Pumped storage power station rock mass engineering parameter identification method and system
Patent Information
- Application Number
- CN202611182200.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-05
- Publication Date
- 2026-11-03
AI Technical Summary
Existing rock mass engineering safety analysis methods are insufficient to reflect the rheological characteristics of rock masses during the operation of pumped storage power stations while maintaining consistency with geomechanical mechanisms, and lack a unified framework for rheological deformation analysis and parameter identification of various types of rock mass engineering units.
A dual-driven approach based on mechanistic data was adopted, combining a geomechanical numerical model with operational monitoring data. Key mechanical parameters of rock mass rheological deformation behavior were dynamically identified through Bayesian inference and Markov chain Monte Carlo methods, and a surrogate model was constructed for verification analysis.
It enables dynamic identification of rock mass rheological deformation behavior under the constraints of geomechanical mechanisms, improves the utilization efficiency of monitoring data, reduces computational costs, and supports unified rheological deformation analysis of multiple types of rock mass engineering units.
Smart Images
Figure CN123030872A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of safety monitoring and analysis technology for water conservancy and hydropower engineering and rock mass engineering, and in particular to a method and system for identifying rock mass engineering parameters of pumped storage power stations based on dual-driven mechanism data. Background Technology
[0002] As a crucial infrastructure for large-scale peak shaving and energy storage, pumped-storage power stations operate under complex conditions where their rock mass engineering units, such as slopes, high-pressure waterways, and underground powerhouses, alternate between pumping and power generation. Influenced by factors such as periodic water level fluctuations, changes in operating loads, and complex rock mass structures, the rock mass exhibits not only instantaneous mechanical responses during operation but may also undergo time-dependent deformation, leading to cumulative deformation, stress redistribution, and continuous changes in local stability conditions, potentially impacting the safety of the power station's operation. Therefore, how to continuously monitor and uniformly analyze the state of different rock mass engineering units during operation is a key issue in the safety management of pumped-storage power stations.
[0003] Existing methods for rock mass engineering safety analysis mainly include mechanistic analysis methods based on geomechanical numerical models and empirical or statistical analysis methods based on monitoring data. The former usually relies on mechanical parameters determined during the design or initial exploration stage, making it difficult to reflect the rheological properties of the rock mass and the evolution of its parameters during operation, and its computational efficiency is low. The latter, although able to reflect the actual response of the project, lacks geomechanical mechanism constraints, making it difficult to reasonably interpret long-term deformation data and support rheological deformation prediction analysis during the operation stage.
[0004] In recent years, some studies have attempted to combine numerical models with monitoring data to improve analytical accuracy through parameter identification or model calibration. However, these studies mostly target single engineering units or single operating conditions, lacking a unified framework for rheological deformation analysis and parameter identification applicable to various rock mass engineering units such as engineering slopes, high-pressure waterways, and underground powerhouses. Furthermore, achieving efficient utilization of monitoring data and integrated processing of operational phase analysis and prediction while ensuring consistency in geomechanical mechanisms still requires further research.
[0005] Therefore, there is an urgent need for a method and system that can integrate numerical models and field monitoring data to uniformly identify and analyze the rheological deformation behavior and key parameters of different rock mass engineering units during the operation of pumped storage power stations, while maintaining the consistency of geomechanical mechanisms. Summary of the Invention
[0006] The main objective of this invention is to provide a method for identifying rock mass engineering parameters in pumped-storage power stations based on a dual-driven approach of mechanistic data. By integrating numerical models and operational monitoring data under the constraints of geomechanical mechanisms, this method achieves dynamic identification and uncertainty characterization of key mechanical parameters controlling the rheological deformation behavior of rock masses. Based on this, verification analysis and trend prediction of the rheological deformation response of rock masses during the operational phase are carried out. This overcomes the problems of static parameter values, difficulty in reflecting the time-dependent deformation characteristics of rock masses, and insufficient efficiency and reliability of analysis during the operational phase in the current method.
[0007] To achieve the above objectives, the present invention provides the following technical solution: On the one hand, the present invention provides a method for identifying rock mass engineering parameters of pumped storage power stations based on dual-driven mechanistic data, comprising the following steps: S1) Based on engineering geological data, construct a geomechanical numerical model and determine the a priori parameter range of rheology-related rock mass mechanics parameters; S2) Within the range of the prior parameters, for the pumping operation and power generation operation of the pumped storage power station, perform multi-set parameter combination calculations on the geomechanical numerical model to obtain the corresponding rock mass rheological deformation response results and form a training dataset for proxy modeling. S3) Based on the training dataset, construct a data-driven surrogate model to characterize the rheological deformation response characteristics of rock mass engineering under different parameter conditions and different operating conditions; S4) Introduce on-site monitoring data during the operation of the pumped storage power station, conduct joint analysis of the data-driven proxy model and the on-site monitoring data, identify the key mechanical parameters controlling the rheological deformation behavior of the rock mass, and obtain the posterior distribution of parameters reflecting the true working state of the rock mass. S5) Introduce the posterior distribution of the parameters into the geomechanical numerical model, and conduct comparative verification analysis on the mechanical response results of the rock mass engineering in the current operation stage under pumping or power generation conditions. S6) Based on the validated geomechanical numerical model, the rock mass engineering response during the operation of a pumped storage power station is predicted.
[0008] Preferably, the engineering geological data includes at least stratigraphic structure information, rock mass zoning results, structural features, and underground powerhouse spatial layout information; the rock mass mechanical parameters include rheological deformation parameters and strength parameters.
[0009] Preferably, the training dataset consists of at least one of the displacement response, stress response, plastic zone distribution, or seepage response calculated by the geomechanical numerical model under different parameter combinations and different operating conditions.
[0010] Preferably, the data-driven proxy model is used to establish a mapping relationship between rock mass rheological parameters and rock mass engineering rheological deformation response.
[0011] Preferably, the field monitoring data includes operational deformation monitoring data, stress monitoring data, seepage pressure monitoring data, or a combination thereof, reflecting the characteristics of rock mass changes over time.
[0012] Preferably, in step S4, the surrogate model and monitoring data are jointly analyzed using Bayesian inference principles and Markov chain Monte Carlo methods.
[0013] Preferably, the process of obtaining the posterior distribution characteristics of the parameters is as follows: Based on the Bayesian inference principle, and combining the prior parameter distribution with field monitoring data, the mechanical parameters controlling the rheological deformation behavior of the rock mass are identified to obtain the posterior distribution characteristics of the parameters; the posterior distribution of the parameters satisfies the following relationship:
[0014] in, This represents the a priori distribution of rock mass mechanical parameters. This represents the likelihood function of the monitored data under given parameter conditions. This represents the posterior distribution of the identified parameters.
[0015] Preferably, the Markov chain Monte Carlo method is used to sample and calculate the posterior distribution characteristics of the parameters to obtain a posterior sample set of rock mechanics parameters, and the mean, variance, confidence interval and correlation characteristics of the parameters are determined based on the posterior sample set.
[0016] The theoretical basis for parameter identification in this invention is the Bayesian inference principle (mathematical theorem), which constructs a posterior distribution of parameters based on prior parameter distributions and field monitoring data. The Markov Chain Monte Carlo (MCMC) method is a specific sampling algorithm used to solve for this posterior distribution. The relationship between the two is that Bayesian inference serves as the theoretical framework, while MCMC serves as the implementation method for posterior distribution calculation and sampling.
[0017] Preferably, the comparative verification analysis in step S5 includes comparing the consistency of the calculation results of the geomechanical numerical model with the actual field observation results to evaluate the rationality of the rock mass rheological deformation response in the current operating stage.
[0018] Preferably, the rock mass engineering response prediction includes predictive analysis of the rheological deformation response of the rock mass during the operation phase and its related stress evolution characteristics.
[0019] To achieve the above objectives, on the other hand, the present invention provides a rock mass engineering parameter identification system for pumped storage power stations based on dual-driven mechanistic data.
[0020] The system includes: The geological modeling module is used to construct geomechanical numerical models of slopes, high-pressure waterways, or underground powerhouses of pumped storage power station projects based on engineering geological data, and to set the prior range of rheology-related rock mechanics parameters. The proxy modeling module is used to construct a data-driven proxy model based on the calculation results of the numerical model, so as to approximately express the mapping relationship between parameters and rock mass rheological deformation response; The monitoring data processing module is used to acquire and process on-site monitoring data reflecting the characteristics of rock mass changes over time during the operation of the pumped storage power station. The parameter identification module is used to construct the posterior distribution of parameters based on the Bayesian inference principle, and to sample the posterior distribution of mechanical parameters that control the rheological deformation behavior of rock mass using the Markov chain Monte Carlo method. The state analysis and prediction module is used to input the posterior parameter samples into the geomechanical numerical model or surrogate model to analyze and predict the rheological deformation response state and its development trend during the operation of rock mass engineering.
[0021] Compared with the prior art, the beneficial effects of the present invention include: (1) This invention integrates numerical models and operational monitoring data under the constraints of geomechanical mechanisms to achieve dynamic identification of mechanical parameters that control the rheological deformation behavior of rock mass during the operational phase, overcoming the problem that the parameters in existing methods are static and difficult to reflect the time-dependent deformation characteristics of rock mass.
[0022] (2) This invention introduces data-driven proxy and Bayesian inference methods, which improves the utilization efficiency of monitoring data during operation in rheological deformation analysis while maintaining the physical constraints of the geomechanical model, and significantly reduces the computational cost of repeated numerical calculations during operation.
[0023] (3) The present invention constructs an integrated analysis process that combines rheological parameter identification, model verification and rheological deformation response analysis, which can perform unified analysis on the rheological deformation behavior of different rock mass engineering units such as engineering slopes, high-pressure waterways and underground powerhouses during operation.
[0024] (4) The method of the present invention has good versatility and scalability. It can be flexibly configured according to the structural characteristics and monitoring conditions of different engineering units, providing technical support for the long-term deformation control and operation management of rock mass engineering during the operation phase of pumped storage power stations.
[0025] (5) The present invention uses the comparison between the rock mass deformation response obtained by recalculating the identified parameters and the field monitoring data to illustrate that the parameter identification results can improve the consistency between the model response and the measured data. Attached Figure Description
[0026] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be 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 the structures shown in these drawings without creative effort.
[0027] Figure 1 This is an overall flowchart of the method for identifying rock mass engineering parameters of pumped storage power stations based on dual-drive mechanism data according to the present invention. Figure 2 This is a schematic diagram of the overall process of the method for identifying rock mass engineering parameters of pumped storage power stations based on dual-drive mechanism data according to the present invention. Figure 3 A schematic diagram of the numerical model of the rheological geomechanics of the engineering slope and the layout of monitoring points; Figure 4 A schematic diagram comparing the a priori distribution of rock mass rheological deformation control parameters with field monitoring data; Figure 5 For the posterior distribution of the parameters; Figure 6 This is a schematic diagram of the output of the proxy model and the geomechanical numerical model provided by the present invention; Figure 7 This is a schematic diagram showing the results of analyzing the evolution of the rheological deformation response of an engineering slope during its operation based on posterior rheological parameters. Detailed Implementation
[0028] The following section uses a pumped storage power station slope as an example to specifically illustrate the proposed method for identifying rock mass engineering parameters of pumped storage power stations based on dual-driven mechanistic data. Those skilled in the art should understand that this invention is also applicable to rock mass engineering units such as high-pressure waterways and underground powerhouses in pumped storage power stations. The technical principles, implementation processes, and analysis methods remain consistent, with differences only in engineering structural forms, boundary conditions, and monitoring arrangements.
[0029] Example 1: Rock Mass Parameter Identification Method for Slopes in Pumped Storage Power Station Projects In this embodiment 1, the slope of a pumped storage power station is selected as the research object. During the operation of the power station, the slope is affected by the periodic rise and fall of the reservoir water level, the switching of operating conditions, and the rock mass structure conditions. The mechanical state of the surrounding rock changes over time, exhibiting typical time-dependent deformation and deformation accumulation characteristics, and has representative rock mass rheological deformation behavior.
[0030] (I) Geomechanical Model Construction and Prior Parameter Setting First, engineering geological data of the engineering slope is obtained, including information on topography, stratigraphic structure, lithological distribution, structural features, and boundary conditions. Based on this, a geomechanical numerical model is established to reflect the geometric morphology, layered structure, mechanical boundary conditions, and time-dependent deformation characteristics of the rock mass of the engineering slope. Monitoring locations corresponding to those in the actual project are then placed within the model to extract the rheological deformation response of the engineering slope during operation.
[0031] Based on the engineering survey results and design data, the a priori value ranges of rock mass elastic modulus, Poisson's ratio, cohesion, internal friction angle, and other relevant parameters controlling the rheological deformation behavior of the rock mass are determined, and a priori parameter distribution is constructed to characterize the uncertainty of rock mass rheological parameters during the operation phase.
[0032] (II) Sample Data Generation and Proxy Model Construction Within the range of the prior parameters, rock mechanics parameters are sampled in combination for pumping and power generation operation conditions of the pumped storage power station. The sampled parameter sets are then input into the geomechanical numerical model for multi-parameter combination calculation to obtain the corresponding engineering response data. The engineering response data reflects the rheological deformation response of the engineering slope over time, including time history data of monitoring point displacement, thereby forming a sample dataset for training the surrogate model.
[0033] The geomechanical numerical model can be abstractly represented as a mapping relationship between parameters and engineering responses, and its mathematical form is as follows:
[0034] in, This is a vector of rock mass mechanics parameters. This represents the response mapping function composed of a geomechanical numerical model or a data-driven surrogate model. This represents the corresponding rock mass rheological deformation response. This represents the observation error or model error term.
[0035] Based on the sample dataset, a data-driven proxy model is constructed to approximate the mapping relationship between parameters and responses, thereby replacing costly repetitive numerical calculations and improving the efficiency of analysis during the runtime phase.
[0036] Specifically, a data-driven proxy model is constructed to characterize the rheological deformation response characteristics of rock mass engineering under different parameter conditions and different operating conditions; (III) Monitoring data introduction and rheological parameter identification During the operation of the pumped storage power station, on-site monitoring data of the engineering slope is acquired. This monitoring data reflects the deformation monitoring characteristics of the rock mass over time, including time history data of displacement at monitoring points. The Markov chain Monte Carlo method is used to correlate the monitoring data with the output results of a data-driven surrogate model, constructing the data foundation required for rheological parameter identification.
[0037] Based on Bayesian inference principles, the prior distribution of rheology-related rock mechanics parameters is used as initial constraints, and a data-driven surrogate model is used as a fast computation model for the parameters to the rock mass rheological deformation response. For each set of candidate parameters generated by the Markov chain Monte Carlo method, the candidate parameters are input into the data-driven surrogate model to obtain the corresponding predicted rock mass rheological deformation response. The predicted response is then compared with field deformation monitoring data to construct the likelihood function corresponding to the candidate parameter. Based on the likelihood function and the prior distribution of the parameters, the posterior probability of the candidate parameters is calculated, and the candidate parameters are accepted or rejected according to the acceptance criteria of the Markov chain Monte Carlo method, thereby achieving iterative updates of the rheology-related rock mechanics parameters. The posterior distribution of the parameters satisfies the following relationship:
[0038] in, This represents the a priori distribution of rheology-related rock mechanics parameters. Indicates that under given parameters Under the given conditions, the probability of consistency between the rock mass rheological deformation response calculated by the data-driven surrogate model and the field deformation monitoring data is... To incorporate the updated posterior distribution of parameters based on on-site deformation monitoring data.
[0039] Through the above process, the Markov chain Monte Carlo method continuously generates candidate parameters in the parameter space and updates the parameters by combining the deviation between the output of the surrogate model and the field deformation monitoring data. Finally, a set of parameter samples that conforms to the characteristics of the posterior probability distribution is obtained to characterize the uncertainty of rock mass rheological parameters during the operation phase.
[0040] (iv) Model verification and rheological deformation response analysis during operation The parameter sample set is input into a geomechanical numerical model or surrogate model to calculate the rheological deformation response of the engineering slope during the operation phase. The calculation results are then compared and analyzed with the field deformation monitoring data to verify the model's ability to characterize the actual engineering rheological deformation behavior under the condition of rheological parameter identification.
[0041] Based on the validated geomechanical numerical model or dual data-driven surrogate model, the rock mass engineering response during the operation phase of the pumped storage power station is predicted, and the long-term rheological deformation development trend of the rock mass engineering is analyzed.
[0042] Based on this, the rheological parameter samples are used to analyze the changes and development trends of the rheological deformation response during the operation of the engineering slope, so as to characterize the long-term deformation characteristics of the rock mass under the action of the operating conditions and provide a basis for deformation control and engineering management during the operation phase.
[0043] Example 2: Parameter Identification System for Rock Mass Engineering of Underground Powerhouse in Pumped Storage Power Station In this embodiment 2, a parameter identification system for rock mass engineering in pumped storage power stations is provided. The system can identify the mechanical parameters controlling the rheological deformation behavior of the rock mass in underground powerhouse rock mass engineering objects by combining engineering geological data, geomechanical numerical models, data-driven surrogate models, and on-site monitoring data during operation, and obtain the posterior distribution characteristics of the parameters. The system includes a mechanism model construction module, a training data generation module, a surrogate model construction module, a monitoring data import module, a parameter identification module, a verification analysis module, and a prediction and state analysis module.
[0044] (I) Mechanism Model Construction Module The mechanism model construction module is used to obtain engineering geological data of the underground powerhouse of the pumped storage power station, and to construct a geomechanical numerical model based on the engineering geological data that reflects the rock mass structure characteristics, stratum distribution, boundary conditions and time-dependent deformation characteristics of the rock mass.
[0045] The engineering geological data shall include at least one or more of the following: topography, stratigraphic structure, lithological distribution, rock mass zoning results, structural features, boundary conditions, underground powerhouse spatial layout information, high-pressure waterway layout information, and engineering operation status information.
[0046] The mechanism model construction module is also used to determine the a priori range of rock mass mechanics parameters based on engineering survey results, design data, and rock mechanics test results, and to construct the a priori distribution of parameters. The rock mass mechanics parameters include elastic modulus, Poisson's ratio, cohesion, internal friction angle, and related parameters controlling the rheological deformation behavior of the rock mass.
[0047] (ii) Training data generation module The training data generation module is used to sample and combine the rock mechanics parameters within the a priori value range of the rock mechanics parameters, and input the parameter combinations into the geomechanics numerical model for multi-set parameter combination calculations for pumped storage power station pumping operation, power generation operation, or periodic water level change operation, to obtain the corresponding rock mass engineering response data and generate a sample dataset for surrogate model training.
[0048] The sample dataset consists of engineering response results calculated by a geomechanical numerical model under different parameter combinations and operating conditions. The engineering response results include at least one of the following: displacement response, stress response, plastic zone distribution, seepage response, or time history response data of monitoring points.
[0049] (III) Proxy Model Construction Module The proxy model construction module is used to construct a data-driven proxy model based on the sample dataset to approximately express the mapping relationship between rock mass mechanical parameters and rock mass rheological deformation response in the geomechanical numerical model.
[0050] The surrogate model is used to characterize the rheological deformation response characteristics of the rock mass engineering of pumped storage power stations under different parameter conditions and different operating conditions, and is used to replace the high-cost repetitive numerical calculations during the operation phase, thereby improving the efficiency of parameter identification and engineering response analysis.
[0051] The mapping relationship between the rock mass mechanical parameters and the rock mass rheological deformation response can be expressed as:
[0052] in, This is a vector of rock mass mechanics parameters. This represents the response mapping function composed of a geomechanical numerical model or a data-driven surrogate model. This represents the corresponding engineering response quantity of rock mass rheological deformation. This represents the observation error or model error term.
[0053] (iv) Monitoring data import module The monitoring data import module is used to acquire on-site monitoring data during the operation phase of the pumped storage power station. This on-site monitoring data is used to characterize the engineering response features of the rock mass over time, including deformation monitoring data, stress monitoring data, seepage pressure monitoring data, or a combination thereof during operation.
[0054] The monitoring data import module is also used to compare the field monitoring data with the output results of the geomechanical numerical model or surrogate model to form error evaluation information for parameter identification. This error evaluation information is used to construct the likelihood function in Bayesian inference, providing data constraints for subsequent calculations of the posterior distribution of rock mass mechanics parameters.
[0055] (v) Parameter Identification Module The parameter identification module is used to identify the mechanical parameters that control the rheological deformation behavior of rock mass based on the Bayesian inference principle, combining the prior distribution of rock mass mechanical parameters with field monitoring data, and obtaining the posterior distribution of rock mass mechanical parameters.
[0056] The posterior distribution of the parameters satisfies the following relationship:
[0057] in, This represents the a priori distribution of rock mass mechanical parameters. This represents the likelihood function of field monitoring data under given parameter conditions. This represents the posterior distribution of the identified parameters. This indicates on-site monitoring data.
[0058] The parameter identification module employs the Markov chain Monte Carlo method to numerically sample the posterior distribution of the parameters, obtaining a parameter sample set that conforms to the characteristics of the posterior probability distribution. This parameter sample set is used to characterize the uncertainty of rock mass mechanics parameters during the operational phase and can be used to determine the parameter mean, variance, confidence interval, and main distribution range.
[0059] The Markov chain Monte Carlo method is not used to build surrogate models, nor is it used directly for ordinary correlation analysis between surrogate models and monitoring data. Instead, it is used to sample and calculate the posterior distribution of parameters within a Bayesian inference framework.
[0060] (vi) Verification and Analysis Module The verification and analysis module is used to input the posterior distribution of parameters or the parameter sample set obtained by the parameter identification module into the geomechanical numerical model or surrogate model to calculate the rheological deformation response of the underground powerhouse of the pumped storage power station in the current operating stage.
[0061] The verification and analysis module is also used to compare and analyze the calculated rock mass engineering response results with the actual field monitoring results to verify the ability of the geomechanical numerical model or surrogate model to characterize the actual engineering rheological deformation behavior under parameter identification conditions.
[0062] (vii) Prediction and Status Analysis Module The prediction and state analysis module is used to calculate and analyze the rheological deformation response of rock mass and its development trend during the operation of pumped storage power stations based on the validated geomechanical numerical model or surrogate model.
[0063] The prediction and state analysis module is also used to analyze the rheological deformation response of rock mass engineering under pumping operation, power generation operation or periodic water level changes by using a set of parameter posterior samples. This is used to characterize the long-term deformation characteristics of rock mass under operating conditions and to provide a basis for deformation control and engineering management during the operation phase.
[0064] Example 3: Rock Mass Engineering Parameter Identification System for High-Pressure Water Channels of Pumped Storage Power Stations In this embodiment 3, a method for identifying rock mass engineering parameters for high-pressure waterways in pumped storage power stations is provided. The system is used to identify the mechanical parameters of the surrounding rock in high-pressure waterways during operation. It can combine engineering geological data, geomechanical numerical models, data-driven surrogate models, and on-site monitoring data during operation to identify the mechanical parameters controlling the rheological deformation behavior of the surrounding rock in high-pressure waterways and obtain the posterior distribution characteristics of the parameters. The method includes a mechanism model construction module, a training data generation module, a surrogate model construction module, a monitoring data import module, a parameter identification module, a verification analysis module, and a prediction and state analysis module.
[0065] (I) Mechanism Model Construction Module The mechanism model construction module is used to obtain engineering geological data of the high-pressure waterway of the pumped storage power station, and to construct a geomechanical numerical model based on the engineering geological data to reflect the structural characteristics, stratigraphic distribution, boundary conditions, groundwater conditions and time-dependent deformation characteristics of the surrounding rock of the high-pressure waterway.
[0066] The engineering geological data shall include at least one or more of the following: high-pressure waterway axis layout, burial depth, tunnel diameter, lithological distribution, surrounding rock zoning results, fault fracture zones, joint and fissure development characteristics, initial geostress conditions, groundwater conditions, lining structure parameters, internal water pressure boundary conditions, and operating condition information.
[0067] The mechanism model construction module is also used to determine the a priori range of mechanical parameters of the surrounding rock in high-pressure waterways based on engineering survey results, design data, and rock mechanics test results, and to construct the a priori distribution of parameters. These surrounding rock mechanical parameters include elastic modulus, Poisson's ratio, cohesion, internal friction angle, and related parameters controlling the rheological deformation behavior of the surrounding rock.
[0068] (ii) Training data generation module The training data generation module is used to sample and combine the mechanical parameters of the surrounding rock of the high-pressure waterway within the a priori value range of the surrounding rock mechanical parameters. For pumped storage power station pumping operation, power generation operation, shutdown operation, or periodic change of internal water pressure, the parameter combination is input into the geomechanical numerical model to perform multi-set parameter combination calculations to obtain the corresponding high-pressure waterway surrounding rock engineering response data and generate a sample dataset for surrogate model training.
[0069] The sample dataset consists of engineering response results calculated by a geomechanical numerical model under different parameter combinations and operating conditions. The engineering response results include at least one of the following: high-pressure waterway surrounding rock displacement response, tunnel wall convergence deformation, lining stress response, distribution of surrounding rock plastic zone, seepage pressure response, contact pressure response, or time history response data of monitoring points.
[0070] (III) Proxy Model Construction Module The proxy model construction module is used to construct a data-driven proxy model based on the sample dataset to approximately express the mapping relationship between the mechanical parameters of the surrounding rock in the high-pressure waterway and the rheological deformation response of the surrounding rock in the geomechanical numerical model.
[0071] The surrogate model is used to characterize the rheological deformation response characteristics of the surrounding rock of high-pressure waterways under different parameter conditions and different operating conditions, and is used to replace the high-cost repetitive numerical calculations during the operation phase, thereby improving the efficiency of parameter identification and engineering response analysis of the surrounding rock of high-pressure waterways.
[0072] The mapping relationship between the mechanical parameters of the surrounding rock in the high-pressure waterway and the rheological deformation response of the surrounding rock can be expressed as:
[0073] in, This is a vector of mechanical parameters of the surrounding rock in a high-pressure waterway. This represents the response mapping function composed of a geomechanical numerical model or a data-driven surrogate model. This represents the corresponding engineering response quantity of the rheological deformation of the surrounding rock in the high-pressure waterway. This represents the observation error or model error term.
[0074] (iv) Monitoring data import module The monitoring data import module is used to acquire on-site monitoring data during the operation phase of the high-pressure waterway of the pumped storage power station. The on-site monitoring data is used to characterize the engineering response features of the surrounding rock and lining structure of the high-pressure waterway over time, including surrounding rock deformation monitoring data, tunnel wall convergence monitoring data, lining stress monitoring data, anchor bolt or anchor cable stress monitoring data, seepage pressure monitoring data, or a combination thereof.
[0075] The monitoring data import module is also used to compare the field monitoring data with the output results of the geomechanical numerical model or surrogate model to form error evaluation information for parameter identification. This error evaluation information is used to construct the likelihood function in Bayesian inference, providing data constraints for subsequent calculations of the posterior distribution of the surrounding rock mechanical parameters in the high-pressure waterway.
[0076] (v) Parameter Identification Module The parameter identification module is used to identify the mechanical parameters that control the rheological deformation behavior of the surrounding rock of the high-pressure waterway based on the Bayesian inference principle, combining the prior distribution of mechanical parameters of the surrounding rock with field monitoring data, and obtaining the posterior distribution of mechanical parameters of the surrounding rock of the high-pressure waterway.
[0077] The posterior distribution of the parameters satisfies the following relationship:
[0078] in, This represents the prior distribution of the mechanical parameters of the surrounding rock in a high-pressure waterway. This represents the likelihood function of field monitoring data under given parameter conditions. This represents the posterior distribution of the identified parameters. This indicates on-site monitoring data for the high-pressure waterway.
[0079] The parameter identification module employs the Markov chain Monte Carlo method to numerically sample the posterior distribution of the parameters, obtaining a parameter sample set that conforms to the characteristics of the posterior probability distribution. This parameter sample set is used to characterize the uncertainty of the surrounding rock mechanical parameters of the high-pressure waterway during operation and can be used to determine the parameter mean, variance, confidence interval, and main distribution range.
[0080] (vi) Verification and Analysis Module The verification and analysis module is used to input the posterior distribution of parameters or the parameter sample set obtained by the parameter identification module into the geomechanical numerical model or surrogate model to calculate the rheological deformation response of the surrounding rock in the current operating stage of the pumped storage power station's high-pressure waterway.
[0081] The verification and analysis module is also used to compare and analyze the calculated response results of the surrounding rock of the high-pressure waterway with the actual monitoring results on site, so as to verify the ability of the geomechanical numerical model or surrogate model to characterize the actual rheological deformation behavior of the high-pressure waterway under parameter identification conditions.
[0082] (vii) Prediction and Status Analysis Module The prediction and state analysis module is used to calculate and analyze the rheological deformation response and development trend of the surrounding rock during the operation of the high-pressure waterway of a pumped storage power station, based on a validated geomechanical numerical model or surrogate model.
[0083] The prediction and state analysis module is also used to analyze the rheological deformation response of the surrounding rock of the high-pressure waterway under pumping operation, power generation operation, shutdown conditions or periodic changes in internal water pressure by using a set of parameter posterior samples. This is used to characterize the long-term deformation characteristics of the surrounding rock of the high-pressure waterway under operating conditions and to provide a basis for the stability analysis, deformation control and engineering management of the surrounding rock during the operation of the high-pressure waterway.
[0084] In this invention Figure 4 , Figure 5 and Figure 7 They correspond to different stages in the parameter identification process and have different functions. Figure 4 This section compares the calculated rock mass deformation response based on prior parameter samples with field monitoring data before parameter identification. It illustrates the uncertainty of the initial parameters and the discrepancy between the calculated prior parameters and the measured data, thus necessitating parameter identification. This demonstrates the necessity of parameter identification. Figure 5This figure represents the posterior distribution of rock mass mechanical parameters obtained based on Bayesian inference and Markov chain Monte Carlo sampling, illustrating the probability distribution characteristics of each control parameter after parameter identification. Specifically, it shows the posterior distribution results of rock mass mechanical parameters obtained after constraints based on field monitoring data. Each sub-figure corresponds to a different rock mass mechanical parameter to be identified; the red curve represents the posterior probability density distribution of the parameter, and the black vertical line represents the parameter reference value or engineering calibration value. This figure characterizes the main distribution range and uncertainty level of each rock mass mechanical parameter after parameter identification, thus demonstrating that the present invention can probabilistically identify key mechanical parameters controlling the rheological deformation behavior of rock mass under the conditions of monitoring data error and model uncertainty, further illustrating the importance of probabilistic identification. Figure 7 This represents a comparison between the rock mass deformation response recalculated using the identified parameters and the field monitoring data, illustrating that the parameter identification results can improve the consistency between the model response and the measured data.
[0085] It should be noted that this embodiment uses an engineering slope as an example for illustration, but the method of the present invention is also applicable to rock mass engineering units such as high-pressure waterways and underground powerhouses in pumped storage power stations. For different engineering units, only the geomechanical model and rheological deformation response indicators need to be set accordingly based on their structural characteristics, stress conditions, and monitoring layout. The rheological parameter identification, model verification, and rheological deformation analysis processes remain consistent. Those skilled in the art can make reasonable adjustments or equivalent substitutions to the implementation method without departing from the core idea of the present invention, and all such adjustments should fall within the protection scope of the present invention.
[0086] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.
Claims
1. A method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data, characterized in that, Includes the following steps: S1) Based on engineering geological data, construct a geomechanical numerical model and determine the a priori parameter range of rheology-related rock mass mechanics parameters; S2) Within the range of the prior parameters, for the pumping operation and power generation operation of the pumped storage power station, perform multi-set parameter combination calculations on the geomechanical numerical model to obtain the corresponding rock mass rheological deformation response results and form a training dataset for proxy modeling. S3) Based on the training dataset, construct a data-driven surrogate model to characterize the rheological deformation response characteristics of rock mass engineering under different parameter conditions and different operating conditions; S4) Introduce on-site monitoring data during the operation of the pumped storage power station, conduct joint analysis of the data-driven proxy model and the on-site monitoring data, identify the key mechanical parameters controlling the rheological deformation behavior of the rock mass, and obtain the posterior distribution of parameters reflecting the true working state of the rock mass. S5) Introduce the posterior distribution of the parameters into the geomechanical numerical model, and conduct comparative verification analysis on the mechanical response results of the rock mass engineering in the current operation stage under pumping or power generation conditions. S6) Based on the validated geomechanical numerical model, the rock mass engineering response during the operation of a pumped storage power station is predicted.
2. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 1, characterized in that, The engineering geological data includes at least stratigraphic structure information, rock mass zoning results, structural features, and underground powerhouse spatial layout information; the rock mass mechanical parameters include rheological deformation parameters and strength parameters.
3. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 1, characterized in that, The training dataset consists of at least one of the displacement response, stress response, plastic zone distribution, or seepage response calculated by the geomechanical numerical model under different parameter combinations and different operating conditions.
4. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 1, characterized in that, The data-driven proxy model is used to establish the mapping relationship between rock mass rheological parameters and rock mass engineering rheological deformation response.
5. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 1, characterized in that, The on-site monitoring data includes operational deformation monitoring data, stress monitoring data, seepage pressure monitoring data, or a combination thereof, reflecting the characteristics of rock mass changes over time.
6. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 1, characterized in that, In step S4, the surrogate model and monitoring data are jointly analyzed using Bayesian inference principles and Markov chain Monte Carlo methods.
7. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 6, characterized in that, The process of obtaining the posterior distribution characteristics of the parameters is as follows: Based on the Bayesian inference principle, combined with the prior parameter distribution and field monitoring data, the mechanical parameters controlling the rheological deformation behavior of the rock mass are identified to obtain the posterior distribution characteristics of the parameters. Its posterior distribution of parameters satisfies the following relationship: in, This represents the a priori distribution of rock mass mechanical parameters. This represents the likelihood function of the monitored data under given parameter conditions. This represents the posterior distribution of the identified parameters.
8. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 7, characterized in that, The Markov chain Monte Carlo method is used to sample and calculate the posterior distribution characteristics of the parameters to obtain the posterior sample set of rock mechanics parameters, and the mean, variance, confidence interval and correlation characteristics of the parameters are determined based on the posterior sample set.
9. The method for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data as described in claim 1, characterized in that, The comparative verification analysis in step S5 includes comparing the consistency of the calculation results of the geomechanical numerical model with the actual field observation results to evaluate the rationality of the rock mass rheological deformation response in the current operation stage.
10. A system for identifying rock mass engineering parameters of pumped-storage power stations based on dual-driven mechanistic data, characterized in that, include: The mechanism model construction module is used to construct a numerical model of rock mass engineering geology and mechanics reflecting pumped storage power station based on engineering geological data, and to determine the a priori parameter range of rock mass mechanics parameters. The training data generation module is used to perform multiple parameter combination calculations on the geomechanical numerical model within the range of the prior parameters, for pumping operation and power generation operation, to generate a training dataset for proxy modeling. The proxy model construction module is used to construct a data-driven proxy model based on the training dataset to characterize the rheological deformation response characteristics of rock mass engineering under different parameter conditions and different operating conditions. The data assimilation module is used to introduce on-site monitoring data during the operation of pumped storage power stations. The Markov chain Monte Carlo method is used to jointly analyze the surrogate model and the monitoring data to identify the key mechanical parameters controlling the rheological deformation behavior of rock mass and obtain the posterior distribution of the parameters. The verification and analysis module is used to introduce the posterior distribution of the parameters into the geomechanical numerical model and to compare and verify the rheological deformation response results of the rock mass engineering in the current operation stage under pumping or power generation conditions. The prediction and state analysis module is used to predict the rheological deformation response and development trend of rock mass during the operation of pumped storage power stations based on the validated geomechanical numerical model, and to conduct engineering rheological deformation state analysis of rock mass.