Ground stress field inversion method and device based on Bayesian optimization and medium

Abstract

The invention discloses a crustal stress field inversion method and device based on Bayesian optimization and a medium. The inversion efficiency and precision can be improved. The method comprises the following steps: acquiring parameters generated by an agent model, wherein the parameters comprise boundary parameters and physical and mechanical parameters; the prior probability of Bayesian model parameters is obtained through calculation according to the parameters; simulation stress is obtained according to the parameter simulation, and a Bayesian model parameter likelihood probability is obtained based on the simulation stress and the corresponding actually measured stress; obtaining an objective function based on the prior probability and the likelihood probability, and adding the objective function and corresponding parameters into an optimization data set; adding the parameters and the objective function into a sample set of an agent model to update the agent model, and circulating after new parameters are obtained; and finally, determining a minimum value of an objective function in the optimized data set, and performing simulation based on parameters corresponding to the objective function to obtain a crustal stress field.

Description

Technical Field

[0001] The present invention belongs to the technical field of geostress inversion, and in particular relates to a geostress field inversion method, device and medium based on Bayesian optimization. Background Art

[0002] Field testing of geostress is expensive and time-consuming, and the number of tests is limited. It is difficult to grasp the geostress state of the study area based on a small amount of measured data. Therefore, based on limited measurement data, the use of numerical simulation technology and inverse analysis methods is currently a common way to obtain geostress fields. The invention patent with application number 202210216125.2 discloses an "accelerated optimization inversion method for initial geostress field under complex geological conditions". The inversion method is specifically as follows: establish a three-dimensional geological model for stress field inversion calculation; determine the application method and range value of the tectonic stress boundary, and then obtain the stress calculation value of the measuring point under each stress boundary condition through numerical calculation; use the stress boundary condition as the network input and the stress calculation value of the measuring point as the network output, establish the learning sample of the proxy model, and bring the learning sample into the proxy model for training; based on the trained proxy model, use the genetic algorithm to quickly search for the optimal solution within the trained sample range to obtain the optimal stress boundary condition; invert the optimal stress boundary condition to obtain the initial geostress field within the regional range.

[0003] However, the existing method has the following problems:

[0004] 1. The inversion of the existing geostress field sets the physical and mechanical parameters of the rock mass as constants, and the setting is artificially subjective, so the inversion result of the geostress field is not accurate.

[0005] 2. The inversion efficiency of the existing geostress field is low. Summary of the invention

[0006] In order to solve the problem that the artificial setting of physical and mechanical parameters in the existing geostress inversion results in low accuracy and low inversion efficiency, the present invention proposes a geostress field inversion method, device and medium based on Bayesian optimization, which can improve the inversion efficiency and accuracy.

[0007] The objective of the present invention is achieved through the following technical solutions:

[0008] The first aspect of the present invention discloses a method for inversion of geostress field based on Bayesian optimization, comprising the following steps:

[0009] Step A: obtaining parameters generated by the proxy model and initializing the number of cycles, wherein the parameters include boundary parameters and physical and mechanical parameters, and the physical and mechanical parameters include elastic modulus;

[0010] Step B: Calculating the prior probability of the Bayesian model parameters according to the parameters;

[0011] Step C: obtaining a simulated stress according to the parameter simulation and obtaining a Bayesian model parameter likelihood probability based on the simulated stress and the corresponding measured stress;

[0012] Step D: obtaining an objective function based on the prior probability and the likelihood probability and adding the objective function and its corresponding parameters to the optimization data set;

[0013] Step E: updating the number of cycles and determining whether the number of cycles is greater than or equal to a threshold;

[0014] Step F: in response to the number of cycles being less than a threshold, adding the parameters and the objective function to the sample set of the proxy model to update the proxy model, and then looping step BE after obtaining new parameters;

[0015] Step G: In response to the number of cycles being greater than or equal to a threshold, determining the minimum value of the objective function in the optimization data set and performing simulation based on the parameters corresponding to the objective function to obtain the geostress field.

[0016] The second aspect of the present invention discloses a geostress field inversion device based on Bayesian optimization, comprising a memory and a controller which are communicatively connected in sequence, wherein a computer program is stored in the memory, and the controller is used to read the computer program and execute the geostress field inversion method based on Bayesian optimization described in the first aspect.

[0017] The third aspect of the present invention discloses a computer-readable storage medium having instructions stored thereon. When the instructions are run on a computer, a Bayesian optimization-based geostress field inversion method described in the first aspect and any possible design thereof is executed.

[0018] The beneficial effects of the present invention are:

[0019] 1. By adopting the method of this scheme, physical and mechanical parameters are included in the optimization scope, and experimental information and engineering experience are incorporated in a priori form, the accuracy of inversion of the geostress field can be improved.

[0020] 2. Using the method of this scheme, the inversion optimization is carried out based on the Bayesian optimization method. In the iterative optimization process, historical data and prior information are used to guide the search process, which greatly improves the inversion efficiency of the geostress field. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0026] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0027] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, or the positions or positional relationships in which the product of the invention is usually placed when in use, or the positions or positional relationships commonly understood by those skilled in the art, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0028] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0029] The first aspect of the present invention discloses a geostress field inversion method based on Bayesian optimization, which can be performed by, but is not limited to, a geostress field inversion device, which can be software or a combination of software and hardware, and which can be integrated in smart devices such as smart mobile terminals, tablets, and computers. Figure 1 As shown, the method includes steps A to G. It should be noted that the step identifiers in this solution are only for the convenience of explaining the method and do not constitute a limitation on the sequence. The sequence of each step is based on its language description and the sequence of each signal.

[0030] Step A: Obtain parameters generated by the proxy model and initialize the number of cycles, wherein the parameters include boundary parameters and physical and mechanical parameters, wherein the physical and mechanical parameters include elastic modulus, and the boundary parameters are stress boundary condition parameters or displacement boundary condition parameters.

[0031] The initial proxy model is trained by the initial sample set, which takes the parameters in the initial sample set as input and the objective function as output. The parameters in the initial sample set include boundary parameters and physical and mechanical parameters. A random sampling algorithm is used to extract t0 groups of samples from the parameter set search space. Call FLAC3D numerical simulation to calculate the simulated stress in sequence Then the objective function is calculated. FLAC3D (Fast Lagrangian Analysis of Continua) is a simulation software developed by ITASCA, USA.

[0032] Using t0 group samples After the initial proxy model is trained, the proxy model can generate the next set of parameters.

[0033] The optimization parameters in this scheme include not only the boundary parameter X1, but also the physical and mechanical parameter X2. The entire inversion process does not require subjective setting of physical and mechanical parameters. Through continuous optimization, the accuracy of physical and mechanical parameters is improved, and the accuracy of geostress field inversion is further improved. The physical and mechanical parameters include at least one of elastic modulus, Poisson's ratio, cohesion, and internal friction angle.

[0034] The more types of physical and mechanical parameters that are added for optimization, the higher the accuracy of the inversion of the geostress field. However, the corresponding calculation efficiency will be lost. Therefore, when taking into account both accuracy and calculation efficiency, when there is only one physical and mechanical parameter, the elastic modulus is preferred; when there are two physical and mechanical parameters, the physical and mechanical parameters include the elastic modulus and Poisson's ratio.

[0035] Step B: Calculate the prior probability of the Bayesian model parameters based on the parameters.

[0036] Parameter X includes boundary parameter X1 and physical and mechanical parameter X2. The boundary parameter prior is determined based on the researcher's subjective experience of the regional geostress formation process and influencing factors. Generally speaking, there is a lack of reference information for boundary parameters, and their possible range can only be roughly estimated. Therefore, weak information prior is specified for the boundary parameters.

[0037] Assuming boundary parameters Any parameter x i The value range is x i ∈(a i ,b i ), N1 is the number of boundary parameters, ai is the lower boundary of the boundary parameter, bi is the upper boundary of the boundary parameter, and T represents transposition. ai is the lower boundary of the boundary parameter, and bi is the upper boundary of the boundary parameter, which can be set according to experience. Within the reference range, the parameters obey a uniform distribution; when the sampling exceeds the reference range, it is truncated. In summary, the boundary parameter x i The probability density function can be expressed as:

[0038]

[0039] According to basic geological knowledge, it is assumed that each boundary parameter condition is independently distributed, so the parameter set The joint probability density function can be expressed as:

[0040]

[0041] The physical and mechanical parameters X2 are specified a priori based on physical and mechanical tests, engineering analogies, or experience. ( and X1∪X2=X) any parameter x j , assuming it follows a Gaussian distribution Physical and mechanical parameters x j The mean μ j and variance Determined based on empirical information, its probability density function is as follows:

[0042]

[0043] Assuming that the physical and mechanical parameters of each rock mass are independent of each other, the parameter set The joint probability density function can be expressed as:

[0044]

[0045] Where N 1+1 is the number of the first physical and mechanical parameter; N is the total number of boundary parameters and physical and mechanical parameters.

[0046] The parameter set X1 follows a uniform distribution, and the parameter set X2 follows a Gaussian distribution, and they are independent events. Therefore, the joint probability density distribution of the parameter set X for the inversion of the geostress field can be expressed as:

[0047]

[0048] The prior probability P(X) of the Bayesian model parameters can be obtained.

[0049] Step C: obtain the simulated stress according to the parameter simulation and obtain the Bayesian model parameter likelihood probability P (σ m |X).

[0050] Assuming that the measurement error of ground stress obeys a normal distribution with zero mean, the likelihood function is defined in the form of Gaussian probability density:

[0051]

[0052] Among them, G i (Θ,X), and σ i are the simulated stress, measured stress and standard deviation of the i-th parameter, and the standard deviation is set according to the confidence level of the simulation error.

[0053] Step D: Obtain an objective function based on the prior probability and the likelihood probability and add the objective function and its corresponding parameters to the optimization data set.

[0054] make is k field test values ​​of geostress, X=[x1,x2,…x N ] T Optimize parameters. According to Bayes' theorem, the model parameter P(X|σ m ) can be expressed as:

[0055]

[0056] According to Bayes' theorem, to maximize the posterior probability P(X|σ m ) to solve the parameter X to achieve the inversion of the geostress field. That is:

[0057]

[0058] Combining the prior probability formula and the posterior probability formula, in order to follow the optimization principle and ensure numerical stability, the negative logarithm of the maximized posterior probability is taken as the objective function of model optimization:

[0059]

[0060] Among them, the objective function is:

[0061]

[0062] Among them, β i , β j is the coefficient constant and C is a constant.

[0063] Step E: Update the number of cycles and determine whether the number of cycles is greater than or equal to a threshold.

[0064] The threshold value of the number of cycles can be set according to the specific situation.

[0065] Step F: In response to the number of loops being less than a threshold, the parameters and the objective function are added to the sample set of the proxy model to update the proxy model, and after obtaining new parameters, steps BE are repeated until the number of loops is greater than or equal to a threshold.

[0066] Step G: In response to the number of cycles being greater than or equal to a threshold, determining the minimum value of the objective function in the optimization data set and performing simulation based on the parameters corresponding to the objective function to obtain the geostress field.

[0067] Specifically, after determining the parameters corresponding to the minimum value of the objective function in this step, FLAC3D is called to perform forward calculation to obtain the optimal geostress field.

[0068] By adopting the method of this scheme, physical and mechanical parameters are included in the optimization scope, and experimental information and engineering experience are incorporated in a priori form, the accuracy of the inversion of the geostress field can be improved.

[0069] With the method of this scheme, the inversion optimization is carried out based on the Bayesian optimization method. In the iterative optimization process, historical data and prior information are used to guide the search process, which greatly improves the inversion efficiency of the geostress field.

[0070] The second aspect of the present invention discloses a ground stress field inversion device based on Bayesian optimization, including a memory and a controller connected in sequence, wherein a computer program is stored on the memory, and the controller is used to read the computer program and execute a ground stress field inversion method based on Bayesian optimization described in the first aspect. Specifically, the memory may include, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a flash memory (Flash Memory), a first-in-first-out memory (FIFO) and or a first-in-last-out memory (FILO), etc.; the controller may not be limited to a microcontroller of the STM32F105 series. In addition, the computer device may also include, but is not limited to, a power supply unit, a display screen and other necessary components.

[0071] The third aspect of the present invention discloses a computer-readable storage medium having instructions stored thereon. When the instructions are run on a computer, a Bayesian optimization-based geostress field inversion method described in the first aspect and any possible design thereof is executed.

[0072] The operating principles of the devices and media disclosed in the second and third aspects of the present invention are detailed in the first aspect and will not be repeated here.

[0073] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for inversion of geostress field based on Bayesian optimization, characterized in that: The following steps are involved: Step A: obtaining parameters generated by the proxy model and initializing the number of cycles, wherein the parameters include boundary parameters and physical and mechanical parameters, wherein the physical and mechanical parameters include elastic modulus, and the boundary parameters are stress boundary condition parameters or displacement boundary condition parameters; Step B: Calculating the prior probability of the Bayesian model parameters according to the parameters; Step C: obtaining a simulated stress according to the parameter simulation and obtaining a Bayesian model parameter likelihood probability based on the simulated stress and the corresponding measured stress; Step D: obtaining an objective function based on the prior probability and the likelihood probability and adding the objective function and its corresponding parameters to the optimization data set; Step E: updating the number of cycles and determining whether the number of cycles is greater than or equal to a threshold; Step F: in response to the number of cycles being less than a threshold, adding the parameters and the objective function to the sample set of the proxy model to update the proxy model, and then looping step BE after obtaining new parameters; Step G: In response to the number of cycles being greater than or equal to a threshold, determining the minimum value of the objective function in the optimization data set and performing simulation based on the parameters corresponding to the objective function to obtain the geostress field.

2. The inversion method of geostress field based on Bayesian optimization according to claim 1 is characterized in that: The physical and mechanical parameters also include at least one of Poisson's ratio, cohesion, and internal friction angle.

3. The inversion method of geostress field based on Bayesian optimization according to claim 1 is characterized in that: The prior probability of the Bayesian model parameters calculated according to the parameters is: Calculate the boundary parameter x based on the lower boundary, upper boundary, and number of boundary parameters i The joint probability density function P(X1), In the formula, a i is the lower boundary of the boundary parameter, b i is the upper boundary of the boundary parameter, N1 is the number of boundary parameters; Calculate the physical and mechanical parameters x j The joint probability density function P(X2), In the formula, μ j is the physical and mechanical parameter x j The mean of is the physical and mechanical parameter x j Variance of 1+1 is the number of the first physical and mechanical parameter; N is the total number of boundary parameters and physical and mechanical parameters; According to the boundary parameter x i The joint probability density function P(X1) and the physical and mechanical parameters x j The joint probability density function P(X2) of is used to calculate the prior probability P(X) of the Bayesian model parameters.

4. The inversion method of geostress field based on Bayesian optimization according to claim 1 is characterized in that: The simulated stress obtained according to the parameter simulation and the Bayesian model parameter likelihood probability obtained based on the simulated stress and the corresponding measured stress are: In the formula, P(σ m |X) is the likelihood probability of the Bayesian model parameters, G i (Θ,X), and σ i are the simulated stress, measured stress and standard deviation of the i-th parameter respectively.

5. The inversion method of geostress field based on Bayesian optimization according to claim 1 is characterized in that: The objective function obtained based on the prior probability and likelihood probability is: Where M(X) is the objective function, β i , β j is the coefficient constant and C is a constant.

6. A geostress field inversion device based on Bayesian optimization, comprising a memory and a controller that are communicatively connected in sequence, wherein a computer program is stored in the memory, and the controller is used to read the computer program and execute a geostress field inversion method based on Bayesian optimization as described in any one of claims 1-5.

7. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed on a computer, a method for inversion of a geostress field based on Bayesian optimization as described in any one of claims 1 to 5 is executed.

Images