Parameter inversion and updating method of digital twin model of long-distance water pipeline system

Abstract

The application discloses a parameter inversion and updating method of a long-distance pipeline water delivery system digital twin model, which couples an actual water delivery system, a BIM-GIS digital twin model, a numerical calculation model, a proxy model and field monitoring data, uses measured data generated in a construction stage and an operation and maintenance stage to invert key state parameters in the water delivery system which are difficult to directly measure or change over time, and feeds back the inversion results to the digital twin model and the numerical calculation model, so that dynamic synchronization of the digital model and the entity engineering state is realized. The application enables the water delivery system digital twin model to be continuously corrected according to changes in the entity engineering state, and improves the reliability of construction quality control, operation state identification, leakage risk judgment and operation and maintenance decision-making.

Description

Technical Field

[0001] This invention belongs to the field of information processing technology, specifically relating to a method for parameter inversion and updating of a digital twin model of a long-distance pipeline water conveyance system. Background Technology

[0002] With the continuous development of information and digital technologies in the field of water conservancy engineering, Building Information Modeling (BIM), Geographic Information System (GIS), numerical simulation, Internet of Things (IoT) monitoring, and digital twin technologies are increasingly being applied to the design, construction, and operation and maintenance management of pumping station projects, water conveyance projects, and related water conservancy facilities. By constructing three-dimensional information models and integrating operational monitoring data, water conveyance systems can, to a certain extent, achieve visualized representation of engineering components, spatial environment integration, operational status display, and collaborative management, which plays a positive role in improving engineering design quality, construction management efficiency, and operation and maintenance safety. However, for linear projects such as long-distance water conveyance systems, existing technologies still have the following limitations: (1) Existing BIM or BIM-GIS models mainly focus on the geometric representation and static attribute management of pipe sections, valves, pumps, supports and ancillary facilities. The model parameters are usually derived from the design phase data. After the water conveyance system enters the construction and operation and maintenance phase, parameters such as actual burial depth, installation deviation, backfill quality, interface status, pipe bottom support conditions, pipe roughness coefficient, actual valve opening degree and pump performance will change with construction conditions and operation time. However, existing digital models are difficult to absorb these measured information in a timely manner, which leads to the gradual deviation between the BIM model and the actual project status.

[0003] (2) Existing hydraulic calculation, transient flow calculation, pipe-soil coupling calculation, and structural safety analysis models are usually established independently of the BIM digital model. The geometric parameters, material parameters, boundary conditions, and operating conditions used in the calculations mostly rely on manual compilation and design assumptions. The pressure test data, settlement monitoring data, and pipeline deformation data generated during the construction phase, as well as the pressure, flow rate, water level, vibration, and leakage alarm data generated during the operation and maintenance phase, are difficult to use to correct the key parameters in the numerical calculation model, making it difficult for the calculation model to reflect the real operating conditions of the water conveyance system in a timely manner.

[0004] (3) Existing SCADA systems, sensor monitoring systems, hydraulic simulation software, and BIM platforms are usually independent of each other. Operational monitoring data is mostly displayed in the form of tables, curves, or alarm information, which can indicate abnormal pressure, abnormal flow, or leakage risks, but it is difficult to further identify the reasons for parameter changes behind the anomalies. For example, the system is unable to invert the actual state parameters such as changes in pipe roughness coefficient, abnormal local resistance, suspected leakage location, leakage volume, pump performance degradation, or structural stiffness degradation based on monitoring data, thus limiting the depth of application of digital twin models in risk diagnosis and predictive maintenance.

[0005] (4) Although high-precision numerical calculation can analyze the pressure distribution, water hammer pressure, pipeline deformation, settlement response and structural safety status of water conveyance system under different operating conditions, the calculation of the complete numerical model is time-consuming and cannot meet the rapid response requirements of construction verification and operation and maintenance diagnosis. Existing technologies lack a parameter inversion mechanism that combines numerical calculation samples, surrogate models and field monitoring data, and cannot achieve continuous correction between the digital model, numerical calculation model and physical entity of water conveyance system. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention provides a method for parameter inversion and updating of a digital twin model of a long-distance pipeline water conveyance system.

[0007] To achieve the above objectives, the present invention provides the following solution: A parameter inversion and update method for a digital twin model of a long-distance pipeline water conveyance system is proposed. This method is used for real-time status monitoring, key parameter inversion, dynamic model updating, and predictive operation and maintenance management of the water conveyance system. It couples the actual water conveyance system, BIM-GIS digital twin model, numerical calculation model, proxy model, and field monitoring data. Using measured data generated during the construction and operation and maintenance phases, it inverts key status parameters in the water conveyance system that are difficult to measure directly or change over time. The inversion results are then fed back to the digital twin model and the numerical calculation model, achieving dynamic synchronization between the digital model and the physical project status.

[0008] In this method, a BIM-GIS digital twin model is used to represent the spatial location, component attributes, topographic and geological environment, and operational functions of the water conveyance system's pipe sections, valves, pumps, interfaces, supports, and ancillary facilities. A numerical calculation model is used to simulate the pressure, flow rate, water hammer pressure, pipe deformation, interface displacement, and settlement response of the water conveyance system under various conditions, including normal water conveyance, emergency pump shutdown, rapid valve closure, localized leakage, pipe settlement, changes in foundation stiffness, and structural degradation. A proxy model is used to replace the time-consuming numerical calculation process and establish a rapid mapping relationship between key parameters and system responses. Field monitoring data is used to reverse-correct the actual state parameters during the construction and operation / maintenance phases. Through the synergistic effect of these models and data, the system can identify deviations between design assumptions and actual construction conditions, diagnose the causes of abnormal pressure, abnormal flow, leakage risks, and structural degradation during operation, and generate model update and operation / maintenance decision results. Unlike existing technologies that only perform BIM visualization, SCADA data binding, or standalone hydraulic simulation, this invention focuses on a closed-loop mechanism: "monitoring data drives parameter inversion, inversion parameters drive model updates, and updated models support risk assessment." This mechanism ensures that the digital twin model of the water conveyance system is not statically represented but continuously corrected as the actual water conveyance system's state changes. This improves the accuracy of construction quality verification, operational status identification, leakage risk location, structural degradation assessment, and maintenance decisions for the water conveyance system. Specifically, this includes: Step S1: Construct a multi-source data foundation for the water conveyance system; Step S2: Based on the multi-source data base of the water conveyance system, obtain the basic model of BIM-GIS fusion digital twin; Step S3: Generate a numerical calculation model of the water conveyance system based on the BIM-GIS integrated digital twin basic model; Step S4: Based on the numerical calculation model of the water conveyance system, construct a multi-condition sample library and train the surrogate model; Step S5: Perform key parameter inversion based on on-site monitoring data and the trained agent model; Step S6: Perform inversion parameter feedback and dynamic update of the digital twin model.

[0009] Preferably, step S1 includes: Collect multi-source data generated during the design, construction, and operation and maintenance phases of the water conveyance system. The multi-source data includes pipeline centerline coordinates, node elevations, pipe diameters, wall thicknesses, pump station parameters, valve parameters, actual burial depths, installation deviations, pressure test results, foundation settlement, pipeline deformation, pressure, flow rates, water levels, valve openings, pump operating status, and leakage alarm data. A unique code is assigned to each pipe section, valve, pump, interface, support, monitoring point, and numerical calculation unit. The correspondence between physical entities, BIM components, GIS spatial objects, sensor data, and numerical calculation units is established through a code mapping table.

[0010] Preferably, step S2 includes: A BIM model is established based on the centerline coordinates of the water pipeline, node elevations, pipe diameter, pipe material, valve location, pump station location, and ancillary facility parameters. Pipe sections, valves, pumps, manholes, supports, interfaces, and pressure regulating facilities are expressed as parametric components. Import GIS data such as digital elevation models, topography, roads, rivers, strata distribution, and groundwater levels along the water conveyance route. Through coordinate transformation and spatial registration, form a BIM-GIS integrated digital twin basic model to carry engineering parameters, monitoring data, calculation results, and inversion results.

[0011] Preferably, step S3 includes: Numerical calculation parameters such as pipe segment length, pipe diameter, node elevation, roughness coefficient, valve local resistance coefficient, pump station head, pipe material elastic modulus, pipe wall thickness, interface stiffness, soil elastic modulus, backfill parameters, pipe bottom support stiffness, groundwater level, and boundary conditions are extracted from the BIM-GIS integrated digital twin basic model. Based on the analysis objectives, hydraulic calculation models, transient flow calculation models, or pipe-soil coupling calculation models are established to analyze the pressure, flow rate, head loss, water hammer pressure, pipeline deformation, interface displacement, settlement response, and structural safety status of the water conveyance system under different operating conditions.

[0012] Preferably, step S4 includes: Based on the numerical calculation model, simulations are performed on various working conditions, including normal water conveyance, maximum flow rate, low flow rate, pump stoppage due to an accident, rapid valve closure, local leakage, pipeline settlement, reduced foundation stiffness, insufficient backfill quality, and pipeline structure degradation, forming a multi-condition calculation sample library. Using pipe diameter, flow rate, valve opening, pump station operating parameters, pipe roughness coefficient, local resistance coefficient, leakage location, leakage amount, foundation support stiffness, interface stiffness, and structural degradation coefficient as inputs, and nodal pressure, pipe section velocity, head loss, peak water hammer pressure, pipe deformation, interface displacement, settlement, leakage risk, and structural safety status as outputs, a surrogate model for water conveyance system is trained for rapid prediction and parameter inversion.

[0013] Preferably, step S5 includes: The system integrates data on actual burial depth, installation deviation, pressure test results, interface displacement, pipeline deformation, foundation settlement, and backfill quality during the construction phase, as well as pressure, flow rate, water level, valve opening, pump operating status, pipeline vibration, and leakage alarm data during the operation and maintenance phase. It compares the on-site monitoring response with the predicted response from the proxy model. When discrepancies exist, it aims to minimize the error and invert the pipe bottom support stiffness, backfill soil compression parameters, interface stiffness, pipeline roughness coefficient, local resistance coefficient, valve opening deviation, pump performance attenuation coefficient, suspected leakage location, leakage volume, and structural degradation coefficient.

[0014] As a preferred embodiment, S6 includes: The corrected parameters obtained from the inversion are fed back to the BIM-GIS digital twin model and numerical calculation model to update the extended attribute fields, material parameters, boundary conditions, pipe-soil action parameters and operating condition parameters of the corresponding components. When the inversion results exceed the preset threshold, the system generates backfill quality verification, local reinforcement, interface review, suspected leakage area investigation, valve adjustment, pump station operation optimization, maintenance route recommendation, or emergency valve closure plan, thus forming a closed-loop update relationship between the physical entity of the water transmission system, digital twin model, numerical calculation model, proxy model, and on-site monitoring data.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention utilizes on-site monitoring data from the construction and operation / maintenance phases to invert key state parameters such as pipe bottom support stiffness, backfill soil parameters, interface stiffness, pipe roughness coefficient, local resistance coefficient, suspected leakage, and structural degradation coefficient. This allows the model parameters to be corrected from design assumptions to actual parameters that are closer to the actual state of the physical project, thereby improving the accuracy of water conveyance system state identification and risk assessment.

[0016] 2. This invention feeds back the corrected parameters obtained from the inversion to the BIM digital model and the numerical calculation model, enabling the digital twin model to be dynamically updated as the construction and operation status of the physical water conveyance system changes. This avoids the digital model remaining at the design parameter or static display level for a long time, and reduces the deviation between the model and the actual engineering status.

[0017] 3. This invention constructs a training sample library through multi-condition numerical calculation and uses a surrogate model to replace the time-consuming numerical calculation process, enabling the system to quickly predict pressure distribution, flow rate changes, water hammer pressure, pipeline deformation, settlement response, leakage risk and structural safety status, thereby improving the response efficiency of construction verification and operation and maintenance diagnosis.

[0018] 4. This invention can identify the impact of construction deviations such as actual burial depth, backfill quality, interface status, pipe bottom support conditions, and foundation settlement on the safety status of the water conveyance system based on construction monitoring data, providing a basis for backfill quality verification, interface review, local reinforcement, and construction parameter correction, thereby improving the pertinence of quality control during the construction phase.

[0019] 5. This invention can identify operational problems such as abnormal pressure, abnormal flow, valve opening deviation, pump performance degradation, suspected leakage, and structural degradation based on operational monitoring data, and generate risk warning information or operation and maintenance solutions when the inversion results exceed a preset threshold, thereby improving the initiative and safety of water transmission system operation and maintenance. Attached Figure Description

[0020] To more clearly illustrate the technical solution of the present invention, the drawings used in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating the parameter inversion and update method for a digital twin model of a long-distance pipeline water conveyance system according to an embodiment of the present invention.

[0022] Figure 2 This is a flowchart illustrating the training and evaluation process of a proxy model based on a multi-condition sample library, as described in an embodiment of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0025] Example 1 like Figure 1As shown, this invention also provides a method for parameter inversion and updating of a digital twin model of a long-distance pipeline water conveyance system, used for real-time status monitoring, key parameter inversion, dynamic model updating, and predictive operation and maintenance management of the water conveyance system. This method couples the actual water conveyance system, the BIM-GIS digital twin model, the numerical calculation model, the proxy model, and field monitoring data. Using measured data generated during the construction and operation and maintenance phases, it inverts key state parameters in the water conveyance system that are difficult to measure directly or change over time, and feeds the inversion results back to the digital twin model and the numerical calculation model, achieving dynamic synchronization between the digital model and the physical project status. To achieve the above-mentioned objectives, this invention adopts the following technical solution: S1. Construction of a multi-source data base for the water conveyance system The system collects multi-source data generated during the design, construction, and operation and maintenance phases of the water conveyance system and establishes a unified data management foundation. Design foundation data includes pipeline planar path, longitudinal profile elevation, pipeline centerline coordinates, node elevations, pipe segment length, pipe diameter, pipe material, wall thickness, pipe material elastic modulus, Poisson's ratio, density, design roughness coefficient, joint type, joint stiffness, pump station head, design flow rate, pump speed, valve type, valve specifications, valve opening, pressure regulating facility parameters, support dimensions, and inspection well parameters. Geographic environment data includes digital elevation model, topography, roads, rivers, geological strata distribution, groundwater level, soil elastic modulus, compression modulus, unit weight, cohesion, and internal friction angle. The permeability coefficient; construction measurement data including construction and installation deviations, actual burial depth, interface displacement, interface closure status, pressure test results, backfill material type, backfill compaction degree, backfill quality inspection results, foundation settlement, pipeline deformation, and construction acceptance records; operation monitoring data including pressure monitoring data, flow monitoring data, water level data, pump start-up and shutdown status, pump unit operating power, valve opening, pipeline vibration data, leakage alarm data, and operation scheduling records; maintenance data including maintenance time, maintenance location, maintenance method, replaced components, post-maintenance status, and retest results. All of the above data collectively constitute the data foundation for digital twin modeling, numerical calculation, surrogate model training, and parameter inversion of the water conveyance system.

[0026] During the data import process, the system assigns unified codes to pipe sections, valves, pumps, interfaces, supports, manholes, monitoring points, and numerical calculation units. Each pipe section corresponds to a component code in the BIM model, a spatial line object in the GIS model, a pipe segment unit in the hydraulic calculation model, a structural calculation unit in the pipe-soil coupling calculation model, and several pressure, flow, settlement, or vibration monitoring points in the sensor system. The system establishes the correspondence between physical entities, BIM components, GIS spatial objects, sensor data, and numerical calculation units through a coding mapping table. Subsequent digital twin model construction, numerical calculation model generation, proxy model training, construction phase parameter inversion, operation and maintenance phase parameter inversion, and dynamic model updates are all based on this data foundation, thereby ensuring that data from different stages, different systems, and different models can be traced back to each other and used collaboratively.

[0027] Establishment of S2 and BIM-GIS integrated digital twin basic model After establishing a multi-source data foundation, the system builds a BIM-GIS integrated digital twin basic model based on the spatial location, component type, geometric dimensions, material properties, connection relationships, and operational functions of the physical entities of the water conveyance system. The system establishes the BIM model based on pipeline centerline coordinates, node elevations, pipe diameters, pipe materials, valve locations, pump station locations, manhole locations, support locations, and pressure regulating facility parameters. In the BIM model, pipe sections, valves, pumps, manholes, supports, interfaces, and pressure regulating facilities are all represented as identifiable parametric components. Each component includes geometric dimensions, material properties, equipment parameters, installation status, and a unique code. For pipe components, the extended attribute fields should include at least the design burial depth, actual burial depth, pipe diameter, wall thickness, pipe material type, roughness coefficient, pipe bottom support stiffness, backfill quality evaluation, settlement state, leakage risk level, and structural degradation coefficient. For valve components, the extended attribute fields should include at least the valve type, design opening degree, real-time opening degree, opening degree deviation, local resistance coefficient, and operating status. For pump components, the extended attribute fields should include at least the rated head, design flow rate, actual head, operating power, start-stop status, and pump set performance degradation coefficient.

[0028] Simultaneously, the system imports GIS data along the water conveyance route, including digital elevation models, topography, roads, rivers, geological strata, groundwater levels, and information on surrounding structures. Through coordinate transformation and spatial registration, the engineering components in the BIM model are integrated with their spatial locations in the GIS environment, ensuring that the water conveyance pipeline and its ancillary facilities correspond to the terrain, geology, and surrounding environmental conditions along the route. For pipeline sections crossing roads, rivers, soft strata, high embankment areas, high groundwater levels, or areas sensitive to foundation deformation, the system can set environmental sensitivity markers in the GIS environment and bind these markers to the corresponding BIM component codes. This BIM-GIS integrated digital twin basic model is not only used for 3D display and component attribute management but also for receiving construction process data, operational monitoring data, numerical calculation results, surrogate model prediction results, parameter inversion results, and risk assessment results. Thus, the BIM model is transformed from a static information model into a digital twin basic model that can continuously reflect the changing state of the physical water conveyance system.

[0029] S3. Generation of the numerical calculation model of the water conveyance system After the BIM-GIS integrated digital twin basic model is established, the system extracts the geometric parameters, material parameters, boundary conditions, and operating condition parameters required for numerical calculation from the model, and generates corresponding numerical calculation models of the water conveyance system according to different analysis objectives. The geometric parameters include pipeline centerline coordinates, node elevations, pipe segment lengths, pipe diameters, wall thicknesses, burial depths, interface locations, valve locations, pump station locations, and pressure regulating facility locations; the material parameters include pipe elastic modulus, Poisson's ratio, density, pipe roughness coefficient, interface stiffness, structural degradation coefficient, soil elastic modulus, compression modulus, backfill parameters, groundwater level, and pipe bottom support stiffness; the operating condition parameters include pump station head, flow rate, rotational speed, valve opening, local resistance coefficient, inlet head, outlet flow rate, design flow rate, maintenance conditions, emergency pump shutdown conditions, and valve closure conditions.

[0030] When the analysis objective is steady-state water conveyance capacity and pressure distribution, the system extracts pipe segment length, pipe diameter, node elevation, roughness coefficient, valve local resistance coefficient, pump station head, inlet head, outlet flow rate, and node boundary conditions to establish a hydraulic calculation model. This model is used to calculate node pressure, pipe segment velocity, and head loss under normal water conveyance, maximum flow rate, and low flow rate conditions. When the analysis objective is emergency pump shutdown, rapid valve closure, or pressure regulating facility verification, the system further extracts pipe material elastic modulus, pipe wall thickness, pump station start-up and shutdown curves, valve closing patterns, pressure regulating facility parameters, air valve parameters, and boundary water level conditions based on the hydraulic calculation model. A transient flow calculation model is established to analyze the pressure wave propagation process, water hammer pressure peak value, negative pressure risk, and the protective effect of pressure regulating facilities. When the analysis target is pipeline structural deformation, interface displacement, settlement response, and structural safety status, the system extracts pipeline burial depth, pipe material elastic modulus, interface stiffness, soil elastic modulus, backfill compression parameters, pipe bottom support stiffness, groundwater level, and foundation deformation conditions to establish a pipe-soil coupled calculation model or a structural safety calculation model. This model is used to analyze pipeline deformation, interface displacement, settlement response, and structural safety status under the effects of foundation settlement, changes in backfill quality, changes in pipe bottom support conditions, and structural degradation. During model generation, the system establishes a correspondence between numerical calculation units and BIM components through unified coding, enabling numerical calculation results to be written back to the digital twin model. The correction parameters obtained from subsequent inversion can also be synchronously updated to the numerical calculation model.

[0031] S4: Establishment of a multi-condition sample library and training of the surrogate model After the numerical calculation model is established, the system conducts multi-condition simulations based on the hydraulic calculation model, transient flow calculation model, and pipe-soil coupling calculation model to form a calculation sample library required for training the surrogate model. The multi-condition simulations include normal water conveyance, maximum flow, low flow, emergency pump shutdown, rapid valve closure, local leakage, pipeline settlement, foundation stiffness variation, backfill quality variation, and pipeline structure degradation. If necessary, pump performance degradation, abnormal local resistance, valve opening deviation, and different maintenance and scheduling conditions can also be added.

[0032] Each operating condition uses pipe diameter, flow rate, valve opening, pump station operating parameters, pipe roughness coefficient, local resistance coefficient, leakage location, leakage amount, foundation support stiffness, backfill parameters, interface stiffness, burial depth deviation, and structural degradation coefficient as input parameters, and nodal pressure, pipe section velocity, head loss, peak water hammer pressure, pipe deformation, interface displacement, settlement, leakage risk level, and structural safety status as output results. The system organizes the above multi-condition calculation results into a sample library, and performs outlier removal, missing value completion, dimension unification, normalization, and training, validation, and test set division on the sample data. For continuous outputs such as pressure, flow rate, head loss, peak water hammer pressure, pipe deformation, and settlement, a regression surrogate model can be established using the limit gradient boosting model; for discrete outputs such as leakage risk level and structural safety status, a classification surrogate model can be established using random forest, limit gradient boosting model, or classification neural network.

[0033] Furthermore, the specific training process for establishing a classification surrogate model using the Extreme Gradient Boosting (XGBoost) model is as follows: Input parameter feature matrices and corresponding target vectors are extracted from a multi-condition numerical computation sample library. Input parameters include pipe diameter, flow rate, valve opening, pump station operating parameters, pipe roughness coefficient, local resistance coefficient, leakage location, leakage amount, foundation support stiffness, backfill parameters, interface stiffness, burial depth deviation, and structural degradation coefficient, etc. Output results include leakage risk level and structural safety status, etc. After outlier removal, missing value completion, dimension unification, and normalization, the sample data is randomly divided into a training set and an independent test set at an 8:2 ratio. The training set is used for model training and parameter optimization, while the test set is used to test the model's generalization ability under unknown conditions. The core hyperparameters of the model mainly include the learning rate, the maximum depth of the decision tree, and the number of weak learners. The learning rate ranges from 0.01 to 0.30, the maximum depth of the decision tree ranges from 3 to 10, and the number of weak learners ranges from 50 to 500. The system employs a grid search combined with five-fold cross-validation for automated optimization. It compares the prediction performance of models under different hyperparameter combinations and selects the parameter combination with the smallest cross-validation error as the final model configuration. During training, XGBoost uses decision trees as base learners, updating the model round-by-round through additive ensemble. It utilizes the first and second gradients of the objective function to determine the weights of feature split points and leaf nodes, thereby continuously reducing classification loss and improving the model's ability to discriminate leakage risk levels and structural safety status. Model updates cease when the validation set loss stabilizes or the preset number of iterations is reached.

[0034] After model training, its classification performance is evaluated using an independent test set. For discrete outputs, metrics such as accuracy, precision, recall, F1 score, and confusion matrix can be used for evaluation. Accuracy reflects the overall proportion of correctly classified samples; precision reflects the proportion of samples correctly classified as a certain risk level or safe state; recall reflects the proportion of samples correctly identified in that category; and the F1 score, which considers both precision and recall, is suitable for situations where the number of samples at different risk levels is imbalanced. Its calculation formula can be expressed as: In the formula, TP represents the number of samples correctly identified as the target category; TN represents the number of samples correctly identified as non-target categories; FP represents the number of samples incorrectly identified as the target category; and FN represents the number of samples in the target category that were not correctly identified.

[0035] It should be noted that the above process uses a categorical surrogate model with discrete output as an example. For continuous outputs such as node pressure, pipe segment velocity, head loss, peak water hammer pressure, pipe deformation, interface displacement, and settlement, the same data processing, sample partitioning, cross-validation, and hyperparameter optimization process can be used to establish a regression surrogate model, and the prediction accuracy can be evaluated using indicators such as the coefficient of determination and root mean square error. To intuitively represent the surrogate model training process, the steps of sample construction, data preprocessing, dataset partitioning, hyperparameter optimization, model training, and performance evaluation can be organized into a flowchart, as shown in the following figure. Figure 2 As shown.

[0036] When the surrogate model's prediction error meets preset requirements, it is used as a rapid prediction model and parameter inversion tool. The surrogate model replaces the time-consuming complete numerical calculation process, enabling the system to quickly complete predictions for a large number of operating conditions, meeting the response speed requirements of construction verification and operation and maintenance diagnosis. Through this step, the system establishes a rapid mapping relationship between key parameters and the water conveyance system response, and can also be used to reverse-identify the true state parameters when the on-site monitoring response is known.

[0037] S5: Key Parameter Inversion Based on Field Monitoring Data During the construction phase, the system integrates field data such as pipeline installation deviations, actual burial depth, pressure test results, interface displacements, pipeline deformation, foundation settlement, and backfill quality. It then uses a surrogate model to invert key parameters for the construction period. The system first compares the field monitoring response with the surrogate model's predicted response based on design parameters. When the deviation is less than a preset threshold, it indicates that the current design parameters accurately reflect the actual construction status; when the deviation exceeds the preset threshold, the system initiates construction period parameter inversion. The construction period parameter inversion aims to minimize the error between the field monitoring response and the surrogate model's predicted response, identifying parameters such as pipe bottom support stiffness, backfill soil compression parameters, interface stiffness, foundation deformation parameters, boundary constraints, and actual burial status. For example, if the pressure test results for a certain pipe section meet the requirements, but subsequent settlement monitoring values ​​are significantly higher than the surrogate model's predicted values, the system can adjust the pipe bottom support stiffness and backfill soil compression parameters to gradually bring the surrogate model's predicted settlement closer to the measured settlement, thereby determining whether there are issues such as insufficient backfill quality, weak pipe bottom support, or excessive foundation deformation in that section. After the inversion is completed, the system will write the correction parameters into the extended attribute fields of the corresponding components in the BIM digital model, and simultaneously update the soil parameters, boundary conditions and pipe-soil action parameters in the numerical calculation model.

[0038] During the operation and maintenance phase, the system receives real-time monitoring data on pressure, flow rate, water level, valve opening, pump operating status, pipeline vibration, and leakage alarms. It then uses a surrogate model to invert operating status parameters and structural degradation parameters. The system compares the real-time monitoring response with the response predicted by the surrogate model based on the current model parameters. When abnormal pressure drops, increased upstream and downstream flow differences, mismatch between valve opening and flow response, decreased pump station output capacity, or continuous leakage alarm triggering occur, the system initiates operational parameter inversion. The operational parameter inversion aims to minimize the error between the surrogate model predictions and the field monitoring data, inverting parameters such as pipeline roughness coefficient, local resistance coefficient, actual valve opening deviation, pump performance degradation coefficient, suspected leakage location, leakage volume, pipeline structural stiffness degradation coefficient, and pipe-soil contact state change parameters. For example, when the upstream pressure of a certain pipe section is basically normal, the downstream pressure continues to decrease, and the difference in flow between the upstream and downstream increases significantly, the system can invert the suspected leakage location and leakage amount based on the surrogate model; when the output flow of a certain pump station continues to decrease under the same opening degree and the same operating conditions, the system can invert the pump set performance degradation coefficient; when the settlement of a certain section increases over a long period of time and the pipeline vibration is abnormal, the system can invert the changes in the pipe-soil contact state and the degradation of structural stiffness.

[0039] Parameter inversion can employ iterative search, genetic algorithms, particle swarm optimization, Bayesian optimization, or other optimization methods. After inversion, the system outputs the parameter combination that best matches the on-site monitoring status and associates this parameter combination with the corresponding BIM component code, monitoring point code, and numerical calculation unit code. When the inversion results show that a certain pipe section exhibits an abnormally increased roughness coefficient, abnormally decreased pressure, significant flow rate differences, increased leakage risk, or accelerated structural degradation, the system feeds back the abnormal information to the BIM digital model and visualizes it through color coding, label prompts, risk zoning, or alarm information.

[0040] S6: Inversion Parameter Feedback and Dynamic Update of Digital Twin Model After the parameter inversion is completed, the system feeds back the corrected parameters obtained from the construction and operation and maintenance phases to the BIM digital model, numerical calculation model, and on-site management process. For the BIM digital model, the system writes the actual burial depth, backfill quality evaluation results, pipe bottom support stiffness, interface stiffness, pipe roughness coefficient, local resistance coefficient, valve opening deviation, pump performance degradation coefficient, suspected leakage location, leakage amount, leakage risk level, settlement status, structural degradation coefficient, maintenance status, and safety evaluation results into the extended attribute fields of the corresponding components, transforming the BIM model from a static information model into a digital twin model that can reflect the real-time status of the physical project.

[0041] For the numerical calculation model, the system updates pipe parameters, interface parameters, soil parameters, boundary conditions, pipe-soil interaction parameters, and operating condition parameters based on the inversion results. The updated numerical calculation model is used for subsequent steady-state hydraulic analysis, water hammer analysis, pipe-soil coupling analysis, and structural safety evaluation, making the subsequent calculation results closer to the actual state of the water conveyance system. For on-site management, when the inversion results exceed a preset threshold, the system automatically generates construction adjustment suggestions or operation and maintenance handling suggestions. During the construction phase, when the inversion results indicate abnormal backfill soil compression parameters, insufficient pipe bottom support stiffness, or low interface stiffness in a certain section, the system generates suggestions for backfill quality verification, local reinforcement, interface review, or construction parameter correction. During the operation and maintenance phase, when the inversion results indicate a leakage risk in a certain pipe section, an abnormally increased roughness coefficient, abnormal local resistance, pump performance degradation, or accelerated structural degradation, the system generates suggestions for suspected leakage area investigation, valve adjustment, pump station operation optimization, maintenance route recommendations, or emergency valve closure plans.

[0042] During the design phase, the system performs scheme verification, operational condition analysis, and parameter optimization based on the BIM digital model and numerical calculation model. During the construction phase, the system corrects key construction parameters based on on-site monitoring data and the results of the proxy model inversion, identifying the impact of construction deviations on the system's safety status. During the operation and maintenance phase, the system continuously inverts state parameters and degradation parameters based on real-time operational data, identifies abnormal operating conditions, determines risk locations and impact ranges, and generates corresponding operation and maintenance response plans. Thus, a cyclical update relationship is formed between the physical entity of the water conveyance system, the BIM digital model, the numerical calculation model, the proxy model, and the on-site monitoring data, enabling the digital model to dynamically correct itself as the physical project status changes. Through the above implementation methods, the measured data generated during the construction and operation and maintenance phases are no longer limited to individual recording, report display, or alarm prompts, but can be transformed into a basis for model parameter correction. Therefore, the measured data generated during the construction and operation and maintenance phases can be transformed into a basis for model parameter correction, and the inverted parameters can be further fed back to the digital twin model and numerical calculation model, providing support for subsequent calculation analysis, risk identification, and response decisions.

[0043] Through the implementation of the above six stages, this invention, within a unified digital twin framework, achieves an information closed loop for long-distance water conveyance systems, encompassing data acquisition, model building, numerical calculation, proxy inversion, dynamic updates, and operation and maintenance decisions. This method relies on a BIM-GIS fusion model to complete spatial representation and component management, combines numerical calculation and proxy models to achieve rapid prediction and parameter inversion, and uses on-site monitoring data to drive continuous correction of model parameters, forming an intelligent management and control solution for water conveyance systems oriented towards construction verification, operational diagnosis, and predictive operation and maintenance.

[0044] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for parameter inversion and updating of a digital twin model of a long-distance pipeline water conveyance system, characterized in that, include: Step S1: Construct a multi-source data foundation for the water conveyance system; Step S2: Based on the multi-source data base of the water conveyance system, obtain the basic model of BIM-GIS fusion digital twin; Step S3: Generate a numerical calculation model of the water conveyance system based on the BIM-GIS integrated digital twin basic model; Step S4: Based on the numerical calculation model of the water conveyance system, construct a multi-condition sample library and train the surrogate model; Step S5: Perform key parameter inversion based on on-site monitoring data and the trained agent model; Step S6: Perform inversion parameter feedback and dynamic update of the digital twin model.

2. The parameter inversion and update method for the digital twin model of a long-distance pipeline water conveyance system as described in claim 1, characterized in that, Step S1 includes: Collect multi-source data generated during the design, construction, and operation and maintenance phases of the water conveyance system. The multi-source data includes pipeline centerline coordinates, node elevations, pipe diameters, wall thicknesses, pump station parameters, valve parameters, actual burial depths, installation deviations, pressure test results, foundation settlement, pipeline deformation, pressure, flow rates, water levels, valve openings, pump operating status, and leakage alarm data. A unique code is assigned to each pipe section, valve, pump, interface, support, monitoring point, and numerical calculation unit. The correspondence between physical entities, BIM components, GIS spatial objects, sensor data, and numerical calculation units is established through a code mapping table.

3. The parameter inversion and update method for the digital twin model of a long-distance pipeline water conveyance system as described in claim 2, characterized in that, Step S2 includes: A BIM model is established based on the centerline coordinates of the water pipeline, node elevations, pipe diameter, pipe material, valve location, pump station location, and ancillary facility parameters. Pipe sections, valves, pumps, manholes, supports, interfaces, and pressure regulating facilities are expressed as parametric components. Import GIS data such as digital elevation models, topography, roads, rivers, strata distribution, and groundwater levels along the water conveyance route. Through coordinate transformation and spatial registration, form a BIM-GIS integrated digital twin basic model to carry engineering parameters, monitoring data, calculation results, and inversion results.

4. The parameter inversion and update method for the digital twin model of a long-distance pipeline water conveyance system as described in claim 3, characterized in that, Step S3 includes: Numerical calculation parameters such as pipe segment length, pipe diameter, node elevation, roughness coefficient, valve local resistance coefficient, pump station head, pipe material elastic modulus, pipe wall thickness, interface stiffness, soil elastic modulus, backfill parameters, pipe bottom support stiffness, groundwater level, and boundary conditions are extracted from the BIM-GIS integrated digital twin basic model. Based on the analysis objectives, hydraulic calculation models, transient flow calculation models, or pipe-soil coupling calculation models are established to analyze the pressure, flow rate, head loss, water hammer pressure, pipeline deformation, interface displacement, settlement response, and structural safety status of the water conveyance system under different operating conditions.

5. The parameter inversion and update method for the digital twin model of a long-distance pipeline water conveyance system as described in claim 4, characterized in that, Step S4 includes: Based on the numerical calculation model, simulations are performed on various working conditions, including normal water conveyance, maximum flow rate, low flow rate, pump stoppage due to an accident, rapid valve closure, local leakage, pipeline settlement, reduced foundation stiffness, insufficient backfill quality, and pipeline structure degradation, forming a multi-condition calculation sample library. Using pipe diameter, flow rate, valve opening, pump station operating parameters, pipe roughness coefficient, local resistance coefficient, leakage location, leakage amount, foundation support stiffness, interface stiffness, and structural degradation coefficient as inputs, and nodal pressure, pipe section velocity, head loss, peak water hammer pressure, pipe deformation, interface displacement, settlement, leakage risk, and structural safety status as outputs, a surrogate model for water conveyance system is trained for rapid prediction and parameter inversion.

6. The parameter inversion and update method for a digital twin model of a long-distance pipeline water conveyance system as described in claim 5, characterized in that, Step S5 includes: The system integrates data on actual burial depth, installation deviation, pressure test results, interface displacement, pipeline deformation, foundation settlement, and backfill quality during the construction phase, as well as pressure, flow rate, water level, valve opening, pump operating status, pipeline vibration, and leakage alarm data during the operation and maintenance phase. It compares the on-site monitoring response with the predicted response from the proxy model. When discrepancies exist, it aims to minimize the error and invert the pipe bottom support stiffness, backfill soil compression parameters, interface stiffness, pipeline roughness coefficient, local resistance coefficient, valve opening deviation, pump performance attenuation coefficient, suspected leakage location, leakage volume, and structural degradation coefficient.

7. The parameter inversion and update method for a digital twin model of a long-distance pipeline water conveyance system as described in claim 6, characterized in that, S6 include: The corrected parameters obtained from the inversion are fed back to the BIM-GIS digital twin model and numerical calculation model to update the extended attribute fields, material parameters, boundary conditions, pipe-soil action parameters and operating condition parameters of the corresponding components. When the inversion results exceed the preset threshold, the system generates backfill quality verification, local reinforcement, interface review, suspected leakage area investigation, valve adjustment, pump station operation optimization, maintenance route recommendation, or emergency valve closure plan, thus forming a closed-loop update relationship between the physical entity of the water transmission system, digital twin model, numerical calculation model, proxy model, and on-site monitoring data.

8. The parameter inversion and update method for a digital twin model of a long-distance pipeline water conveyance system as described in claim 7, characterized in that, It is suitable for real-time status monitoring, key parameter inversion, dynamic model updating, and predictive operation and maintenance management of water conveyance systems.

Images