Rock mass deformation modulus accurate prediction method, system and device based on adaptive Kalman filtering and medium

Abstract

The invention provides a rock mass deformation modulus accurate prediction method, system and device based on adaptive Kalman filtering and a medium. The method comprises the following steps: according to dam foundation engineering rock mass rigid bearing plate in-situ experiment data and sound wave test experiment data, calculating a rock mass deformation modulus of a dam foundation; and taking the acoustic wave velocity and the deformation modulus of the rock mass as state variables required by adaptive Kalman filtering. And performing real-time estimation on the dynamic noise variance matrix according to the existing information, and comprehensively adjusting the noise matrix to obtain a system filtering value so as to predict and evaluate the system state. The method is implemented on the basis of an adaptive Kalman filtering algorithm by referring to a wave velocity-deformation modulus prediction formula of engineering experience. According to the method, the prediction model is corrected according to the adaptive method, data divergence of the system in the filtering process can be effectively reduced, model errors are reduced, and prediction evaluation precision is improved.

Description

Technical Field

[0001] This invention pertains to the fields of data assimilation and rock mechanics, and relates to an optimization method for a data assimilation algorithm, particularly to a method, system, device, and medium for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering. Background Technology

[0002] The excavation relaxation disturbance effect of rock masses is crucial in dam foundation engineering, especially for basalt masses with strong anisotropy and unloading relaxation characteristics. These masses have unique rock structures and internal longitudinal and transverse hidden fractures. Therefore, analyzing and mining field test results and monitoring data during the excavation of such rock masses is an important way to study their unloading relaxation characteristics. Previous studies have mostly relied on traditional statistical methods for processing field test data. However, these methods are not perfect at handling errors and noise in monitoring data, and the function curves fitted by data points are not ideal, resulting in poor evaluation indicators such as R0. 2 The requirements for establishing a model could not be met; therefore, a more in-depth analysis and exploration of the test results data during excavation is needed.

[0003] Currently, commonly used monitoring data processing methods include regression analysis, time series analysis, grey system analysis, and Kalman filtering. Among these, Kalman filtering is a widely used method. It can not only reduce measurement errors but also predict the monitoring data for the next period based on the filtered values ​​of the state variables from the previous period. Kalman filtering is a dynamic data processing method that does not require storing all previous data; it only needs the current observation data and the prediction model to achieve data filtering and prediction. This type of method is well-suited for processing field test data in dam foundation engineering. However, traditional Kalman filtering, when processing observation data, suffers from algorithmic limitations: for example, it may only process noise as Gaussian white noise, which cannot be modified during the filtering process, leading to divergent error results. Summary of the Invention

[0004] The first objective of this invention is to provide an accurate method for predicting the deformation modulus of rock mass based on adaptive Kalman filtering, in order to improve the accuracy of deformation modulus measurement data of rock mass in dam foundation engineering and to establish a wave velocity-deformation modulus prediction model.

[0005] Therefore, the above-mentioned objective of the present invention is achieved through the following technical solution:

[0006] A method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering, characterized in that the method includes the following steps:

[0007] S1. Based on the in-situ test data of the deformation of the rigid bearing plate of the rock mass in the dam foundation project and the test data of the acoustic wave test, the acoustic wave velocity V and deformation modulus E of the rock mass are obtained and used as the state vector X0 and its corresponding covariance matrix D0 of the adaptive Kalman filter model.

[0008] S2. Based on the existing information, establish the covariance matrix D of the observation noise. Δ The covariance transition matrix D of dynamic noise Ω ;

[0009] S3. Establish a variance-compensated adaptive Kalman filter model and determine the values ​​of the system's state transition matrix φ(k|k-1), dynamic noise matrix T(k|k-1), and observation matrix B(k);

[0010] S4. Input the system's state transition matrix, dynamic noise matrix, and observation matrix; begin calculation to obtain the predicted value. Forecast covariance matrix D X(k丨k-1) and the gain matrix J(k);

[0011] S5. Input a set of in-situ test data of rigid bearing plate deformation in rock mass and acoustic wave test data as observation data, perform adaptive Kalman filtering, and obtain the best predicted value of the set of observations. and covariance matrix D X(k丨k) Simultaneously, the covariance matrix D of the dynamic noise is calculated. Ω ;

[0012] S6. Return to step S4 and proceed to the next iterative calculation.

[0013] S7. Obtain the best prediction value for this set after each iteration. and covariance matrix D X(k丨k) The process continues until the iteration count k reaches the last system state q, i.e., k = q. The resulting filtered value reflects the rock mass acoustic wave velocity and deformation modulus that are closest to the current system state.

[0014] S8. Based on the filtered values ​​after iteration, optimize the adaptive Kalman filter prediction model in step S3 to obtain the predicted values ​​of the two state variables, rock mass wave velocity and deformation modulus, of the system in the next state.

[0015] While adopting the above technical solutions, the present invention may also adopt or combine the following technical solutions:

[0016] As a preferred technical solution of the present invention: In step S1, the in-situ test data of the deformation of the rigid bearing plate of the rock mass and the test data of the acoustic wave test are used as input state variables, which are collected from the in-situ test data of the deformation of the rigid bearing plate and the test data of the wave velocity test, including the acoustic wave velocity and deformation modulus of the rock mass.

[0017] As a preferred technical solution of the present invention: In order to improve the accuracy of model prediction, since the deformation modulus is not only controlled by the sound wave as an influencing factor, but also by various other factors that are not convenient to measure directly, in steps S4 to S5, as new observation data is added, the noise covariance matrix is ​​dynamically adjusted to reflect the real-time changes in the observed noise and the latest estimate of the prediction error. After the noise matrix changes dynamically according to the filtering results, the accuracy of model prediction can be significantly improved.

[0018] The second objective of this invention is to provide an accurate prediction system for rock mass deformation modulus based on adaptive Kalman filtering.

[0019] Therefore, the above-mentioned objective of the present invention is achieved through the following technical solution:

[0020] A rock mass deformation modulus accurate prediction system based on adaptive Kalman filtering, characterized in that the system includes the following modules:

[0021] - Test data acquisition module, the test data acquisition module is used to obtain the rock mass acoustic wave velocity V and deformation modulus E based on the in-situ test data of the deformation of the rigid bearing plate of the rock mass and the acoustic wave test data, and use them as the state vector X0 and its corresponding covariance matrix D0 of the adaptive Kalman filter model;

[0022] - A variance-compensated adaptive Kalman filter model building module, which is used to build the covariance matrix D of the observation noise based on existing information. Δ The covariance transition matrix D of dynamic noise Ω A variance-compensated adaptive Kalman filter model is established to determine the values ​​of the system's state transition matrix φ(k|k-1), dynamic noise matrix T(k|k-1), and observation matrix B(k).

[0023] - Iterative recursive calculation module, which is used to input data into the variance-compensated adaptive Kalman filter model established by the variance-compensated adaptive Kalman filter model establishment module, and calculate to obtain the predicted value. Forecast covariance matrix D X(k丨k-1) The gain matrix J(k) is obtained; the observed data are input into the variance-compensated adaptive Kalman filter model, adaptive Kalman filtering is performed, and the best predicted value of the set of observations is obtained. and covariance matrix D X(k丨k) Simultaneously, the covariance matrix D of the dynamic noise is calculated. Ω Then, iterative calculations are performed to obtain the best predicted value for that set after each iteration. and covariance matrix D X(k丨k) Continue until the iteration ends;

[0024] - Adaptive Kalman Filter Prediction Model Optimization Module: Based on the filter value obtained by the iterative recursive calculation module, the adaptive Kalman filter prediction model optimization module optimizes the variance-compensated adaptive Kalman filter model established by the variance-compensated adaptive Kalman filter model establishment module, and obtains the predicted values ​​of the two state variables, rock mass wave velocity and deformation modulus, of the system in the next state.

[0025] The third objective of this invention is to provide an accurate prediction device for rock mass deformation modulus based on adaptive Kalman filtering.

[0026] Therefore, the above-mentioned objective of the present invention is achieved through the following technical solution:

[0027] An accurate prediction device for rock mass deformation modulus based on adaptive Kalman filtering, characterized in that the device comprises:

[0028] - At least one processor;

[0029] - At least one memory for storing at least one computer program;

[0030] The processor executes a computer program in the memory to implement the steps of the method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering as described above.

[0031] A third objective of this invention is to provide a computer storage medium.

[0032] Therefore, the above-mentioned objective of the present invention is achieved through the following technical solution:

[0033] A computer storage medium, characterized in that: the computer storage medium stores a computer program, which is executed by a computer to implement the steps of the method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering as described above.

[0034] This invention provides a method, system, device, and medium for accurate prediction of rock mass deformation modulus based on adaptive Kalman filtering. The method includes the following steps: Based on in-situ experimental data and acoustic wave test data of the rigid bearing plate of the rock mass in dam foundation engineering, the acoustic wave velocity and deformation modulus of the rock mass are used as state variables required for adaptive Kalman filtering. The dynamic noise variance matrix is ​​estimated in real time based on existing information, and the filtered system value is obtained after comprehensive adjustment of the noise matrix, thereby predicting and evaluating the system state. This invention is based on the wave velocity-deformation modulus prediction formula derived from engineering experience and is implemented on the algorithm of adaptive Kalman filtering. This invention uses an adaptive method to correct the prediction model, which can effectively reduce data divergence during the filtering process, reduce model errors, and improve the accuracy of prediction and evaluation.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] (1) The rock wave velocity and deformation modulus obtained from the rigid bearing plate test and acoustic wave test of the rock mass of the dam foundation project are filtered, and a prediction model is established through the test data. This method only requires the current test data and continuously derives the state variable value corresponding to the next state from the initial value of the model. Therefore, it will not occupy a lot of memory and can still run well even when facing large and numerous monitoring data processing.

[0037] (2) The adaptive Kalman filtering method used in this invention can dynamically adjust the process noise, so that the model can always make relatively accurate predictions of the system state and the error will not diverge and cause the model to collapse.

[0038] (3) The core of this invention is to determine the system state and state variable parameters to establish a state transition matrix. Therefore, it has good portability to other parameters and states of dam foundation rock mass. For example, filtering and prediction of dam foundation displacement deformation, temperature field, stress field, etc. can be carried out in a similar way to achieve a more comprehensive and specific data assimilation effect, qualitatively and quantitatively analyze dam foundation rock mass, improve the accuracy of parameters, and further deepen the understanding of dam foundation rock mass.

[0039] (4) This invention provides an accurate prediction method for rock mass deformation modulus based on adaptive Kalman filtering. Compared with commonly used statistical processing and evaluation methods in engineering, the method provided by this invention can achieve better system state prediction results; it can correct the system's state variables at each step and accurately predict the system's state at each step. During the adaptive Kalman filtering process, when system noise and error change, the noise matrix will change dynamically accordingly. As filtering progresses, the model is gradually optimized, the error value gradually converges, and the filtering effect is good. Feeding the filtered rock mass wave velocity-deformation modulus noise matrix back to the system state transition matrix can optimize the prediction model of rock mass wave velocity-deformation modulus under the current state.

[0040] (5) The filtered value after iteration can be used for the optimal estimation of the next system state, and the filtered whole process wave velocity-deformation modulus prediction model can be used for similar dam foundation projects to predict the deformation modulus value of the rock mass of the dam foundation project under different wave velocities, providing a reference for the project. Attached Figure Description

[0041] Figure 1 Flowchart of adaptive Kalman filtering

[0042] Figure 2 This is a graph showing the error deviation of the filtered acoustic data.

[0043] Figure 3 This is a graph showing the error deviation of the deformation modulus data after filtering. Detailed Implementation

[0044] The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

[0045] like Figure 1 As shown, a method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering includes the following steps:

[0046] S1. Based on the in-situ test data of the deformation of the rigid bearing plate of the rock mass in the dam foundation project and the test data of the acoustic wave test, the acoustic wave velocity V and deformation modulus E of the rock mass are obtained and used as the state vector X0 and its corresponding covariance matrix D0 of the adaptive Kalman filter model.

[0047] S2. Based on the existing information, establish the covariance matrix D of the observation noise. Δ The covariance transition matrix D of dynamic noise Ω ;

[0048] S3. Establish a variance-compensated adaptive Kalman filter model and determine the values ​​of the system's state transition matrix φ(k|k-1), dynamic noise matrix T(k|k-1), and observation matrix B(k);

[0049] S4. Input the system's state transition matrix, dynamic noise matrix, and observation matrix; begin calculation to obtain the predicted value. Forecast covariance matrix D X(k丨k-1) and the gain matrix J(k);

[0050] S5. Input a set of in-situ test data of rigid bearing plate deformation in rock mass and acoustic wave test data as observation data, perform adaptive Kalman filtering, and obtain the best predicted value of the set of observations. and covariance matrix D X(k丨k) Simultaneously, the covariance matrix D of the dynamic noise is calculated. Ω ;

[0051] S6. Return to step S4 and proceed to the next iterative calculation.

[0052] S7. Obtain the best prediction value for this set after each iteration. and covariance matrix D X(k丨k) The process continues until the iteration count k reaches the last system state q, i.e., k = q. The resulting filtered value reflects the rock mass acoustic wave velocity and deformation modulus that are closest to the current system state.

[0053] S8. Based on the filtered values ​​after iteration, optimize the adaptive Kalman filter prediction model in step S3 to obtain the predicted values ​​of the two state variables, rock mass wave velocity and deformation modulus, of the system in the next state.

[0054] The field test data were iterated into the variance-compensated adaptive Kalman filter model established in this invention, and the predicted data of wave velocity and deformation modulus were obtained as shown in Table 1 below.

[0055] Table 1

[0056]

[0057] This invention employs an adaptive Kalman filter (AKF) algorithm to accurately analyze in-situ test data of rigid bearing plates in rock masses. The advantage of this algorithm lies in its ability to automatically adjust the noise covariance matrix in response to continuous changes in real-time data, thereby ensuring continuous optimization of the estimation results even when data uncertainties exist.

[0058] As the number of system iterations increases, the accuracy of the established prediction model is significantly improved. For example... Figure 2 and Figure 3 As shown, the AKF model effectively captured the changing trends of wave velocity and deformation modulus during continuous observation. Despite data fluctuations, the model exhibited gradual stability and refined its trend prediction capabilities through continuous iteration.

[0059] In the adaptive Kalman filter model created in this invention, the self-correction mechanism plays a crucial role. The model's adaptive parameter adjustment function ensures effective adaptation to and identification of changes in actual measurements. This characteristic is particularly important when performing complex geological data analysis and prediction.

[0060] Through Table 1 and Figure 2 , Figure 3 Through detailed analysis, the present invention can accurately predict wave velocity and deformation modulus from complex test data containing noise and outliers. This ability proves that the present invention is not only theoretically sound, but also has high value in practical applications.

[0061] This invention also provides an accurate prediction system for rock mass deformation modulus based on adaptive Kalman filtering, comprising the following modules:

[0062] - Test data acquisition module, the test data acquisition module is used to obtain the rock mass acoustic wave velocity V and deformation modulus E based on the in-situ test data of the deformation of the rigid bearing plate of the rock mass and the acoustic wave test data, and use them as the state vector X0 and its corresponding covariance matrix D0 of the adaptive Kalman filter model;

[0063] - A variance-compensated adaptive Kalman filter model building module, which is used to build the covariance matrix D of the observation noise based on existing information. Δ The covariance transition matrix D of dynamic noise Ω A variance-compensated adaptive Kalman filter model is established to determine the values ​​of the system's state transition matrix φ(k|k-1), dynamic noise matrix T(k|k-1), and observation matrix B(k).

[0064] - Iterative recursive calculation module, which is used to input data into the variance-compensated adaptive Kalman filter model established by the variance-compensated adaptive Kalman filter model establishment module, and calculate to obtain the predicted value. Forecast covariance matrix D X(k丨k-1) The gain matrix J(k) is obtained; the observed data are input into the variance-compensated adaptive Kalman filter model, adaptive Kalman filtering is performed, and the best predicted value of the set of observations is obtained. and covariance matrix D X(k丨k) Simultaneously, the covariance matrix of dynamic noise is calculated; and iterative recursive calculations are performed to obtain the best prediction values ​​for that set after each iteration. and covariance matrix D X(k丨k) Continue until the iteration ends;

[0065] - Adaptive Kalman Filter Prediction Model Optimization Module: Based on the filter value obtained by the iterative recursive calculation module, the adaptive Kalman filter prediction model optimization module optimizes the variance-compensated adaptive Kalman filter model established by the variance-compensated adaptive Kalman filter model establishment module, and obtains the predicted values ​​of the two state variables, rock mass wave velocity and deformation modulus, of the system in the next state.

[0066] The present invention also provides a device for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering, comprising:

[0067] - At least one processor;

[0068] - At least one memory for storing at least one computer program;

[0069] The processor executes a computer program in the memory to implement the steps of the method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering as described above.

[0070] The present invention also provides a computer storage medium storing a computer program, which is executed by a computer to implement the steps of the method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering as described above.

[0071] The above specific embodiments are used to explain and illustrate the present invention, and are only preferred embodiments of the present invention, not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made to the present invention within the spirit and scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering, characterized in that: The method includes the following steps: S1. Based on the in-situ test data of the deformation of the rigid bearing plate of the rock mass in the dam foundation project and the test data of the acoustic wave test, the acoustic wave velocity V and deformation modulus E of the rock mass are obtained and used as the state vector X0 and its corresponding covariance matrix D0 of the adaptive Kalman filter model. S2. Based on the existing information, establish the covariance matrix D of the observation noise. Δ The covariance transition matrix D of dynamic noise Ω ; S3. Establish a variance-compensated adaptive Kalman filter model and determine the values ​​of the system's state transition matrix φ(k|k-1), dynamic noise matrix T(k|k-1), and observation matrix B(k); S4. Input the system's state transition matrix, dynamic noise matrix, and observation matrix; begin calculation to obtain the predicted value. Forecast covariance matrix D X(k丨k-1) and the gain matrix J(k); S5. Input a set of in-situ test data of rigid bearing plate deformation in rock mass and acoustic wave test data as observation data, perform adaptive Kalman filtering, and obtain the best predicted value of the set of observations. and covariance matrix D X(k丨k) Simultaneously, the covariance matrix D of the dynamic noise is calculated. Ω ; S6. Return to step S4 and proceed to the next iterative calculation. S7. Obtain the best prediction value for this set after each iteration. and covariance matrix D X(k丨k) This continues until the iteration count k reaches the last system state q, i.e., k = q; S8. Based on the filtered values ​​after iteration, optimize the adaptive Kalman filter prediction model in step S3 to obtain the predicted values ​​of the two state variables, rock mass wave velocity and deformation modulus, of the system in the next state.

2. The method for accurately predicting rock mass deformation modulus based on adaptive Kalman filtering according to claim 1, characterized in that: In step S1, the in-situ test data of rigid bearing plate deformation and the test data of acoustic wave test are used as input state variables. They are obtained from the collected in-situ test data of rigid bearing plate deformation and the test data of wave velocity test, including the acoustic wave velocity and deformation modulus of the rock mass.

3. The method for accurately predicting rock mass deformation modulus based on adaptive Kalman filtering according to claim 1, characterized in that: In steps S4 to S5, as new observation data is added, the noise covariance matrix is ​​dynamically adjusted to reflect the real-time changes in observation noise and the latest estimate of prediction error.

4. A rock mass deformation modulus accurate prediction system based on adaptive Kalman filtering, characterized in that: The system includes the following modules: - Test data acquisition module, the test data acquisition module is used to obtain the rock mass acoustic wave velocity V and deformation modulus E based on the in-situ test data of the deformation of the rigid bearing plate of the rock mass and the acoustic wave test data, and use them as the state vector X0 and its corresponding covariance matrix D0 of the adaptive Kalman filter model; - A variance-compensated adaptive Kalman filter model building module, which is used to build the covariance matrix D of the observation noise based on existing information. Δ The covariance transition matrix D of dynamic noise Ω A variance-compensated adaptive Kalman filter model is established to determine the values ​​of the system's state transition matrix φ(k|k-1), dynamic noise matrix T(k|k-1), and observation matrix B(k). - Iterative recursive calculation module, which is used to input data into the variance-compensated adaptive Kalman filter model established by the variance-compensated adaptive Kalman filter model establishment module, and calculate to obtain the predicted value. Forecast covariance matrix D X(k丨k-1) The gain matrix J(k) is obtained; the observed data are input into the variance-compensated adaptive Kalman filter model, adaptive Kalman filtering is performed, and the best predicted value of the set of observations is obtained. and covariance matrix D X(k丨k) Simultaneously, the covariance matrix D of the dynamic noise is calculated. Ω Then, iterative calculations are performed to obtain the best predicted value for that set after each iteration. and covariance matrix D X(k丨k) Continue until the iteration ends; - Adaptive Kalman Filter Prediction Model Optimization Module: Based on the filter value obtained by the iterative recursive calculation module, the adaptive Kalman filter prediction model optimization module optimizes the variance-compensated adaptive Kalman filter model established by the variance-compensated adaptive Kalman filter model establishment module, and obtains the predicted values ​​of the two state variables, rock mass wave velocity and deformation modulus, of the system in the next state.

5. A device for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering, characterized in that, The device includes: - At least one processor; - At least one memory for storing at least one computer program; The processor executes a computer program in the memory to implement the steps of the method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering as described in any one of claims 1-3.

6. A computer storage medium, characterized in that: The computer storage medium stores a computer program, which is executed by a computer to implement the steps of the method for accurately predicting the deformation modulus of rock mass based on adaptive Kalman filtering as described in any one of claims 1-3.

Images