An online closed-loop calibration method and system for a soil-rockfill dam filler constitutive model

Abstract

This invention relates to the field of earth-rock dam engineering technology and discloses an online closed-loop calibration method for the constitutive model of earth-rock dam fill material, comprising the following steps: S1, applying a preset coded excitation sequence to the vibratory compaction equipment; S2, constructing multi-source observation data under a unified time base; S3, constructing an input-output relationship within a preset time window; S4, evaluating the identifiability of the fill material constitutive model parameters under the current construction state based on the dynamic micro-experiment characteristics; S5, obtaining the posterior estimation results of the fill material constitutive model parameters and their uncertainty description; S6, jointly correcting the fill material constitutive model parameters in spatial and temporal dimensions; S7, forming an online closed-loop calibration of the fill material constitutive model. By applying a coded excitation sequence to the vibratory compaction equipment, the construction process forms controlled input conditions with identifiable characteristics under the premise of meeting process constraints, thereby constructing a stable input-response correspondence during the filling construction process.

Description

Technical Field

[0001] This invention relates to the field of earth-rock dam engineering technology, specifically to an online closed-loop calibration method and system for constitutive models of earth-rock dam fill materials. Background Technology

[0002] Earth-rock dams, a common type of dam in water conservancy and hydropower projects, rely heavily on the mechanical behavior of the dam fill material for their overall stability and deformation characteristics. The constitutive model of the fill material, reflecting the stress-strain relationship, is a crucial component in numerical analysis, construction simulation, and operational safety assessment of earth-rock dams. Currently, the parameters of the constitutive model are typically obtained through laboratory tests and adjusted in engineering applications using empirical corrections or inversion analysis. During construction, some technical solutions incorporate on-site monitoring data such as settlement and displacement, using numerical forward modeling and parametric inversion methods to verify or correct the constitutive model parameters, aiming to improve the consistency between the model calculation results and the actual construction response.

[0003] However, the aforementioned existing technologies are mostly based on the construction response data naturally generated during the construction process for parameter inversion. The input conditions during the construction process are mainly determined by the construction technology and equipment operation experience, and there is a lack of controlled excitation process designed for parameter identification needs. In this case, the correspondence between the parameters of the fill constitutive model and the field observation data is often unclear, and it is easy for multiple sets of parameters to produce similar calculated responses, thereby reducing the identifiability of parameter calibration. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an online closed-loop calibration method and system for constitutive models of earth-rock dam fill materials. This solves the problem that the correspondence between existing fill material constitutive model parameters and field observation data is often unclear, and multiple sets of parameters are prone to producing similar calculation responses, thereby reducing the identifiability of parameter calibration.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an online closed-loop calibration method for the constitutive model of earth-rock dam fill material, comprising the following steps:

[0006] S1. During the construction of earth-rock dam filling, a preset coded excitation sequence is applied to the vibratory compaction equipment. The coded excitation sequence is a time-series control quantity formed by segmenting or changing at least one control parameter of the vibratory compaction equipment under the condition of meeting the construction process constraints. Simultaneously, the equipment end observation data and site end observation data corresponding to the coded excitation sequence are collected.

[0007] S2. Time alignment and preprocessing are performed on the collected equipment-side observation data and site-side observation data to construct multi-source observation data under a unified time base;

[0008] S3. Based on the coded excitation sequence and the multi-source observation data, construct an input-output relationship within a preset time window, and extract dynamic micro-experiment features reflecting the dynamic response behavior of the packing from the input-output relationship;

[0009] S4. Based on the dynamic micro-test characteristics, evaluate the identifiability of the filler constitutive model parameters under the current construction state, and determine the set of updatable parameters for online updating based on the identifiability evaluation results.

[0010] S5. For the set of updatable parameters, perform fast-loop online assimilation using the dynamic micro-experiment features as observations to obtain the posterior estimation results of the packing constitutive model parameters and their uncertainty descriptions.

[0011] S6. Based on the construction response data accumulated over multiple construction periods, perform slow-loop parameter field updates and jointly correct the parameters of the fill constitutive model in both spatial and temporal dimensions.

[0012] S7. Based on the results of the fast-loop online assimilation and the slow-loop parameter field update, perform uncertainty gating judgment. When the judgment passes, update the parameters of the packing constitutive model and use the updated parameters of the packing constitutive model in the subsequent construction stage calibration process, thereby forming the online closed-loop calibration of the packing constitutive model.

[0013] Preferably, the coded excitation sequence includes a time series formed by segmenting, sweeping, or pseudo-randomizing at least one parameter of the vibration amplitude, vibration frequency, and travel speed of the vibratory compaction equipment.

[0014] Preferably, the multi-source observation data includes equipment-side observation data acquired synchronously with the coded excitation sequence, and site-side observation data generated by the dam fill material under the action of the coded excitation sequence.

[0015] Preferably, the time alignment is achieved by using the segmented switching times in the coded excitation sequence as a time base to perform unified time base mapping on device-side observation data and site-side observation data with different sampling frequencies and timestamps.

[0016] Preferably, the dynamic micro-experiment feature is a feature vector obtained by performing time-domain analysis, frequency-domain analysis, or time-frequency joint analysis on the input-output relationship between the coded excitation sequence and the multi-source observation data within the preset time window.

[0017] Preferably, the updatable parameter set is a subset of parameters obtained by analyzing the sensitive relationship between the parameters of the packing constitutive model and the characteristics of the dynamic micro-experiment, and filtering the parameters based on the analysis results, and only the packing constitutive model parameters corresponding to the parameter subset are updated online.

[0018] Preferably, the fast-loop online assimilation includes a prediction step and an update step, wherein:

[0019] In the prediction step, corresponding dynamic micro-experiment feature prediction values ​​are generated based on the current packing constitutive model parameters;

[0020] In the update step, the set of updatable parameters is updated online based on the deviation between the predicted value of the dynamic micro-experiment feature and the actual dynamic micro-experiment feature, and the posterior estimation result and its uncertainty description are output.

[0021] Preferably, the slow-loop parameter field update is based on the settlement data, displacement data or stress-related data accumulated over multiple construction periods, and the parameters of the fill constitutive model are jointly updated under consistency constraints at different spatial locations and construction stages of the dam.

[0022] Preferably, the uncertainty gating determination is a determination process jointly executed based on the dynamic micro-experiment feature deviation statistics obtained by the fast-loop online assimilation and the uncertainty description of the posterior estimation result. When the preset conditions are met simultaneously, updating the packing constitutive model parameters is allowed; when the preset conditions are not met, updating is prohibited and the previous stable parameter state is maintained.

[0023] Preferably, an online closed-loop calibration system for the constitutive model of earth-rock dam fill material includes:

[0024] The coding stimulus control module is used to generate and apply the coding stimulus sequence;

[0025] The multi-source data acquisition module is used to acquire device-side observation data and site-side observation data corresponding to the coded excitation sequence;

[0026] The data alignment and preprocessing module is used to construct multi-source observation data under a unified time base.

[0027] The dynamic micro-experiment feature extraction module is used to extract dynamic micro-experiment features that reflect the dynamic response behavior of the packing material.

[0028] The updatable parameter set determination module is used to determine the updatable parameter set of the packing constitutive model;

[0029] The fast-loop online assimilation module is used to perform fast-loop online assimilation of the parameters of the packing constitutive model;

[0030] The slow-loop parameter field update module is used to perform slow-loop parameter field updates on the parameters of the packing constitutive model.

[0031] The uncertainty gating module is used to perform uncertainty gating decisions and output the updated parameters of the packing constitutive model.

[0032] The output of the uncertainty gating module is used to control the coding excitation control module to generate subsequent coding excitation sequences in order to form an online closed-loop calibration.

[0033] This invention provides an online closed-loop calibration method and system for the constitutive model of earth-rock dam fill material. It has the following beneficial effects:

[0034] 1. This invention applies a coded excitation sequence to the vibratory compaction equipment, enabling the construction process to form controlled input conditions with identifiable characteristics while meeting process constraints. This establishes a stable input-response correspondence during the filling construction process. The calibration of the fill constitutive model parameters no longer relies solely on passively collected construction response data, but enhances the correspondence between parameters and observations through repeatable and recordable excitation switching events, providing a clear identification basis for online calibration.

[0035] 2. By determining the set of updatable parameters, this invention performs online updates only on a subset of parameters that are identifiable under the current construction state, while keeping the remaining parameters unchanged. This avoids estimation drift and instability caused by updating all constitutive model parameters simultaneously when information is insufficient or the parameters are highly correlated, thus ensuring that the online calibration process has clear update boundaries in the parameter dimension.

[0036] 3. This invention divides the parameter update process into fast-loop online assimilation based on dynamic micro-experiment characteristics and slow-loop parameter field update based on construction response data, which correspond to different time scales and data types, respectively. The fast loop is used to respond to short-time scale changes during the construction process, while the slow loop is used to constrain the consistency of parameters in space and construction stages, thereby satisfying both real-time requirements and spatial continuity requirements in the same technical solution.

[0037] 4. This invention introduces an uncertainty gating judgment mechanism before parameter updates. The judgment is jointly performed based on the dynamic micro-experiment characteristic deviation statistics and the uncertainty description of the posterior estimation results. The parameters of the packing constitutive model are allowed to be updated only when preset conditions are met. This avoids writing the parameter update results into the model when there are abnormal observations, insufficient identification, or unstable estimation, thereby ensuring the controllability of parameter evolution during online calibration. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the overall structure of an online closed-loop calibration system for the constitutive model of earth-rock dam fill material according to the present invention;

[0039] Figure 2 This is a schematic diagram of the time alignment and preprocessing process for multi-source observation data in this invention;

[0040] Figure 3This is a schematic diagram of the dynamic micro-experiment feature extraction process of the present invention;

[0041] Figure 4 This is a schematic diagram of the process for determining the updatable parameter set of the present invention;

[0042] Figure 5 This is a schematic diagram of the fast-loop online assimilation process of the present invention;

[0043] Figure 6 This is a schematic diagram of the slow loop parameter field update process of the present invention. Detailed Implementation

[0044] The technical solution of the present invention will now be clearly and completely described 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.

[0045] Please see the appendix Figure 1 This invention provides an online closed-loop calibration method for the constitutive model of earth-rock dam fill material, comprising the following steps:

[0046] S1. During the construction of earth-rock dam filling, a preset coded excitation sequence is applied to the vibratory compaction equipment. The coded excitation sequence is a time-series control quantity formed by segmenting or changing at least one control parameter of the vibratory compaction equipment under the condition of meeting the construction process constraints. Simultaneously, the equipment end observation data and site end observation data corresponding to the coded excitation sequence are collected.

[0047] S2. Time alignment and preprocessing are performed on the collected equipment-side observation data and site-side observation data to construct multi-source observation data under a unified time base;

[0048] S3. Based on the coded excitation sequence and multi-source observation data, an input-output relationship is constructed within a preset time window, and dynamic micro-experimental features reflecting the dynamic response behavior of the packing are extracted from the input-output relationship.

[0049] S4. Based on the characteristics of dynamic micro-experiments, evaluate the identifiability of the filler constitutive model parameters under the current construction state, and determine the set of updatable parameters for online updating based on the identifiability evaluation results.

[0050] S5. For the updatable parameter set, fast-loop online assimilation is performed using dynamic micro-experiment features as observations to obtain the posterior estimation results of the packing constitutive model parameters and their uncertainty descriptions.

[0051] S6. Based on the construction response data accumulated over multiple construction periods, perform slow-loop parameter field updates and jointly correct the parameters of the fill constitutive model in both spatial and temporal dimensions.

[0052] S7. Based on the results of fast-loop online assimilation and slow-loop parameter field update, perform uncertainty gating judgment. When the judgment passes, update the parameters of the packing constitutive model and use the updated parameters of the packing constitutive model in the subsequent construction stage calibration process, thereby forming the online closed-loop calibration of the packing constitutive model.

[0053] Specifically, in one embodiment, the present invention provides an online closed-loop calibration method and system for the constitutive model of earth-rock dam fill material, which is applied to the earth-rock dam filling construction process. The present invention takes the control quantity of the vibratory compaction equipment as input, multi-source observation data composed of equipment-end observation data and site-end observation data as observation, and fill material constitutive model parameters as the calibration object. It executes the following steps in sequence: application of coded excitation sequence, observation acquisition, dynamic micro-experiment feature extraction, determination of updatable parameter set, fast-loop online assimilation, slow-loop parameter field update, uncertainty gating judgment and closed-loop feedback, thereby realizing the online closed-loop calibration of the fill material constitutive model.

[0054] In this embodiment, the system deployment includes three types of computing and communication nodes: equipment-side, site-side, and central-side. The equipment-side is integrated into the vibratory compaction equipment and includes a control unit that executes control commands, a data acquisition unit that collects observation data from the equipment, and an event recording unit that outputs excitation switching events. An edge computing node is set up on the site-side to receive data from the equipment-side and site-side and perform fast-loop online assimilation. A computing node is set up on the central-side to receive aggregated construction response data and perform slow-loop parameter field updates. The above nodes transmit data through communication links, which can include wired or wireless links, and each node provides a timestamp output interface for time alignment.

[0055] In this embodiment, the system functional modules are divided into an coded excitation control module, a multi-source data acquisition module, a data alignment and preprocessing module, a dynamic micro-experiment feature extraction module, an updatable parameter set determination module, a fast-loop online assimilation module, a slow-loop parameter field update module, and an uncertainty gating module. The coded excitation control module is connected to the vibration compaction equipment control unit and is used to generate and apply coded excitation sequences. The multi-source data acquisition module is connected to both the equipment-side sensors and the site-side sensors to acquire equipment-side observation data and site-side observation data. The data alignment and preprocessing module receives the data output by the multi-source data acquisition module and the excitation switching events output by the coded excitation control module to construct multi-source observation data. The dynamic micro-experiment feature extraction module receives... The system comprises several modules: a coding excitation sequence and multi-source observation data, used to output dynamic micro-experiment characteristics; an updatable parameter set determination module, receiving the dynamic micro-experiment characteristics and outputting an updatable parameter set; a fast-loop online assimilation module, receiving the dynamic micro-experiment characteristics and the updatable parameter set, used to output the posterior estimation results of the filler constitutive model parameters and their uncertainty descriptions; a slow-loop parameter field update module, receiving the construction response data accumulated over multiple construction periods, used to output the parameter field update results; and an uncertainty gating module, receiving the outputs from the fast-loop online assimilation module and the slow-loop parameter field update module, used to perform uncertainty gating determination and output parameter update control signals. The output of the uncertainty gating module is used to control the coding excitation control module to generate subsequent coding excitation sequences.

[0056] To ensure consistency in terminology and uniqueness of symbols, this embodiment adopts the following unified notation; the position vector is denoted as... The time is recorded as The control input for the vibratory compaction equipment is denoted as... The observation data from the equipment end is recorded as follows: The field observation data is recorded as The two are concatenated to obtain the observation vector. The observations after time alignment and preprocessing are denoted as The parameter field of the constitutive model of the packing is denoted as... Its initial prior is denoted as .

[0057] Please see the appendix Figure 2 The coded excitation sequence includes a time series formed by segmenting, sweeping, or pseudo-randomizing at least one parameter of the vibration amplitude, vibration frequency, and travel speed of the vibratory compaction equipment.

[0058] Specifically, in this embodiment, the coded excitation sequence is generated by the coded excitation control module and executed by the vibratory compaction equipment; the coded excitation sequence is defined as a time-series control quantity formed by segmented switching or changing at least one control parameter of the vibratory compaction equipment; the control parameter set is defined as... ,in Set the vibration amplitude value. Vibration frequency setting The travel speed is set; the coded excitation control module uses discrete time intervals. Generate control sequence and will The command is sent to the control unit of the vibratory compaction equipment for execution; to achieve reproducible segmented switching, the coded excitation control module outputs an excitation switching event each time a control segment switches, and the excitation switching event is used... It means that, among them To switch timestamps, To switch event numbers, The stimulus switching event is written to the event log by the event recording unit and sent to the data alignment and preprocessing module via the communication link;

[0059] In this embodiment, the coded excitation sequence includes three types of control sequence forms: segmented switching, frequency sweeping change, or pseudo-random change; segmented switching refers to setting the frequency sweeping change within adjacent time periods. Take different constant vectors; frequency sweep refers to making within a preset time period... It changes and remains constant over time according to a preset function. and Within the constraints of construction technology; pseudo-random variation refers to... or The system switches according to a predetermined pseudo-random sequence within a pre-defined finite set, ensuring that the switching time can be recorded by the excitation switching event; the coded excitation control module applies a set of constraints to the control input. ,satisfy ,in It is determined by the allowable range of equipment and the constraints of construction technology.

[0060] Please see the appendix Figure 2 The multi-source observation data includes equipment-side observation data acquired synchronously with the coded excitation sequence, as well as site-side observation data generated by the dam fill material under the action of the coded excitation sequence.

[0061] Time alignment is achieved by using the segmented switching times in the coded excitation sequence as a time base to perform unified time base mapping on device-side observation data and site-side observation data with different sampling frequencies and timestamps.

[0062] Specifically, in this embodiment, the multi-source data acquisition module is used to acquire both device-side observation data and site-side observation data; device-side observation data... This includes the vibration response of the vibratory compaction equipment itself and the control execution feedback. The vibration response includes at least one of the acceleration, velocity, or displacement of the equipment body or wheel end. The control execution feedback includes the actual executed... At least one of the following; site observation data This includes the surface or shallow subsurface response quantities generated by the dam body fill material under the action of a coded excitation sequence. The surface or shallow subsurface response quantities include at least one of acceleration, velocity, displacement, or settlement increment. The multi-source data acquisition module adds a timestamp to each type of data stream and records the data source identifier to form a data stream set. ,in Indicates the data stream number. For timestamps, This is the data vector segment corresponding to the timestamp;

[0063] In this embodiment, the data alignment and preprocessing module receives a set of data streams. With stimulus switching events This is used to construct multi-source observation data under a unified time base; for each data stream The data alignment and preprocessing module switches events based on the stimulus switching event number. Determine the corresponding event time in this data stream And calculate the time offset. ,satisfy

[0064]

[0065] Then, a uniform time base mapping is performed on each timestamp within the data stream. Obtain the aligned data stream ;

[0066] In this embodiment, the data alignment and preprocessing module resamples the aligned multiple data streams to a common time grid. Construct observation vectors Resampling employs interpolation or hold-before strategies, and retains the original sampling identifier for each channel for subsequent quality control; the data alignment and preflight processing module... Denoising filtering and outlier handling are performed to obtain preprocessed observations. The denoising filter operator is denoted as ,satisfy Anomaly handling includes removing channel values ​​that exceed the preset physical range and generating anomaly flags. The anomaly flags are output along with the data to the dynamic micro-experiment feature extraction module.

[0067] Please see the appendix Figure 3 The dynamic micro-experiment features are the feature vectors obtained by performing time-domain analysis, frequency-domain analysis, or time-frequency joint analysis on the input-output relationship between the coded excitation sequence and multi-source observation data within a preset time window.

[0068] Specifically, in this embodiment, the dynamic micro-experiment feature extraction module constructs the input-output relationship and extracts dynamic micro-experiment features within a preset time window; the preset time window is divided according to the excitation switching event, defining the first... The time window is ,in and Taken from the set of excitation switching event timestamps and satisfy The dynamic micro-experiment feature extraction module extracts the input sequence corresponding to the window from the coded stimulus sequence. And extract the output sequence corresponding to this window from the multi-source observation data. This constitutes the input / output data pairs within the window;

[0069] In this embodiment, the dynamic micro-experiment feature extraction module uses feature extraction operators. Map the input-output data pair within the window to a feature vector. ,satisfy

[0070] ;

[0071] Φ consists of three types of operations: time-domain analysis, frequency-domain analysis, and joint time-frequency analysis. Time-domain analysis includes calculating the root mean square, peak value, and cumulative energy for the selected output channel. Frequency-domain analysis includes calculating the dominant frequency, band energy distribution, and phase difference for the selected output channel. Joint time-frequency analysis includes a summary of the time-frequency energy distribution obtained by performing a short-time Fourier transform on the selected output channel within a window. The dynamic micro-experiment feature extraction module outputs... As a feature of the dynamic micro-experimentation of this window, The window represents time, and is taken as...

[0072] Please see the appendix Figure 4 The updatable parameter set is a subset of parameters obtained by analyzing the sensitive relationship between the parameters of the packing constitutive model and the characteristics of dynamic micro-experiments, and only the parameters of the packing constitutive model corresponding to the parameter subset are updated online.

[0073] Specifically, in this embodiment, the updatable parameter set determination module is used to determine the updatable parameter set for online updates; the packing constitutive model parameters are expressed in the form of a parameter field. And expand according to the hierarchical and local basis functions as follows

[0074] ;

[0075] in For the hierarchical basis function matrix, This is a vector of hierarchical coefficients; For local basis function matrices, The local coefficient vector; the coefficient vector to be estimated is defined as...

[0076] ;

[0077] Updatable parameter set determination module The components are selected as candidate update objects, and the set of indices of the candidate components is output. This set of indexes is defined as an updatable set of parameters;

[0078] In this embodiment, the updatable parameter set is obtained by analyzing the sensitive relationship between parameters and dynamic micro-experiment characteristics; defining the window. The feature prediction function is This function is used by the prediction step within the fast-loop online assimilation module; the updatable parameter set determines the module's current coefficient estimates. Calculate the Jacobian matrix at point

[0079] ;

[0080] And construct an information matrix

[0081] ;

[0082] in The feature weight matrix; the updatable parameter set determination module starts from... Select the index set from the component set And generate a selection matrix. To extract subspace variables

[0083] ;

[0084] This subspace variable The corresponding updatable parameter set; the updatable parameter set determination module will... The dimensional information is sent to the fast-loop online assimilation module to limit the online update range to the set of updatable parameters.

[0085] Please see the appendix Figure 5 Fast-loop online assimilation includes a prediction step and an update step, wherein:

[0086] In the prediction step, corresponding dynamic micro-experiment feature prediction values ​​are generated based on the current packing constitutive model parameters;

[0087] In the update step, the set of updatable parameters is updated online based on the deviation between the predicted values ​​of dynamic micro-experiment features and the actual dynamic micro-experiment features, and the posterior estimation results and their uncertainty descriptions are output.

[0088] Specifically, in this embodiment, the fast-loop online assimilation module uses dynamic micro-experimental features as observations to analyze the subspace variables corresponding to the updatable parameter set. Perform online assimilation. Fast loop indexed by window. Discrete execution, denoted For the first Each window represents a subspace variable at a given time. The fast-loop online assimilation module establishes the evolution equation.

[0089] ;

[0090] in The process noise has a covariance of The fast-loop online assimilation module establishes the observation equations.

[0091] ;

[0092] in To observe the noise, the covariance is... The predicted values ​​of dynamic micro-experiment features generated for the prediction step.

[0093] In this embodiment, the prediction step of the fast-loop online assimilation module includes: based on the updated value from the previous window... Generate predicted mean Predicting covariance ,in

[0094] ;

[0095] And based on Calculate the predicted value of dynamic micro-experiment features To generate The fast-loop online assimilation module calls the predictor. The input contains the coded stimulus sequence within the window. with by The parameter field obtained by restoration The output is an observable response sequence within the window, which is then processed by... Feature summarization operation generation

[0096] In this embodiment, the updating step of the fast-loop online assimilation module includes: updating based on the deviation between the predicted value of the dynamic micro-experiment feature and the actual dynamic micro-experiment feature. The posterior estimate. Define the linearized Jacobian matrix.

[0097] ;

[0098] Define Kalman gain

[0099] ;

[0100] Define the deviation vector

[0101] ;

[0102] The updated values ​​yield the posterior mean and posterior covariance.

[0103] ;

[0104] The fast-loop online assimilation module outputs the posterior estimation results. and its uncertainty description And write it as the prior input for subsequent windows.

[0105] Please see the appendix Figure 6 The slow-loop parameter field update is based on the settlement data, displacement data or stress-related data accumulated over multiple construction periods. It performs a joint update of the fill constitutive model parameters under consistency constraints at different spatial locations and construction stages of the dam.

[0106] Specifically, in this embodiment, the slow-loop parameter field update module performs joint correction on the parameter field based on the construction response data accumulated over multiple construction periods. Construction response data is defined as settlement data, displacement data, or stress-related data within the construction period. The time series or its aggregate within the range is denoted as . The slow-loop parameter field update module is indexed by slow window. Execute and construct the slow observation prediction function.

[0107] ;

[0108] in For numerical forward modeling operators, the input contains coefficients. Constructed parameter field Construction load history For slow observation operators, used to obtain the result from the forward state mapping. Predictions in the same dimension.

[0109] In this embodiment, the slow-loop parameter field update module updates the parameters based on consistency constraints. Perform the update. The slow-loop parameter field update module defines the objective function.

[0110] ;

[0111] in For slow observation weight matrix, This is the regularization coefficient. Regularization term. To constrain the consistency of parameters in both spatial and temporal dimensions, it consists of two parts: hierarchical and local.

[0112] ;

[0113] in For discrete smoothing operators with local coefficients, This is the difference operator for the hierarchical coefficients. Solved by the slow-loop parameter field update module.

[0114] ;

[0115] It also outputs the corresponding parameter field update results, which include... and with Corresponding parameter field .

[0116] Please see the appendix Figure 6 Uncertainty gating judgment is a judgment process jointly executed based on the dynamic micro-experiment characteristic deviation statistics obtained by fast-loop online assimilation and the uncertainty description of the posterior estimation results. When the preset conditions are met, updating the packing constitutive model parameters is allowed; when the preset conditions are not met, updating is prohibited and the previous stable parameter state is maintained.

[0117] Specifically, in this embodiment, the uncertainty gating module is used to perform uncertainty gating decisions; the gating input includes the uncertainty description of the dynamic micro-experiment feature deviation statistics and the posterior estimation results; and the fast loop window... The uncertainty gating module takes the deviation vector output by the fast-loop online assimilation module. Linearized Jacobian matrix Predicting covariance Observation noise covariance Calculate the deviation covariance

[0118] ;

[0119] And calculate the deviation statistic.

[0120] ;

[0121] The uncertainty gating module is based on the posterior covariance output by the fast-loop online assimilation module. Calculate uncertainty index

[0122] ;

[0123] In this embodiment, the uncertainty gating module performs uncertainty gating determination based on preset conditions, which include a deviation statistics threshold. With uncertainty threshold When satisfied When the uncertainty gating is successful, the uncertainty gating module outputs a signal allowing updates; when the above conditions are not met, the uncertainty gating is failed, the uncertainty gating module outputs a signal prohibiting updates, and maintains the previous stable parameter state. The previous stable parameter state is defined as the filler constitutive model parameters written when the most recent uncertainty gating was successful, or the initial prior parameters. .

[0124] An online closed-loop calibration system for the constitutive model of earth-rock dam fill material includes:

[0125] The coding stimulus control module is used to generate and apply the coding stimulus sequence;

[0126] The multi-source data acquisition module is used to acquire equipment-side observation data and site-side observation data corresponding to the coded excitation sequence;

[0127] The data alignment and preprocessing module is used to construct multi-source observation data under a unified time base.

[0128] The dynamic micro-experiment feature extraction module is used to extract dynamic micro-experiment features that reflect the dynamic response behavior of the packing material.

[0129] The updatable parameter set determination module is used to determine the updatable parameter set of the packing constitutive model;

[0130] The fast-loop online assimilation module is used to perform fast-loop online assimilation of the parameters of the packing constitutive model;

[0131] The slow-loop parameter field update module is used to perform slow-loop parameter field updates on the parameters of the packing constitutive model.

[0132] The uncertainty gating module is used to perform uncertainty gating decisions and output the updated parameters of the packing constitutive model.

[0133] The output of the uncertainty gating module is used to control the coding excitation control module to generate subsequent coding excitation sequences in order to form an online closed-loop calibration.

[0134] Specifically, in this embodiment, the output of the uncertainty gating module is used to control the coding excitation control module to generate subsequent coding excitation sequences; the allowable update control signal is used to trigger the coding excitation control module to write the parameters output by the fast-loop online assimilation module and the slow-loop parameter field update module into the current packing constitutive model parameters, and use these parameters to generate dynamic micro-experiment feature prediction values ​​during the calibration process in the subsequent construction stage; the prohibitive update control signal is used to trigger the coding excitation control module to maintain the previous stable parameter state and generate subsequent coding excitation sequences according to the preset sequence scheduling rules. The sequence scheduling rules select at least one control sequence form among segmented switching, frequency sweeping change, or pseudo-random change with the excitation switching event as the boundary, and maintain the coding excitation sequence satisfying the constraint set. .

[0135] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. An online closed-loop calibration method for a soil-rockfill dam filler constitutive model, characterized in that, Includes the following steps: S1. During the construction of earth-rock dam filling, a preset coded excitation sequence is applied to the vibratory compaction equipment. The coded excitation sequence is a time-series control quantity formed by segmenting or changing at least one control parameter of the vibratory compaction equipment under the condition of meeting the construction process constraints. Simultaneously, the equipment end observation data and site end observation data corresponding to the coded excitation sequence are collected. S2. Time alignment and preprocessing are performed on the collected equipment-side observation data and site-side observation data to construct multi-source observation data under a unified time base; S3. Based on the coded excitation sequence and the multi-source observation data, construct an input-output relationship within a preset time window, and extract dynamic micro-experiment features reflecting the dynamic response behavior of the packing from the input-output relationship; S4. Based on the dynamic micro-experiment characteristics, evaluate the identifiability of the filler constitutive model parameters under the current construction state, and determine the set of updatable parameters for online updating based on the identifiability evaluation results. S5. For the set of updatable parameters, perform fast-loop online assimilation using the dynamic micro-experiment features as observations to obtain the posterior estimation results of the packing constitutive model parameters and their uncertainty descriptions. S6. Based on the construction response data accumulated over multiple construction periods, perform slow-loop parameter field updates and jointly correct the parameters of the fill constitutive model in both spatial and temporal dimensions. S7. Based on the results of the fast-loop online assimilation and the slow-loop parameter field update, perform uncertainty gating judgment. When the judgment passes, update the parameters of the packing constitutive model and use the updated parameters of the packing constitutive model in the subsequent construction stage calibration process, thereby forming the online closed-loop calibration of the packing constitutive model.

2. The online closed-loop calibration method for the soil-rockfill dam filling material constitutive model according to claim 1, characterized in that, The coded excitation sequence includes a time series formed by segmenting, sweeping, or pseudo-randomizing at least one parameter of the vibration amplitude, vibration frequency, and travel speed of the vibratory compaction equipment.

3. The online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to claim 1, characterized in that, The multi-source observation data includes equipment-side observation data acquired synchronously with the coded excitation sequence, and site-side observation data generated by the dam fill material under the action of the coded excitation sequence.

4. The online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to claim 1, characterized in that, The time alignment is achieved by using the segmented switching times in the coded excitation sequence as a time base to perform unified time base mapping on device-side observation data and site-side observation data with different sampling frequencies and timestamps.

5. The online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to claim 1, characterized in that, The dynamic micro-experiment features are feature vectors obtained by performing time-domain analysis, frequency-domain analysis, or time-frequency joint analysis on the input-output relationship between the coded excitation sequence and the multi-source observation data within the preset time window.

6. The online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to claim 1, characterized in that, The updatable parameter set is a subset of parameters obtained by analyzing the sensitive relationship between the parameters of the packing constitutive model and the characteristics of the dynamic micro-experiment, and only the parameters of the packing constitutive model corresponding to the parameter subset are updated online.

7. The online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to claim 1, characterized in that, The fast-loop online assimilation includes a prediction step and an update step, wherein: In the prediction step, corresponding dynamic micro-experiment feature prediction values ​​are generated based on the current packing constitutive model parameters; In the update step, the set of updatable parameters is updated online based on the deviation between the predicted value of the dynamic micro-experiment feature and the actual dynamic micro-experiment feature, and the posterior estimation result and its uncertainty description are output.

8. The online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to claim 1, characterized in that, The slow-loop parameter field update is based on the settlement data, displacement data or stress-related data accumulated over multiple construction periods, and is a joint update of the fill constitutive model parameters under consistency constraints across different spatial locations and construction stages of the dam.

9. The online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to claim 1, characterized in that, The uncertainty gating decision is a decision process jointly executed based on the dynamic micro-experiment characteristic deviation statistics obtained by the fast-loop online assimilation and the uncertainty description of the posterior estimation result. When the preset conditions are met, updating the packing constitutive model parameters is allowed; when the preset conditions are not met, updating is prohibited and the previous stable parameter state is maintained.

10. An online closed-loop calibration system for the constitutive model of earth-rock dam fill material, characterized in that, An online closed-loop calibration method for the constitutive model of earth-rock dam fill material according to any one of claims 1-9, comprising: The coding stimulus control module is used to generate and apply the coding stimulus sequence; The multi-source data acquisition module is used to acquire device-side observation data and site-side observation data corresponding to the coded excitation sequence; The data alignment and preprocessing module is used to construct multi-source observation data under a unified time base; The dynamic micro-experiment feature extraction module is used to extract dynamic micro-experiment features that reflect the dynamic response behavior of the packing material. The updatable parameter set determination module is used to determine the updatable parameter set of the packing constitutive model; The fast-loop online assimilation module is used to perform fast-loop online assimilation of the parameters of the packing constitutive model; The slow-loop parameter field update module is used to perform slow-loop parameter field updates of the packing constitutive model parameters; The uncertainty gating module is used to perform uncertainty gating decisions and output the updated parameters of the packing constitutive model. The output of the uncertainty gating module is used to control the coding excitation control module to generate subsequent coding excitation sequences in order to form an online closed-loop calibration.

Images