlap, connector transition), topological connectivity processing (building a connection diagram structure between components for subsequent structural simulation or construction simulation), and then integrating the geometric elements, spatial information, and connection relationships of each component into a complete structural model, and converting it to a unified spatial reference coordinate system (such as WGS-84, building coordinate system or custom engineering local coordinate system); at the same time, integrating the overall model attributes, including global metadata such as project number, path sequence number, start and end points, and registering the completed tunnel design digital twin model to the digital twin platform or middle office system, and associating metadata, design stage labels, task numbers, etc., to realize the model's call and dynamic update capabilities in subsequent visualization, simulation, and analysis.

[0026] The specific formula for calculating the tunnel feasibility index of the tunnel design digital twin model is as follows: ;in, Tunnel feasibility index for designing digital twin models of tunnels, Tunnel construction risk index for digital twin model design of tunnels, is the tunnel construction adjustment coefficient stored in the data, A comprehensive index of hydraulic risk for tunnel design digital twin models, is the hydraulic adjustment coefficient stored in the data, is the overlay adjustment coefficient stored in the data, is the tunnel construction coefficient stored in the data, is the hydraulic action coefficient stored in the data, .

[0027] It needs to be explained that , , It can be obtained through the following steps: using historical simulation data, combined with the tunnel construction risk index and the comprehensive hydraulic risk index, statistical regression analysis is carried out to quantify the specific impact of each factor on the tunnel feasibility index, so as to fit the initial weight value. Secondly, the sensitivity analysis method is used to adjust the value range of each coefficient, and observe its impact on the tunnel feasibility assessment results to ensure the stability and rationality of the model. Based on the characteristics and actual conditions of the tunnel construction area, the preliminary fitting coefficients are corrected and optimized, and finally the coefficient values ​​applicable to the tunnel construction area are determined.

[0028] , It can be obtained through the following steps: read the tunnel construction risk index and hydraulic risk comprehensive index of the tunnel design digital twin model, and perform sum analysis to obtain the feasible sum value of the tunnel, and perform proportion analysis on the tunnel construction risk index and hydraulic risk comprehensive index of the tunnel design digital twin model with the reasonable design value respectively, and use the proportion analysis results as the corresponding coefficients.

[0029] Specifically, the design requirement information includes the functional index requirements of the tunnel, the preset spatial path coordinates and longitudinal slope range, and the construction process constraints. The building recognition model is specifically a structure generation network, which includes an input layer, a parameter aggregation layer, a component classification reasoning layer, a size and material generation layer, and a spatial layout output layer.

[0030] Among them, the input layer is used to format and vectorize the design requirement information of the water diversion tunnel to be designed. The input includes tunnel functional indicators (such as passage width, hydraulic flow), spatial path coordinate sequence, longitudinal slope range, construction process constraints, etc., which are converted into a unified format of design input embedded vector set by normalization, multi-dimensional embedding mapping, position encoding, etc. according to the parameter type.

[0031] The parameter aggregation layer is used to extract the dependencies and interaction logic between design parameters (such as the impact of path changes on component configuration and the constraints of construction methods on material selection), and output a contextual semantic feature vector representing the semantic state of the current design target.

[0032] The component classification reasoning layer is used to infer the types of structural components that should be included in the current design scenario, and output functional adaptation scores for each type of component to form a set of structural component-adaptation score pairs, which are used for component screening and weight control in the subsequent generation process.

[0033] The size and material generation layer is used for the selected target components. The model further infers their geometric size parameters (such as length, width, height, thickness) and structural material properties (such as concrete strength, corrosion resistance level, coating type, etc.) based on the context information and adaptation score values, and outputs the parameter configuration plan for each component in a structured form.

[0034] The spatial layout output layer is used to map the various types of generated structural components to the spatial path, determine their three-dimensional position coordinates, direction, connection node relationship, etc., and generate the component layout order and spatial logical topology, and finally form a spatial component layout map.

[0035] like Figure 2As shown in FIG. 1 , the specific steps for obtaining the spatial layout map of the water diversion tunnel to be designed are as follows: in the input coding layer of the structure generation network, the design requirement information of the water diversion tunnel to be designed is encoded (functional indicators are numerically normalized and vectorized, spatial path coordinates are sequence modeled and encoded, construction process constraints are classified and vectorized, and longitudinal slope interval parameters are split and standardized. Finally, the various parameter embedding vectors after the above processing are spliced ​​into a multidimensional input feature vector set in a unified format as the input of the structure generation network), and the multidimensional design input embedding vector set of the water diversion tunnel to be designed is obtained; in the parameter aggregation layer of the structure generation network, the multidimensional design input of the water diversion tunnel to be designed is encoded. The embedded vector set is input for constraint relationship learning processing (context semantic modeling processing is performed on the multi-dimensional design requirement embedding vector from the input coding layer, that is, based on Transformer encoding, the semantic association, regulatory effect and constraint coupling relationship between parameters such as functional indicators, spatial path coordinates, longitudinal slope range and construction restrictions are learned to build a unified design target representation space. During the processing, the network automatically identifies the mutual influence relationship between key parameters through the attention mechanism, such as the restrictions of construction technology on the layout of path components, the regulatory effect of hydraulic flow on the spacing and thickness of components, etc., and finally generates a set of high-dimensional semantic vectors used to represent the semantic state of the overall design task), and the design context semantics of the water diversion tunnel to be designed are obtained. feature vector; in the component classification reasoning layer of the structure generation network, the design context semantic feature vector of the water diversion tunnel to be designed is classified into structural component types (based on the design requirements of typical water diversion tunnel engineering, a candidate set of structural component types is preset, and the set includes but is not limited to the main tunnel section, ventilation shaft section, maintenance shaft section, turning section, node connection section, support structure unit, lining layer structure, diversion channel component, etc., and each component has a unique identifier in the candidate set, and is indexed and associated with the structure size range and function attribute field. Secondly, the design context semantic feature vector is used as input, and the vector is a global embedded feature vector including water diversion function indicators, spatial path coordinate logic, longitudinal slope constraint relationship and construction process restriction conditions. The eigenvalue indicates that, again, the context vector is input into the component classification neural network, which includes at least two perceptron layers. Each layer performs feature mapping on the input semantics through linear mapping and nonlinear activation function, and connects the Sigmoid activation function node at the end to output a relatively independent existence score for each component type. For each candidate component type, the network will output an adaptation score value between [0, 1] based on the semantic matching degree between the design context and the component in the historical modeling. The score value is the confidence that the component "should be included in the plan" in the current design context. The higher the score, the higher the priority of subsequent generation). The adaptation score value of each structural component of the water diversion tunnel to be designed is obtained;In the size and material generation layer of the structure generation network, the adaptation score of each structural component of the water diversion tunnel to be designed is processed by size reasoning and material parameter prediction (according to the score weight of each type of component, the components are ranked in importance, and the subset of components that need to enter the size generation link is screened out according to the set score threshold. At the same time, the score value is passed into the size and material generation submodule as a weight factor to guide the main feature enhancement and redundant feature suppression in the subsequent parameter calculation. For each component to be generated, its type label, score value, context semantic vector, functional target parameter, etc. are used as input and sent to the size generation subnetwork. The subnetwork is composed of a multi-layer perception mechanism, and the output includes but is not limited to the length, width, height, thickness, radius and other geometric size parameters of the component. For example: for a ventilation shaft component, the model will comprehensively consider the ventilation efficiency, longitudinal space restrictions and construction technology, etc., and output the reasonable diameter value and structural height of the ventilation shaft, etc., and further perform material selection and performance level reasoning on each component based on the component function type, construction restrictions, historical design samples, etc. Features, predict the corresponding material combination, including but not limited to concrete strength grade, impermeability grade, anti-corrosion coating type, composite structural material type, etc., and the material attribute parameters are represented in a structured form, as the auxiliary configuration field of the component bound to the size parameter, and finally, the model encapsulates the size parameters and material attribute parameters of each type of component generated into a structural component parameter set), and obtains the multi-dimensional structural parameter set of the water diversion tunnel to be designed; in the spatial layout output layer of the structure generation network, the multi-dimensional structural parameter set of the water diversion tunnel to be designed is subjected to layout reasoning processing (reading the preset spatial path coordinate sequence and longitudinal slope range information in the design requirement information, combining the size parameters and functional characteristics of each structural component, and determining the spatial projection reference line of the layout path; all components are bound to the path section in sequence according to the functional logic, such as the main tunnel section is given priority, and the auxiliary components such as ventilation shafts and inspection shafts are inserted according to the strategy, and the path alignment calibration is performed on the starting position, and for each component to be laid out, based on its size parameters and functional type, the path line direction vector and the longitudinal slope surface are used for projection positioning, and the three-dimensional center coordinate point of the component is calculated;At the same time, combined with the construction process restrictions in the context, such as the minimum turning radius of the shield machine and the maximum layout section, the spatial feasibility is judged. If there is a spatial conflict or crowded layout, the component spacing is dynamically adjusted or the layout position is slightly shifted to meet the layout feasibility. Then, based on the spatial position determination, the model further generates a connection relationship diagram between adjacent components, including connection node numbers, relative direction vectors, interface matching methods, such as plug-in, overlap, transition structure, etc., and fine-tunes the spatial layout plan through graph optimization mechanisms, such as the shortest path algorithm or topological compression, to ensure layout connectivity, functional integrity and structural continuity. Finally, the spatial coordinates, direction orientation, arrangement order, connection relationship and other information of each component are encapsulated as a component spatial layout map) to obtain the spatial layout map of the water diversion tunnel to be designed. ;

[0036] Among them, the pre-training process of the structure generation network is as follows: Obtain an annotated data set, that is, obtain structural design data from multiple historical water diversion tunnels or underground engineering projects, including but not limited to design requirement information (functional indicators, path coordinates, construction technology), types of structural components in the built model, dimensional parameters of each component, material properties, spatial position layout relationships, etc.; organize the above data into input and output pair samples, construct a multimodal structural design sample set, and divide it into a training set and a validation set.

[0037] Perform parameter initialization on each functional module in the structure generation network: use Xavier or He initialization method to set the initial weight for the embedding layer; load preset weight parameters (such as position encoding) for the Transformer structure in the aggregation layer; initialize the bias item of the generation sub-network involved in the size and layout layer to zero, and uniformly set the activation function to ReLU or GELU to enhance nonlinear modeling capabilities.

[0038] Set the loss function, train the loss function of the component classification subnetwork (the goal is to minimize the multi-label cross entropy loss between the predicted component type and the actual component type), the size and material generation submodule loss function (the goal is to minimize the mean square error loss of the structural size and material properties), and the spatial layout submodule loss function (optimize the distance error and layout order similarity between the component layout position and the actual model layout position).

[0039] Training is performed based on the training set, the number of training cycles is set, and forward propagation is performed in each training cycle (the design requirement information of each group of samples in the training set, including functional indicators, path coordinates, construction restrictions, etc., is used as input, and sequentially passes through the input encoding layer, parameter aggregation layer, component classification reasoning layer, size and material generation layer, and spatial layout layer to generate the predicted output results of the structural components, including component type labels, size parameters, material properties, and spatial position layout results), loss function calculation (for the output results of different modules, a multi-task loss function is set for comprehensive evaluation, specifically including: multi-label cross entropy loss for component classification; Mean square error loss of dimensional parameters and material properties; L2 distance error between spatial layout coordinates and true coordinates; structural consistency loss of component layout order, such as edit distance or topological similarity), back propagation (based on the total loss value, the gradient of parameters of each layer in the network is calculated through the back propagation algorithm, and the error signal is transmitted back along the network level to correct the prediction deviation in the model), parameter update (the Adam optimizer is used to update the model weights, and an adaptive learning rate mechanism is set to speed up convergence and prevent oscillation; if necessary, Dropout and regularization methods are introduced to improve the generalization ability of the model and avoid overfitting).

[0040] After each training session, the current model performance is evaluated based on the validation set, and indicators such as accuracy, mean absolute error (MAE), F1 value, and IoU overlap of each output task are calculated. A performance trend chart is drawn for training monitoring. If there is no significant improvement in the evaluation indicators on the validation set, the early stopping mechanism is triggered to prevent the model from overfitting. If the number of training rounds reaches the set upper limit or the performance reaches the set threshold, the training process is terminated.

[0041] A set of network parameter weights that achieves the best performance on the validation set is saved as the final model version for the automatic tunnel structure generation task.

[0042] In this implementation scheme, by constructing a structure generation network and using a pre-training mechanism to enhance the network's parameter learning and modeling capabilities, the intelligence level of digital modeling of tunnel structures and the accuracy of model output are significantly improved. Secondly, the model introduces various types of design requirement information in the input layer, such as functional indicators, spatial path coordinates, construction process restrictions, etc., and realizes interactive correlation modeling between different input parameters through multi-dimensional encoding and semantic fusion technology; the Transformer aggregation mechanism effectively captures the potential dependencies and structural logic between constraints, so that the model has strong generalization and reasoning capabilities. In the training process, the network introduces multi-task loss functions such as component classification, size generation and layout position, so that the model can not only correctly identify the type of structural components to be configured, but also accurately predict the geometric dimensions, material properties and spatial layout relationships of each component. Finally, multi-stage training and early stopping mechanisms are adopted to ensure the stability and comprehensive performance of the model under different task indicators, thereby significantly reducing the manpower input and parameter error problems in engineering design, thereby improving the design response efficiency, and then providing a high-precision, structurally consistent digital twin basic model for subsequent simulation analysis and optimization.

[0043] Specifically, the geological simulation data includes the stratigraphic heterogeneity index, the underground stress distribution index, and the surrounding rock stability index, and the specific steps for obtaining the tunnel construction risk index of the tunnel design digital twin model are as follows: standardizing the stratigraphic heterogeneity index, the underground stress distribution index, and the surrounding rock stability index of the tunnel design digital twin model; and comprehensively analyzing the stratigraphic heterogeneity index, the underground stress distribution index, and the surrounding rock stability index of the tunnel design digital twin model after standardization to obtain the tunnel construction risk index of the tunnel design digital twin model.

[0044] Among them, the stratigraphic heterogeneity index is the stratigraphic complexity of the line where the water diversion tunnel is located. It can be obtained by obtaining the depth value of each stratigraphic layer in multiple sections (50m as a section) of the line where the water diversion tunnel is located (which can be obtained by geological radar, that is, the depth of different underground stratigraphic layers can be inferred by the time difference of the reflected wave and the signal intensity), stratigraphic inclination value (which can be obtained by geological inclinometer), and performing standardization processing, performing weighted processing based on the standardized processing result, and performing average processing based on the weighted processing result. The result is the stratigraphic heterogeneity index.

[0045] The underground stress distribution index is the ground stress state of the strata in multiple sections of the line where the tunnel construction is located. It can be obtained by obtaining the tunnel burial depth values ​​of multiple sections of the line where the water diversion tunnel is located (that is, the vertical depth of the tunnel from the surface to the tunnel top, which can be obtained by geological radar), vertical ground stress (vertical stress acting on the underground rock formation, which can be obtained by stress sensors, that is, when the underground rock formation is subjected to stress, the material of the sensor deforms and generates voltage or current signals, which can be converted into stress values), horizontal ground stress (horizontal stress on the underground rock formation, which can be obtained by stress sensors), and performing standardization processing, weighting processing based on the standardized processing results, and average processing based on the weighted processing results. The result is the underground stress distribution index.

[0046] The surrounding rock stability index is a measure of the stability of the surrounding rock (the rock and soil layer around the tunnel) during tunnel excavation. The stability of the surrounding rock directly affects the construction safety and later operation of the tunnel. It can be obtained by obtaining the uniaxial compressive strength value (the maximum pressure that the rock can withstand without lateral constraints), tensile strength value, internal friction angle value (friction between rock and soil material particles), and cohesion value (the bonding force between rock internal particles formed by chemical or physical action) of multiple sections of the line where the water diversion tunnel is located and performing standardization. Based on the standardized processing results, weighted processing is performed, and based on the weighted processing results, average processing is performed. The result is the surrounding rock stability index. The uniaxial compressive strength value, tensile strength value, internal friction angle value, and cohesion value can all be obtained through the field sampling method, and the results of the field sampling method are uploaded to the simulation database. For example, for the uniaxial compressive strength value, representative rock samples are collected from the tunnel construction area and loaded using a material testing machine. Apply pressure at a slower loading speed until the rock sample is damaged, record stress-strain data through sensors, draw stress-strain curves, and find the peak stress, which is the uniaxial compressive strength; for tensile strength value, collect rock samples from the tunnel construction area, usually select rock samples of different depths and directions to ensure representativeness, cut the rock samples into standard sizes (such as cylindrical samples with a diameter of 50 mm), use a tensile testing machine, gradually apply tension until the rock sample breaks, and record the maximum tensile stress, which is the tensile strength; for internal friction angle value, collect rock and soil samples, and place the rock samples in a direct shear test device, apply shear force until the sample slides, and calculate the internal friction angle through the relationship between shear stress and normal stress; for cohesion value, collect rock and soil samples, and conduct triaxial tests. The samples are compressed under pressure, and the shear strength of the samples is measured under different stress conditions. The cohesion value is obtained by calculating the relationship between shear strength and normal stress.

[0047] The specific formula for calculating the tunnel construction risk index of the tunnel design digital twin model is as follows: ;in, Tunnel construction risk index for digital twin model design of tunnels, The stratigraphic heterogeneity index of the digital twin model for the tunnel design after standardization is is the heterogeneity adjustment coefficient stored in the database, The underground stress distribution index of the digital twin model of the tunnel design after standardization, is the stress distribution adjustment coefficient stored in the database, The surrounding rock stability index of the digital twin model for tunnel design after standardization. is the surrounding rock stability adjustment coefficient stored in the database, is the interaction adjustment coefficient stored in the database.

[0048] It should be explained that the formula This item is used to adjust the superposition effect between the stratum heterogeneity index, underground stress distribution index and surrounding rock stability index to avoid the tunnel construction risk index being too high or too low.

[0049] , , , It can be obtained through the following steps: Based on historical simulation data, the initial influence weights of each variable (stratum heterogeneity index, underground stress distribution index, surrounding rock stability index) on the tunnel construction risk index are determined through statistical regression analysis. Then, the value range of the coefficient is adjusted using the sensitivity analysis method to evaluate the stability and applicability of these parameters to the formula output. Next, the weights are further fitted through model optimization (such as machine learning algorithms or multi-objective optimization) to ensure that the formula can accurately reflect the actual tunnel construction risks. The coefficients are fine-tuned based on the characteristics of the tunnel construction area to ensure that they are suitable for specific tunnel construction risk assessment needs.

[0050] The specific implementation example of calculating the tunnel construction risk index of the tunnel design digital twin model is as follows. The following data are available: the stratum heterogeneity index, underground stress distribution index, and surrounding rock stability index of the tunnel design digital twin model, as shown in Table 1 and Figure 3 Said: Table 1 Examples of geological simulation data for the digital twin model of tunnel design The data in Table 1 are standardized to obtain Table 2: Table 2. Example of geological simulation data of the digital twin model for tunnel design after standardization Heterogeneous adjustment coefficients stored in the database is: 0.268; Adjustment factors for stress distribution stored in the database is: 0.873; Surrounding rock stability adjustment coefficient stored in the database is: 0.493; Interaction adjustment coefficients stored in the database is: 0.265; Substituting the data in Table 2 and the above adjustment coefficients into the specific formula for calculating the tunnel construction risk index of the tunnel design digital twin model, we obtain: Calculate the tunnel construction risk index of the tunnel design digital twin model = ((0.427 0.268 +ln(1+0.873×0.618)) / (1+exp(-0.493×0.562)))×(1+(0.618 / 0.562)) 0.265 ≈0.847.

[0051] In this implementation scheme, the formation heterogeneity index, underground stress distribution index and surrounding rock stability index involved in the geological simulation data are standardized, weighted and integrated to finally form a tunnel construction risk index, so that the potential geological risks in the tunnel construction process can be systematically and quantitatively evaluated, which has important engineering application value. Secondly, multiple geological indicators are integrated into a computable comprehensive risk quantification result, thereby significantly improving the scientificity and accuracy of the evaluation. In particular, the superposition of formation heterogeneity and underground stress distribution often causes sudden geological disasters such as soil collapse and water inrush during construction. This method effectively captures the nonlinear synergistic effect between indicators by introducing an interactive adjustment coefficient, avoiding the misjudgment caused by a single indicator evaluation. Finally, the various adjustment coefficients in the formula can be adaptively adjusted based on historical simulation data through regression and optimization algorithms to ensure that it has good regional adaptability and model generalization ability. For example, in high-stress areas such as when a tunnel passes through a fault fracture zone, the index can significantly identify high-risk sections in advance, and provide accurate references for the design unit to formulate reinforcement plans and optimize support structures in advance, thereby reducing sudden changes, controlling construction costs and ensuring construction safety.

[0052] Specifically, Figure 4As shown in the figure, the hydrological simulation data includes hydraulic slope value, water flow corrosion index, water kinetic energy value, and hydrological vulnerability index, and the specific steps for obtaining the comprehensive index of hydraulic action risk of the tunnel design digital twin model are as follows: the hydrological simulation data of the tunnel design digital twin model are comprehensively analyzed to obtain the hydrological simulation index set of the tunnel design digital twin model, including the hydraulic impact effect index (the ability of water flow to cause impact damage to the tunnel structure) and the hydraulic abrasion sensitivity index (quantifying the tendency of tunnel structure to be damaged under long-term scouring of water flow and chemical erosion of groundwater); and the hydrological simulation index set of the tunnel design digital twin model is comprehensively analyzed to obtain the comprehensive index of hydraulic action risk of the tunnel design digital twin model.

[0053] Among them, the hydraulic gradient value is the average of the groundwater level change rate values ​​of multiple sections of the line where the water diversion tunnel is located. It can be obtained by obtaining the groundwater level values ​​of the starting point and the end point of each section of the line where the water diversion tunnel is located (all of which can be obtained by geological radar) and performing a ratio analysis (that is, the absolute value of the difference between the groundwater level values ​​of the starting point and the end point / section distance). The result is the hydraulic gradient value.

[0054] The water flow corrosion index is the corrosion risk caused by the surrounding groundwater of the water diversion tunnel during the infiltration of the tunnel surrounding rock. It can be obtained by obtaining the pH value, sand content, and corrosive ion concentration of groundwater in multiple sections of the water diversion tunnel, and performing mean analysis. Based on the mean analysis results, standardization processing and weighted processing are performed. The result is the water flow corrosion index. The pH value, sand content, and corrosive ion concentration can all be obtained through the groundwater monitoring report stored in the regional hydrogeological database. For example, the pH value of groundwater at several historical times is obtained and averaged, and the result is the pH value of groundwater.

[0055] The kinetic energy of water is a measure of the impact of water on tunnel structures. It can be obtained by obtaining the flow velocity value of each section of the water diversion tunnel (already built) under the same environment (which can be obtained through a flow velocity sensor), averaging it, and obtaining the water density (directly taking the value as 1000kg / m3). It is then analyzed based on the fluid kinetic energy formula, and the result is the kinetic energy of water.

[0056] The hydrological vulnerability index is used to assess the vulnerability of the tunnel area under extreme hydrological conditions (such as heavy rainfall, floods, etc.). It can be obtained by analyzing the rainfall of several extreme hydrological conditions in history, the groundwater level (i.e., the final water level after the extreme hydrological conditions), the water level rise rate (i.e., the absolute value of the difference between the initial water level and the final water level / the required time), and performing mean analysis. Based on the mean analysis results, it is standardized and weighted. The result is the hydrological vulnerability index. The rainfall can be obtained through the historical meteorological observation reports stored in the meteorological database, and the groundwater level and the water level rise rate can be obtained through the groundwater monitoring reports stored in the hydrogeological database.

[0057] The specific steps for obtaining the hydrological simulation index set of the tunnel design digital twin model are as follows: obtain the water flow value of the tunnel design digital twin model, and conduct a comprehensive analysis in combination with the hydraulic gradient value and the kinetic energy value (i.e., first perform standardization processing, and then perform weighted processing based on the standardized processing results) to obtain the hydraulic impact effect index of the tunnel design digital twin model; obtain the water flow shear stress index of the tunnel design digital twin model, and conduct a comprehensive analysis in combination with the water flow corrosion index and the hydrological vulnerability index (i.e., first perform standardization processing, and then perform weighted processing based on the standardized processing results) to obtain the hydraulic abrasion sensitivity index of the tunnel design digital twin model.

[0058] The water flow value can be obtained by obtaining the water flow value of each section of the water diversion tunnel (already built) under the same environment (which can be obtained through the historical monitoring data of the built tunnels stored in the database, that is, the water diversion tunnels built under the same environment), and performing averaging processing, and the result is the water flow value.

[0059] The water flow shear stress index is the shear stress intensity generated by water flow on the tunnel wall or bottom. It can be obtained by obtaining the shear stress intensity value of each section of the water diversion tunnel (already built) under the same environment (it can be obtained through the historical monitoring data of the built tunnels stored in the database, that is, the water diversion tunnels built under the same environment), and performing average processing. The result is the water flow shear stress index.

[0060] The specific formula for calculating the comprehensive index of hydraulic risk of the digital twin model of tunnel design is as follows: ;in, A comprehensive index of hydraulic risk for tunnel design digital twin models, Design of hydraulic impact index for digital twin model of tunnel, is the shock effect adjustment coefficient stored in the database, Design of hydraulic erosion sensitivity index for digital twin model of tunnel, is the abrasion sensitivity adjustment coefficient stored in the database, is the synergy adjustment coefficient stored in the database, is the risk coupling adjustment coefficient stored in the database, is the smoothing coefficient stored in the database, and in this implementation example, its value is 0.05.

[0061] It should be explained that the formula This item is used to adjust the interaction between the hydraulic impact effect index and the hydraulic abrasion sensitivity index to avoid the comprehensive hydraulic risk index being too high or too low.

[0062] , , , It can be obtained through the following steps: using historical simulation data, combining the hydraulic impact effect index and the hydraulic abrasion sensitivity index, conducting statistical regression analysis, quantifying the specific impact of each factor on the comprehensive index of hydraulic risk, and thus fitting the initial weight value; secondly, using the sensitivity analysis method, adjusting the value range of each coefficient, observing its impact on the comprehensive assessment results of hydraulic risk, ensuring the stability and rationality of the model; based on the characteristics of the tunnel construction area and the actual situation, correcting and optimizing the preliminary fitting coefficients, and finally determining the coefficient values ​​suitable for the characteristics of the tunnel construction area.

[0063] In this implementation plan, a complete hydrological simulation data system is constructed, and a multi-dimensional risk index is introduced to systematically model and quantitatively evaluate the hydrodynamic environment of the area where the water diversion tunnel is located, so as to finally form a comprehensive index of hydraulic action risk, thereby improving the prediction and response capabilities of the tunnel hydrological safety risks, and effectively identifying the impact of extreme weather or groundwater changes on the long-term stability of the structure. Secondly, through standardized processing, weighted combination, and dual-channel modeling paths of flow and shear stress, the hydraulic impact effect index and hydraulic abrasion sensitivity index are extracted respectively, and nonlinear interactive modeling is performed by introducing synergistic adjustment coefficients, risk coupling factors, and smoothing coefficients to ensure that the evaluation results are closer to the actual project operation status, thereby avoiding misjudgment caused by improper weighting of single indexes. Finally, all coefficients can be dynamically fitted and regionally adjusted based on historical simulation and field monitoring data, so that the model has good adaptability and scalability, thereby significantly improving the predictability of the durability and risks of the tunnel in a complex hydrological environment. At the same time, it also provides solid hydrological data support for subsequent feasibility analysis and construction design, and then effectively avoids hidden dangers such as seepage, corrosion, and structural erosion, and ensures the long-term stability and safety of the project operation.

[0064] Specifically, the specific steps for design optimization of the tunnel design digital twin model based on the feasibility index are as follows: the tunnel feasibility index of the tunnel design digital twin model is judged and analyzed with the preset tunnel feasibility index threshold; if the tunnel feasibility index of the tunnel design digital twin model is lower than or equal to the preset tunnel feasibility index threshold, the first optimization measure is taken (i.e., the design scheme of the water diversion tunnel is modified), which is specifically based on the highest influencing factors in the tunnel construction risk index and the hydraulic action risk comprehensive index, the component configuration and path layout information in the model are reversely tracked to identify the design components or path areas that cause excessive risk, and the following operations are performed: if the risk is increased due to the formation heterogeneity index and the surrounding rock stability deviation, the lining thickness and the material strength of the supporting components in the relevant section are increased; if the high risk is caused by excessive hydraulic slope and abnormal hydrological vulnerability indicators, the longitudinal slope form of the section is adjusted or components such as anti-seepage protective layer and drainage channel are added; for areas where the component arrangement in the local spatial layout map is unreasonable, the node spacing is dynamically adjusted and the boundary pressure coupling is reduced; the adjusted component parameters are re-entered into the structure generation network to perform local reconstruction and spatial layout reconfiguration. Optimize, obtain a new layout map, iteratively generate a new version of the digital twin model; and recalculate the feasibility index of the new version of the model until it is above the set safety threshold; if the tunnel feasibility index of the tunnel design digital twin model is higher than the preset tunnel feasibility index threshold, then take the second optimization measure, which is to analyze the design margin of the component based on the existing component layout map and parameter configuration, identify possible material over-matching, structural redundancy or layout density redundancy, and perform the following cost-saving and efficiency-enhancing operations: For redundant safety components (such as material grade exceeding the standard, component size far exceeding the minimum requirement), Appropriately reduce the concrete strength grade (such as C40 to C35), reduce the thickness of the additional layer or the number of components; optimize the spatial layout density, reduce the total number of construction nodes, and adjust the component sequence to adapt to the continuity of the construction process; introduce more efficient construction plan suggestions (such as changing the shield section to the open-cut section), and re-mark the construction process constraints; input the optimized parameters and spatial layout map into the modeling system again to generate a digital twin model version of the tunnel design with better economy and constructability; and retain the original version as the comparison baseline to achieve comparative simulation and deployment selection between different optimization levels.

[0065] In this implementation plan, by introducing a dual-path design optimization mechanism based on the tunnel feasibility index, the intelligent correction and economic improvement of the digital twin model of tunnel design are unifiedly controlled. Secondly, when the model evaluation results show that the feasibility index is lower than the preset threshold, the risk source is automatically identified, and the key influencing factors based on the construction risk index and the hydraulic action risk comprehensive index are located at the component level and path section, so as to dynamically adjust the parameters such as component material, size, and layout density, and then form local design modification suggestions and automatically iterate the optimization model to ensure that the overall design meets the construction and operation safety requirements. When the feasibility index is higher than the threshold, the efficiency improvement path is entered, such as identifying material redundancy, component configuration redundancy, and layout sequence redundancy in the model, and reducing costs and increasing efficiency by lowering the strength grade, optimizing the layout density, and adjusting the construction plan, while retaining the design safety margin. Finally, through the closed-loop design, evaluation, and optimization mechanism, the digital design is leaped from static visual modeling to dynamic adaptive optimization, so that it has strong engineering adaptability, safety controllability, and economic flexibility, thereby improving the tunnel design quality and implementation efficiency.

[0066] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0067] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A digital design method for water conservancy projects based on digital twins, characterized by: The following steps are involved: Obtain the design requirement information of the water diversion tunnel to be designed, and input it into the pre-trained building recognition model for comprehensive analysis, obtain the spatial layout map of the water diversion tunnel to be designed, and build a digital twin model of the tunnel design; and obtaining simulation data of the digital twin model of tunnel design, wherein the simulation data includes geological simulation data and hydrological simulation data; The simulation data of the tunnel design digital twin model are simulated and analyzed respectively to obtain the tunnel construction risk index and hydraulic action risk comprehensive index of the tunnel design digital twin model, and a comprehensive analysis is performed to obtain the tunnel feasibility index of the tunnel design digital twin model; Design optimization of the digital twin model of tunnel design based on the feasibility index.

2. The digital twin-based water conservancy project digital design method according to claim 1 is characterized in that: The design requirement information includes tunnel functional index requirements, preset spatial path coordinates and longitudinal slope range, and construction process constraints. The building recognition model is specifically a structure generation network, which includes an input layer, a parameter aggregation layer, a component classification reasoning layer, a size and material generation layer, and a spatial layout output layer.

3. The digital design method for water conservancy projects based on digital twins according to claim 2 is characterized in that: The specific steps for obtaining the spatial layout map of the water diversion tunnel to be designed are as follows: In the input encoding layer of the structure generation network, the design requirement information of the water diversion tunnel to be designed is encoded to obtain a multi-dimensional design input embedding vector set of the water diversion tunnel to be designed; In the parameter aggregation layer of the structure generation network, the multi-dimensional design input embedding vector set of the water diversion tunnel to be designed is processed by constraint relationship learning to obtain the design context semantic feature vector of the water diversion tunnel to be designed; In the component classification reasoning layer of the structure generation network, the design context semantic feature vector of the water diversion tunnel to be designed is classified into structural component types to obtain the adaptation score value of each structural component of the water diversion tunnel to be designed; In the size and material generation layer of the structure generation network, the adaptation score value of each structural component of the water diversion tunnel to be designed is processed by size reasoning and material parameter prediction to obtain a multi-dimensional structural parameter set of the water diversion tunnel to be designed; In the spatial layout output layer of the structure generation network, the multi-dimensional structural parameter set of the water diversion tunnel to be designed is processed by layout reasoning to obtain the spatial layout map of the water diversion tunnel to be designed.

4. The digital design method for water conservancy projects based on digital twins according to claim 1 is characterized in that: The geological simulation data includes a stratum heterogeneity index, an underground stress distribution index, and a surrounding rock stability index, and the specific steps of obtaining a tunnel construction risk index of a tunnel design digital twin model are as follows: Standardize the stratum heterogeneity index, underground stress distribution index, and surrounding rock stability index of the tunnel design digital twin model; The stratum heterogeneity index, underground stress distribution index and surrounding rock stability index of the standardized tunnel design digital twin model are comprehensively analyzed to obtain the tunnel construction risk index of the tunnel design digital twin model.

5. The digital design method for water conservancy projects based on digital twins according to claim 4 is characterized in that: The specific formula for calculating the tunnel construction risk index of the tunnel design digital twin model is as follows: ； in, Tunnel construction risk index for digital twin model design of tunnels, , , These are the stratum heterogeneity index, underground stress distribution index, and surrounding rock stability index of the digital twin model of tunnel design after standardization. , , , They are the heterogeneous adjustment coefficient, stress distribution adjustment coefficient, surrounding rock stability adjustment coefficient and interaction adjustment coefficient stored in the database.

6. The digital design method for water conservancy projects based on digital twins according to claim 1 is characterized in that: The hydrological simulation data includes hydraulic gradient value, water flow corrosion index, water flow kinetic energy value, and hydrological vulnerability index, and the specific steps of obtaining the comprehensive index of hydraulic action risk of the tunnel design digital twin model are as follows: The hydrological simulation data of the tunnel design digital twin model are comprehensively analyzed to obtain the hydrological simulation index set of the tunnel design digital twin model, including the hydraulic impact effect index and the hydraulic abrasion sensitivity index. A comprehensive analysis is also conducted on the hydrological simulation index set of the tunnel design digital twin model to obtain the comprehensive index of hydraulic action risk of the tunnel design digital twin model.

7. The digital twin-based water conservancy project digital design method according to claim 6 is characterized in that: The specific steps to obtain the hydrological simulation index set of the tunnel design digital twin model are as follows: Obtain the water flow value of the tunnel design digital twin model, and conduct a comprehensive analysis based on the hydraulic gradient value and the kinetic energy value to obtain the hydraulic impact effect index of the tunnel design digital twin model; The water flow shear stress index of the tunnel design digital twin model is obtained, and a comprehensive analysis is performed in combination with the water flow corrosion index and the hydrological vulnerability index to obtain the hydraulic abrasion sensitivity index of the tunnel design digital twin model.

8. The digital twin-based water conservancy project digital design method according to claim 6 is characterized in that: The specific formula for calculating the comprehensive index of hydraulic risk of the digital twin model of tunnel design is as follows: ； in, , , They are the comprehensive hydraulic risk index, hydraulic impact effect index, and hydraulic abrasion sensitivity index of the tunnel design digital twin model. , , , They are the impact effect adjustment coefficient, abrasion sensitivity adjustment coefficient, synergy adjustment coefficient, and risk coupling adjustment coefficient stored in the database. is the smoothing coefficient stored in the database.

9. The digital twin-based water conservancy project digital design method according to claim 1 is characterized in that: The specific formula for calculating the tunnel feasibility index of the tunnel design digital twin model is as follows: ； in, , , They are the tunnel feasibility index, tunnel construction risk index, and hydraulic action risk comprehensive index of the tunnel design digital twin model. , , They are the tunnel construction adjustment coefficient, hydraulic adjustment coefficient, and superposition adjustment coefficient stored in the data. , They are the tunnel construction coefficient and hydraulic action coefficient stored in the data, .

10. The digital twin-based water conservancy project digital design method according to claim 9 is characterized in that: The specific steps for design optimization of the tunnel design digital twin model based on the feasibility index are as follows: The tunnel feasibility index of the tunnel design digital twin model is compared with the preset tunnel feasibility index threshold value for judgment and analysis; If the tunnel feasibility index of the tunnel design digital twin model is lower than or equal to the preset tunnel feasibility index threshold, the first optimization measure is taken; If the tunnel feasibility index of the tunnel design digital twin model is higher than the preset tunnel feasibility index threshold, the second optimization measure is taken.

Images