Nonlinear Geostress Field Inversion Method and Apparatus Based on Physical Data Neural Networks

Abstract

This application provides a nonlinear geostress field inversion method and apparatus based on physical data neural networks. It is used to construct a corresponding backpropagation neural network to optimize the interpretation of the nonlinear geostress field boundary conditions as the optimization objective, analyze the nonlinear characteristics of the geostress field distribution caused by the complexity of the geological structure and the heterogeneity of mechanical parameters, and obtain the optimal nonlinear geostress field boundary conditions. This solves the problem of large errors and unreliable inversion results of nonlinear geostress field inversion in engineering practice.

Description

Technical Field

[0001] This application relates to the field of geostress field inversion, specifically to a nonlinear geostress field inversion method and apparatus based on physical data neural networks. Background Technology

[0002] In-situ stress field analysis can provide important data support for related geological and engineering analysis work. Typically, it can be used to advance further mechanical analysis based on the inverted in-situ stress field.

[0003] However, in reality, the nonlinear mechanical characteristics of the rock and soil medium and the existence of regional geological structures make it difficult for traditional linear geostress inversion methods to reveal the geostress field distribution characteristics of the study area.

[0004] More specifically, in specific application scenarios such as the simulation of surface subsidence caused by shallow mining, numerical analysis of the mechanical response of deep large underground tunnel networks or surrounding rock of large factories, and the numerical simulation of mechanical analysis based on regional geostress field inversion results, the traditional geostress field inversion analysis method is mostly based on elastic constitutive models, which contradicts the inelastic constitutive models used in the mechanical analysis process. This seriously restricts the effective application of geostress field inversion results.

[0005] In other words, existing geostress field inversion analysis methods have drawbacks such as large errors in the results and unreliable inversion results. Summary of the Invention

[0006] This application provides a nonlinear geostress field inversion method and apparatus based on physical data neural networks. It is used to construct a corresponding backpropagation neural network to optimize the interpretation of the nonlinear geostress field boundary conditions as the optimization objective, analyze the nonlinear characteristics of the geostress field distribution caused by the complexity of the geological structure and the heterogeneity of mechanical parameters, and obtain the optimal nonlinear geostress field boundary conditions. This solves the problem of large errors and unreliable inversion results of nonlinear geostress field inversion in engineering practice.

[0007] Firstly, this application provides a nonlinear geostress field inversion method based on physical data neural networks, the method comprising:

[0008] Based on the engineering geological data and measured geostress results of the study area, the geostress field inversion model was constructed and the traditional elastic constitutive model was used for multiple linear regression processing to determine the multiple linear regression geostress field calculation results of the initial boundary conditions.

[0009] On the basis of the orthogonal test table constructed by the initial boundary condition, the nonlinear constitutive model numerical simulation processing is respectively performed on each boundary condition, and the neural network data set is configured after the corresponding nonlinear constitutive model numerical simulation result is normalized;

[0010] The back propagation neural network is trained using the neural network data set as a training sample, and a trained neural network meeting the requirements of a loss function is obtained.

[0011] According to the boundary condition coefficients obtained by the calculation of the trained neural network, the nonlinear constitutive calculation is performed by substituting the model, the stress values of each measured point are inversed, and the root mean square error of the corresponding measured points is calculated. If the root mean square error of the measured points is not less than the threshold value, the corresponding data is added to the neural network data set for cyclic processing until the errors of all measured points meet the requirement that the error is not less than the threshold value. The current boundary condition coefficient is output, and the regional in-situ stress field is obtained by forward calculation.

[0012] In the second aspect, the application provides a nonlinear in-situ stress field inversion device based on physical data neural network, the device comprising:

[0013] The first processing unit is configured to determine the multiple linear regression in-situ stress field calculation result of the initial boundary condition by constructing an in-situ stress field inversion model and performing multiple linear regression processing of the traditional elastic constitutive model on the basis of the engineering geological condition data and the in-situ stress measurement results of the study area.

[0014] The second processing unit is configured to perform nonlinear constitutive model numerical simulation processing on each boundary condition on the basis of the orthogonal test table constructed by the initial boundary condition, and to configure the neural network data set after the corresponding nonlinear constitutive model numerical simulation result is normalized.

[0015] The third processing unit is configured to train the back propagation neural network using the neural network data set as a training sample, and to obtain a trained neural network meeting the requirements of a loss function.

[0016] The fourth processing unit is configured to perform nonlinear constitutive calculation by substituting the model according to the boundary condition coefficients obtained by the calculation of the trained neural network, to inverse the stress values of each measured point and to calculate the root mean square error of the corresponding measured points. If the root mean square error of the measured points is not less than the threshold value, the corresponding data is added to the neural network data set for cyclic processing until the errors of all measured points meet the requirement that the error is not less than the threshold value. The current boundary condition coefficient is output, and the regional in-situ stress field is obtained by forward calculation.

[0017] In the third aspect, the application provides a processing device comprising a processor and a memory, wherein the memory stores a computer program, and the processor executes the method provided in the first aspect of the application when invoking the computer program in the memory.

[0018] In a fourth aspect, the present application provides a computer readable storage medium, which stores a plurality of instructions, and the instructions are adapted to be loaded by a processor to execute the method provided in the first aspect of the present application.

[0019] From the above, the present application has the following beneficial effects:

[0020] In view of the high-precision nonlinear stress inversion requirement, the present application takes the nonlinear stress field boundary condition as the optimization target, constructs the corresponding back propagation neural network to perform optimization interpretation, analyzes the nonlinear characteristic law of the in-situ stress field distribution caused by the complexity of the stratum structure and the inhomogeneity of the mechanical parameters, and thus obtains the optimal nonlinear in-situ stress field boundary condition, so as to solve the problem of large error and unreliable inversion result of the nonlinear in-situ stress field inversion result in engineering practice. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0022] Figure 1 A flowchart of the nonlinear in-situ stress field inversion method based on the physical data neural network of the present application;

[0023] Figure 2 A logic diagram of the nonlinear in-situ stress field inversion method based on the physical data neural network of the present application;

[0024] Figure 3 A structural diagram of the nonlinear in-situ stress field inversion device based on the physical data neural network of the present application;

[0025] Figure 4 A structural diagram of the processing device of the present application. DETAILED DESCRIPTION

[0026] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0027] The terms "first", "second", and the like in the description and in the claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily indicate a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or device that includes a series of steps or modules does not necessarily have to be limited to only those steps or modules clearly listed, but can include other steps or modules that are not clearly listed or inherent to the process, method, product, or device. The naming or numbering of the steps appearing in the present application does not mean that the steps in the method flow must be performed in the time / logical order indicated by the naming or numbering, and the named or numbered flow steps can be changed in execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved.

[0028] The division of modules appearing in the present application is a logical division, and in actual application, there can be another division manner, for example, multiple modules can be combined or integrated in another system, or some features can be ignored or not executed, in addition, the coupling or direct coupling or communication connection between the displayed or discussed each other can be through some interface, the indirect coupling or communication connection between the modules can be electrical or other similar forms, which are not limited in the present application. In addition, the modules or sub-modules described as separate components can or can not be physically separated, can or can not be physical modules, or can be distributed to multiple circuit modules, and part or all of the modules can be selected according to actual needs to achieve the purpose of the present application scheme.

[0029] Before introducing the nonlinear stress field inversion method based on physical data neural network provided by the present application, first introduce the background content involved in the present application.

[0030] The nonlinear stress field inversion method, device and computer readable storage medium based on physical data neural network provided by the present application can be applied to a processing device, for taking the nonlinear stress field boundary condition as the optimization target, constructing the corresponding back propagation neural network to perform optimization interpretation, analyzing the nonlinear characteristic law of the in-situ stress field distribution caused by the complexity of the formation structure and the inhomogeneity of the mechanical parameters, so as to obtain the optimal nonlinear stress field boundary condition, so as to solve the problem that the nonlinear stress field inversion result error is large and the inversion result is not reliable in engineering practice.

[0031] The physical data neural network based nonlinear stress field inversion method mentioned in the present application can be executed by a physical data neural network based nonlinear stress field inversion device, or a server, a physical host or a user equipment (UE) or other types of processing devices integrated with the physical data neural network based nonlinear stress field inversion device. The physical data neural network based nonlinear stress field inversion device can be implemented in hardware or software, and the UE can be a terminal device such as a smartphone, a tablet computer, a notebook computer, a desktop computer or a personal digital assistant (PDA), and the processing device can be set in a device cluster.

[0032] It can be understood that the present application is based on existing data or data that has been collected, so the processing device executing the physical data neural network based nonlinear stress field inversion method of the present application or the corresponding application service of the physical data neural network based nonlinear stress field inversion method of the present application usually only needs to meet the required data processing capability, and the specific device type and device deployment form are flexible.

[0033] If the direct collection of the existing data mentioned above is also involved, the processing device needs to be further adapted in terms of software and hardware to have the corresponding data collection capability.

[0034] In addition, if the content display requirement of the processing process or the processing result is involved, the processing device can also be configured with a display screen (including a touch screen), and the processing device can also use an external display device or other devices with a display screen to complete the content display target.

[0035] Next, the physical data neural network based nonlinear stress field inversion method provided by the present application will be introduced.

[0036] Firstly, referring to Figure 1 , Figure 1 Fig. 1 shows a flowchart of the physical data neural network based nonlinear stress field inversion method of the present application. The physical data neural network based nonlinear stress field inversion method provided by the present application can specifically include the following steps S101 to S104:

[0037] Step S101, based on the engineering geological condition data and the in-situ stress measurement results of the study area, the initial boundary condition multiple linear regression stress field calculation result is determined by constructing a stress field inversion model and performing a traditional elastic constitutive model multiple linear regression processing.

[0038] It can be understood that the implementation of the scheme of the present application is focused on a pre-selected, determined certain research area, engineering area, in which case, the engineering geological condition data and the in-situ stress measurement results (i.e. the in-situ stress test results of the relevant measurement points) of the research area currently required to be subjected to the nonlinear stress field inversion can be obtained to preliminarily construct the in-situ stress field inversion model, and the traditional multivariate linear regression process of the elastic constitutive model is continued to be performed, so as to complete the preliminary determination of the initial boundary conditions of the in-situ stress field inversion model of the research area by the calculated multivariate linear regression in-situ stress field calculation results.

[0039] Among them, the data of the engineering geological condition data and the in-situ stress measurement results are common / often collected data in geological engineering, so they can usually be directly extracted from online systems and the like, of course, real-time collection and processing are not excluded, and the specific situation in the actual application of the scheme of the present application can be flexibly adjusted.

[0040] As an exemplary embodiment here, the corresponding test method or data source of the in-situ stress measurement results can specifically include the hydraulic fracturing method, the stress relief method, the stress recovery method and the acoustic emission method and the like in-situ stress test methods.

[0041] On the other hand, the types of the initial boundary conditions involved here can specifically include the stress boundary conditions and the displacement boundary conditions.

[0042] Further, as an exemplary embodiment, the construction of the in-situ stress field inversion model and the traditional elastic constitutive model multivariate linear regression processing here can specifically involve the following processing contents:

[0043] 1) The azimuth transformation is performed on the in-situ stress measurement results, the corresponding value, azimuth angle and inclination angle are converted to the coordinate system adopted in the numerical simulation, and the measured stress component values of each known point are obtained;

[0044] Specifically, the corresponding principal stress azimuth , inclination and principal stress value can be obtained from the in-situ stress measurement results, and the directional cosines between the principal stress axis and the axes of the coordinate system O-xyz are obtained according to the following formula in combination with the related theories of elastic mechanics:

[0045] ,

[0046] ,

[0047] ,

[0048] Among them, The azimuth of the measured principal stress is taken as 0°, with the clockwise direction as positive (+) and the counterclockwise direction as negative (-). The inclination of the measured principal stress is taken as positive (+) for the elevation angle and negative (-) for the depression angle. , and are the direction cosines of the principal stress axes with respect to the x, y and z axes, respectively.

[0049] Then, the six stress components in the x, y and z coordinates are further converted using the following coordinate conversion formula:

[0050]

[0051] It can be understood that in the exemplary embodiments herein, the data other than the ground stress is extracted from the engineering geological condition information obtained in the foregoing.

[0052] 2) Select the main influencing factors of the initial ground stress field, calculate the stress of each known point, and obtain the calculated stress component values of each known point;

[0053] It is generally believed that the initial ground stress field is a function of the following form:

[0054] ,

[0055] wherein, is the initial ground stress value, which exists in the form of six stress components, , and represent the coordinate position, , and are rock mass parameters, which are the elastic modulus, Poisson's ratio and unit weight, respectively, is the self-weight factor, , and are geological structure factors, is the temperature factor.

[0056] It is generally assumed that the rock mass within the research scope is in the elastic stage before construction, so that the elastic analysis can be taken for research, and on this basis, it is assumed that each factor is linearly superimposed, so that the initial ground stress value can be obtained:

[0057] ,

[0058] wherein, is the regression coefficient (also the target to be solved in the next step), For the selected regional stress field main influencing factors, usually selected factors such as displacement boundary constraint conditions, stress / displacement boundary conditions, cross-sectional coordinate system, X direction uniform action, Y direction uniform action, Z direction uniform action, X direction triangle action, Y direction triangle action, self-weight action, XY plane shear, XZ plane shear, YZ plane shear, self-weight force, etc. For observation error, it is a random variable, and it is considered that its expected value is zero when the number of observation data is large enough.

[0059] 3) On the basis of the measured stress component values and calculated stress component values of each known point, the corresponding regression coefficients are calculated using the least squares analysis method;

[0060] Based on the principle of multiple linear regression, the regression calculation value of the ground stress (i.e. the measured stress component value) and the stress calculation value calculated under each operating condition obtained by numerical simulation (i.e. the calculated stress component value) have the following relationship:

[0061] ,

[0062] Among them, is the observation point number, is the stress component mark, ranging from 1 to 6, is the regression calculation value of the first stress component of the first observation point, is the regression coefficient relative to the independent variable , is the factor of the sub-load, is the numerical calculation value of the first stress component of the first observation point, is the total number of factors including self-weight stress and structural stress.

[0063] If there are measurement points, then the residual sum of squares is:

[0064] ,

[0065] Among them, is the observation value of the stress component of the observation point , according to the least squares principle, the equation that makes is the minimum value is:

[0066] .

[0067] 4) According to the regression coefficient, re-calculate the numerical value on the model, get the initial stress field distribution, complete the determination of the multiple linear regression stress field calculation results of the initial boundary condition.

[0068] By solving the above matrix equation, the regression stress value of any point in the model can be obtained with undetermined coefficients, and then the calculated value calculated by numerical calculation under each working condition is superimposed, and the regression stress value of any point in the model Position can be obtained:

[0069] ,

[0070] According to the multiple linear regression stress field calculation results of the measuring points, the initial data of the subsequent analysis process is determined, and the corresponding data structure can be referred to as follows:

[0071] ,

[0072] Among them, part of the stress component inversion data of the measuring point, part of the boundary regression coefficient data, part of the root mean square error of the measured data and the inversion data of the measuring point.

[0073] Step S102, on the basis of the orthogonal test table constructed by the initial boundary condition, the nonlinear constitutive model numerical simulation processing is respectively carried out on each boundary condition, and the corresponding nonlinear constitutive model numerical simulation result is normalized, and the neural network data set is configured;

[0074] It can be understood that after the initial boundary condition is preliminarily determined (quantified by the corresponding boundary condition coefficient), an orthogonal test table can be constructed to better promote the following data processing.

[0075] Specifically, as an exemplary embodiment, the initial data of the orthogonal test table here can include 1) model measuring point stress inversion value, 2) boundary condition regression coefficient, 3) root mean square error of the inversion result and the measured result of the measuring point These three contents.

[0076] In addition, it should be noted that the present application starts from step S102 here, and involves a loop processing in the following step S104, therefore, the content of the orthogonal test table will also be updated with the loop processing, and therefore, the exemplary embodiment mentioned here refers to the initial data, and the three content items involved are not changed, and the content recorded in the content item will be dynamically changed.

[0077] It can be understood that for the numerical model, the boundary conditions of different positions are involved, which can be further classified into the above-mentioned two boundary conditions of stress boundary condition and displacement boundary condition.

[0078] At this time, the nonlinear constitutive model numerical simulation process of each boundary condition can be carried out on the numerical model, and after obtaining the nonlinear constitutive model numerical simulation result, normalization can be continued, and the neural network data set, that is, the training sample, can be configured according to the configuration scheme of the configured training sample, such as the proportion (such as 7:3) and the extraction mode (such as random extraction) of the training set and the test set, for the subsequent neural network training.

[0079] In addition, for the numerical model of the present application, that is, the nonlinear constitutive model (both steps S102 and subsequent step S104 are involved), as an exemplary embodiment, the Mohr-Coulomb model, the Drucker-Prager model, the strain hardening / softening model, the double yield surface model or the Hoek-Brown model can be specifically used.

[0080] Step S103, using the neural network data set as a training sample to train the back propagation neural network to obtain a trained neural network meeting the requirements of the loss function;

[0081] After obtaining the foregoing training sample, the specific network training work can be promoted, and the back propagation (Backpropagation, BP) neural network is trained in combination with the configured loss function.

[0082] Specifically, the present application also specially designs a configuration scheme for the loss function, which is helpful to promote the efficient and high-precision training effect.

[0083] Correspondingly, as an exemplary embodiment, the overall loss function used to train the back propagation neural network can be specifically expressed as follows:

[0084] ,

[0085] ,

[0086] ,

[0087] Among them, is the total loss, is the weight coefficient, is the physical loss term, is the data loss term, is the number of test set samples in the training sample, is the number of output data, is the jth real output value of the ith test set, The input for the i-th test set and combination The j-th predicted output value is combined. Optimal weight With bias The combination For the j-th physical constraint component of the boundary condition data, The measured ground stress results at the measuring points When combining input data The boundary condition prediction value output by the neural network.

[0088] Specifically, a backpropagation neural network consists of an input layer, multiple hidden layers, and an output layer. The input layer receives external data, the hidden layers perform nonlinear transformations on the data, and the output layer outputs the final result. Through information exchange between neurons, a nonlinear mapping between input and output is achieved. During data processing, for the first... The first hidden layer of the layer Each neuron, through forward propagation, follows the formula based on the input data. (The output of the input layer or the hidden layer above) uses weights. and bias After calculating the weighted sum, the activation function... Activation is the output of neurons in this layer. :

[0089] ,

[0090] After obtaining the outputs of each neuron in the output layer through forward propagation, the error between the output result and the true value is calculated according to the following formula. The first hidden layer of the layer Error signal of each neuron :

[0091] ,

[0092] in, The activation function outputs at the current neuron. The derivative value, These are the weights of neurons in this layer and the next layer. For the first Layer Error signals of each neuron.

[0093] During backpropagation, based on the error signal of the neurons in this layer... and the weights of the input data to the neurons in this layer and bias Update:

[0094] ,

[0095] ,

[0096] wherein, is the learning rate, is the input result of the current layer of neurons (i.e., the output result of the previous layer of neurons).

[0097] To obtain the optimal weight matrix and bias combination, so that the prediction result of the network is best fitted with the true value, a data-driven loss function is constructed to quantify the error between the two, and the optimal weight matrix and bias combination is determined by minimizing the loss function, as shown in the following formula:

[0098] ,

[0099] wherein, is the optimal weight and bias combination, is the data-driven loss function under the current weight matrix and bias combination , is the number of test set samples, is the number of output data, is the th test set, is the th true output value of the th test set, is the th predicted output value under the input and of the

[0100] th test set.

[0101] At the same time, the physical driving data corresponding to the minimum RMSE value is obtained, and the physical driving loss function value is calculated.

[0102] ,

[0103] wherein, represents the physical constraint result of the boundary condition data, represents the measured point stress data of the first elastic calculation result, represents the nonlinear numerical simulation calculation process, represents the error of the measured point stress value of the corresponding process.

[0104] ,

[0105] wherein, is made the minimum number of iterations, representing the corresponding nonlinear numerical simulation process.

[0106] Thus, there is the total loss mentioned above .

[0107] In this case, according to the calculated physical data double constraint loss results, the neural network weights and biases are updated by back propagation, and the iteration is repeated until the loss function is lower than the threshold value, so as to obtain the standard output conforming to the test data set and the physical information.

[0108] It is worth noting that in the design of the scheme of the present application, the so-called physical data neural network, or the physical data double-driven neural network, does not mean that the loss function here involves the physical loss term and the data loss term, and the loss function is only the convergence condition of the model, and its essence is to solve the nonlinear problem of the ground stress field, which is embodied in a series of data processing starting from the step S101 of obtaining the engineering geological condition data and the ground stress measurement results of the study area, or the underlying design idea behind a series of data processing starting from the step S101 of obtaining the engineering geological condition data and the ground stress measurement results of the study area, which corresponds to the processing architecture of the overall scheme, so as to achieve a high-precision nonlinear ground stress field inversion effect.

[0109] Step S104, according to the boundary condition coefficient calculated by the trained neural network, substitute into the model for nonlinear constitutive calculation, to invert each measured point stress value and calculate the corresponding measured point root mean square error, if the measured point root mean square error is not lower than the threshold value, the corresponding data is added to the neural network data set for cyclic processing, until all the errors of the measured points meet the requirement of not lower than the threshold value, output the current boundary condition coefficient, and calculate to obtain the regional ground stress field.

[0110] After the training of the back propagation neural network is completed, at this time, the nonlinear ground stress field inversion based on the physical data neural network specially designed in the present application can be promoted, the boundary condition coefficient determined by the trained neural network, i.e. the calculated boundary condition coefficient, can be directly substituted into the nonlinear constitutive model mentioned in the previous step S102 for nonlinear constitutive calculation to invert the stress value of each measured point.

[0111] Among them, it can be seen that the purpose of training the back propagation neural network is to calculate the boundary condition coefficient that best meets the previous situation (embodied by the training sample) through iterative search.

[0112] At this point, for each measured point, if the root mean square error of the measured stress value obtained by nonlinear constitutive calculation is lower than the threshold, it can be considered to meet the requirements. Otherwise, if it is not lower than the threshold, it can be considered to not meet the requirements, and it is necessary to return to the previous step S102 for loop / iterative calculation. This process continues until the root mean square error of the measured stress value obtained by nonlinear constitutive calculation for all measured points is lower than the threshold and meets the requirements.

[0113] Among them, the root mean square error of the measuring point is the root mean square error of the measured point. It follows the conventional calculation logic of root mean square error and is specifically calculated from the inverted value of the geostress obtained by nonlinear constitutive calculation and the measured value of the geostress.

[0114] Specifically, the root mean square error (RMSE) of the measurement points can be calculated using the following formula:

[0115] ,

[0116] in, Indicates data input The root mean square error, The number of measurement points. The number of stress components to be inverted. For the first The first measuring point Measured values ​​of each stress component For the first The first measuring point Inversion values ​​of stress components.

[0117] After the above-mentioned loop / iteration process, if the root mean square error of the measured stress values ​​obtained by nonlinear constitutive calculation of all measured points is lower than the threshold, the corresponding boundary condition coefficient can be recorded as the current boundary condition coefficient and output as the result. The final regional geostress field of the study area can be obtained through forward calculation, thus realizing the goal of nonlinear geostress field inversion based on physical data neural network.

[0118] In particular, the nonlinear constitutive model numerical simulation processing in step S102 above and the nonlinear constitutive calculation involved in step S104 here can be carried out by the finite difference method as an exemplary embodiment to advance the specific calculation processing.

[0119] Meanwhile, the calculated boundary conditions, such as the boundary condition coefficients and the current boundary condition coefficients, can specifically include boundary conditions in the X direction, Y direction, Z direction, XY plane shear, YZ plane shear, XZ plane shear, and self-weight. In addition, the corresponding boundary condition application types can specifically include uniform distribution type and triangular type.

[0120] It can be understood that the above-mentioned data can also be continuously obtained by means of Figure 2 The application shown is based on a logical schematic diagram of a nonlinear stress field inversion method of a physical data neural network, and is used to more intuitively understand the above-mentioned scheme content.

[0121] Finally, in general, for the demand of high-precision nonlinear stress inversion, the application takes the nonlinear stress field boundary condition as the optimization target, constructs the corresponding back propagation neural network for optimization interpretation, analyzes the nonlinear characteristic law of the in-situ stress field distribution caused by the complexity of the stratum structure and the inhomogeneity of the mechanical parameters, and thus obtains the optimal nonlinear in-situ stress field boundary condition, so as to solve the problem of large error and unreliable inversion result of the nonlinear in-situ stress field inversion in engineering practice.

[0122] The above is an introduction to the nonlinear stress field inversion method based on the physical data neural network provided by the application. In order to better implement the nonlinear stress field inversion method based on the physical data neural network provided by the application, the application also provides a nonlinear stress field inversion device based on the physical data neural network from the functional module perspective.

[0123] Reference Figure 3 , Figure 3 is a structural schematic diagram of the nonlinear stress field inversion device based on the physical data neural network, and in the application, the nonlinear stress field inversion device based on the physical data neural network 300 can specifically include the following structures:

[0124] The first processing unit 301 is configured to determine the multivariate linear regression in-situ stress field calculation result of the initial boundary condition by constructing an in-situ stress field inversion model and performing multivariate linear regression processing of a traditional elastic constitutive model on the basis of the engineering geological condition data and the in-situ stress measurement result of the research area;

[0125] The second processing unit 302 is configured to perform nonlinear constitutive model numerical simulation processing on each boundary condition respectively on the basis of the orthogonal test table constructed by the initial boundary condition, and configure a neural network data set after normalizing the corresponding nonlinear constitutive model numerical simulation result;

[0126] The third processing unit 303 is configured to train the back propagation neural network by using the neural network data set as a training sample, and obtain a trained neural network meeting the requirement of a loss function;

[0127] The fourth processing unit 304 is configured to substitute the boundary condition coefficient obtained by calculation of the trained neural network into the model to perform nonlinear constitutive calculation, to inverse each measured point stress value and calculate the corresponding measured point root mean square error, and if the measured point root mean square error is not lower than the threshold value, the corresponding data is added to the neural network data set for cyclic processing until the errors of all the measured points meet the requirement that the error is not lower than the threshold value, and the current boundary condition coefficient is output, and the regional in-situ stress field is obtained by forward calculation.

[0128] In an exemplary embodiment, the corresponding test methods of the in-situ stress measurement results include hydraulic fracturing method, stress relief method, stress recovery method and acoustic emission method.

[0129] The type of the initial boundary condition includes stress boundary condition and displacement boundary condition.

[0130] In yet another exemplary embodiment, the in-situ stress field inversion model is constructed and traditional elastic constitutive model multiple linear regression processing is performed, involving the following processing contents:

[0131] The azimuth transformation is performed on the in-situ stress measurement results, and the corresponding value, azimuth angle and inclination angle are converted to the coordinate system adopted in the numerical simulation to obtain the measured stress component values of each known point.

[0132] The main influencing factors of the initial in-situ stress field are selected, the stress of each known point is calculated, and the calculated stress component values of each known point are obtained.

[0133] On the basis of the measured stress component values and the calculated stress component values of each known point, the regression coefficient is calculated by using the least square analysis method.

[0134] According to the regression coefficient, the numerical calculation is performed on the model again to obtain the initial in-situ stress field distribution, and the determination of the multiple linear regression in-situ stress field calculation result of the initial boundary condition is completed.

[0135] In yet another exemplary embodiment, the initial data of the orthogonal test table includes three kinds of contents, i.e., the in-situ stress inversion value of the model measurement point, the boundary condition regression coefficient, and the root mean square error between the inversion result and the measured result of the measurement point.

[0136] In yet another exemplary embodiment, the total loss function used for training the back propagation neural network is expressed as follows:

[0137] ,

[0138] ,

[0139] ,

[0140] wherein, is the total loss, These are the weighting coefficients. For physical loss items, For data loss items, The number of test set samples in the training samples. To determine the number of output data, For the j-th true output value of the i-th test set, The input for the i-th test set and combination The j-th predicted output value is combined. Optimal weight With bias The combination For the j-th physical constraint component of the boundary condition data, The measured ground stress results at the measuring points When combining input data The boundary condition prediction value output by the neural network.

[0141] In yet another exemplary embodiment, the nonlinear constitutive model specifically employs a Mohr-Coulomb model, a Drucker-Prag model, a strain hardening / softening model, a double yield surface model, or a Hawke-Brown model.

[0142] In yet another exemplary embodiment, both the numerical simulation processing and the nonlinear constitutive model calculation are specifically performed using the finite difference method.

[0143] The calculated boundary condition coefficients and the specific contents of the current boundary condition coefficients include X-direction compression, Y-direction compression, Z-direction compression, XY-plane shear, YZ-plane shear, XZ-plane shear, and self-weight. The corresponding boundary condition application types include uniform distribution type and triangular type.

[0144] This application also provides a processing device from a hardware architecture perspective, see [link / reference]. Figure 4 , Figure 4 This diagram illustrates a structural schematic of the processing device of this application. Specifically, the processing device may include a processor 401, a memory 402, and an input / output device 403. The processor 401 executes the computer program stored in the memory 402 to implement, for example... Figure 1 The steps of the nonlinear geostress field inversion method based on physical data neural networks in the corresponding embodiments; or, when the processor 401 executes the computer program stored in the memory 402, it implements as follows: Figure 3 Corresponding to the functions of each unit in the embodiment, the memory 402 is used to store the functions executed by the processor 401 as described above. Figure 1 The computer program required for the nonlinear geostress field inversion method based on physical data neural networks in the corresponding embodiment.

[0145] For example, the computer program can be divided into one or more modules / units, one or more modules / units are stored in the memory 402 and executed by the processor 401 to complete the present application. One or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the computer device.

[0146] The processing device can include, but not limited to, the processor 401, the memory 402, the input and output device 403. Those skilled in the art can understand that the schematic is only an example of the processing device, and does not constitute a limitation on the processing device, which can include more or less components than the schematic, or combine certain components, or different components, for example, the processing device can also include a network access device, a bus, etc., and the processor 401, the memory 402, the input and output device 403 are connected through the bus.

[0147] The processor 401 can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The processor is the control center of the processing device, which connects various parts of the entire device through various interfaces and lines.

[0148] The memory 402 can be used to store computer programs and / or modules, and the processor 401 realizes various functions of the computer device by running or executing the computer programs and / or modules stored in the memory 402, and calling the data stored in the memory 402. The memory 402 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, etc.; the data storage area can store data created according to the use of the processing device, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state memory device.

[0149] The processor 401 is used to execute the computer program stored in the memory 402, and can specifically implement the following functions:

[0150] On the basis of the engineering geological condition data of the research area and the measured results of the ground stress, the initial boundary condition multivariate linear regression ground stress field calculation result is determined by constructing a ground stress field inversion model and performing multivariate linear regression processing of the traditional elastic constitutive model.

[0151] On the basis of the orthogonal test table constructed by the initial boundary condition, the numerical simulation of the nonlinear constitutive model is performed on each boundary condition, and the corresponding nonlinear constitutive model numerical simulation result is normalized and configured into a neural network data set.

[0152] The trained neural network that meets the requirements of the loss function is obtained by using the neural network data set as a training sample to train the back propagation neural network.

[0153] According to the boundary condition coefficients obtained by the calculation of the trained neural network, the nonlinear constitutive calculation is performed by substituting the model, the measured point stress values are inverted, and the corresponding measured point root mean square error is calculated. If the measured point root mean square error is not less than the threshold value, the corresponding data is added to the neural network data set for cyclic processing until the errors of all measured points meet the requirement that the error is not less than the threshold value. The current boundary condition coefficient is output, and the regional ground stress field is calculated.

[0154] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described nonlinear ground stress field inversion device based on physical data neural network, processing equipment and corresponding units thereof can be referred to as Figure 1 The description of the nonlinear ground stress field inversion method based on physical data neural network in the corresponding embodiment is not repeated here.

[0155] Those skilled in the art can understand that all or part of the steps in the various methods of the above-described embodiments can be completed by instructions, or controlled by related hardware by instructions, which can be stored in a computer readable storage medium and loaded and executed by a processor.

[0156] Therefore, the present application provides a computer readable storage medium, which stores a plurality of instructions capable of being loaded by a processor to execute the present application as Figure 1 The steps of the nonlinear ground stress field inversion method based on physical data neural network in the corresponding embodiment can be specifically operated by referring to Figure 1 The description of the nonlinear ground stress field inversion method based on physical data neural network in the corresponding embodiment is not repeated here.

[0157] The computer readable storage medium can include a read only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0158] Due to the instructions stored in the computer readable storage medium, the application can be implemented as Figure 1 According to the steps of the nonlinear stress field inversion method based on the physical data neural network in the corresponding embodiment, the application can be implemented as Figure 1 The beneficial effects of the nonlinear stress field inversion method based on the physical data neural network in the corresponding embodiment are described in detail in the foregoing description, which will not be repeated here.

[0159] The nonlinear stress field inversion method based on the physical data neural network, the device, the processing equipment and the computer readable storage medium provided by the application are described in detail above, and the principles and implementation modes of the application are described by applying specific examples. The above description of the embodiments is only used to help understand the core idea of the application; at the same time, for those skilled in the art, according to the idea of the application, the specific implementation mode and application range will be changed, and the above description should not be understood as a limitation of the application.

Claims

1. A nonlinear stress field inversion method based on physical data neural network, characterized in that, The method comprises: On the basis of engineering geological condition data and in-situ stress measurement results of a study area, a preliminary construction of an in-situ stress field inversion model is performed, and a preliminary determination of initial boundary conditions of the in-situ stress field inversion model is continued based on a calculated multi-linear regression in-situ stress field. On the basis of an orthogonal test table constructed by the initial boundary conditions, a numerical simulation processing of a non-linear constitutive model is performed on each boundary condition, and after a normalization of a corresponding numerical simulation result of the non-linear constitutive model, a neural network data set is configured, initial data of the orthogonal test table including three content items of in-situ stress inversion values of model measurement points, regression coefficients of boundary conditions, root mean square errors of measurement point inversion results and measured results, the content items being unchangeable, and contents specifically recorded in the content items being updated subsequently with a loop processing; The neural network data set is used as a training sample to train a back propagation neural network, and a trained neural network meeting a loss function requirement is obtained; According to boundary condition coefficients calculated by the trained neural network, a non-linear constitutive calculation is performed on a corresponding non-linear constitutive model of the numerical simulation processing of the non-linear constitutive model, to invert in-situ stress values of each measurement point and calculate a corresponding measurement point root mean square error, if the measurement point root mean square error is not lower than a threshold value, corresponding data is added to the content item in the orthogonal test table to perform the loop processing, until errors of all measurement points meet a requirement of not being lower than the threshold value, current boundary condition coefficients are output, and a regional in-situ stress field is obtained by calculation.

2. The method of claim 1, wherein, The corresponding test methods of the in-situ stress measurement results include a hydraulic fracturing method, a stress relief method, a stress recovery method and an acoustic emission method; Types of the initial boundary conditions include stress boundary conditions and displacement boundary conditions.

3. The method of claim 1, wherein, The multi-linear regression processing of the traditional elastic constitutive model involves the following processing contents: The in-situ stress measurement results are subjected to a direction transformation, corresponding values, azimuth angles and inclination angles are converted to a coordinate system adopted in numerical simulation, and in-situ stress component values of each known point are obtained; Main influencing factors of an initial in-situ stress field are selected, stresses of the known points are calculated, and calculated stress component values of the known points are obtained; Based on the measured stress component values and the calculated stress component values of the known points, regression coefficients are calculated by using a least square analysis method; Based on the regression coefficients, a numerical calculation is performed again on the in-situ stress field inversion model, an initial in-situ stress field distribution is obtained, and the determination of the multi-linear regression in-situ stress field calculation results of the initial boundary conditions is completed.

4. The method of claim 1, wherein, The overall loss function used in the training of the back propagation neural network is expressed as follows: ， ， ， wherein, is the total loss, is the weight coefficient, is the physical loss term, is the data loss term, is the number of test set samples in the training sample, is the number of output data, is the jth real output value of the ith test set, is the input of the ith test set and the combination of the jth predicted output value under the combination is the optimal weight and the combination of bias, is the jth physical constraint component of the boundary condition data, is the measured stress result at the measurement point is the boundary condition prediction value of the neural network output under the combination when the input data is the measured stress result at the measurement point.​ 5. The method of claim 1, wherein, The non-linear constitutive model specifically adopts a Mohr-Coulomb model, a Drucker-Prager model, a strain hardening / softening model, a double yield surface model or a Hoek-Brown model.

6. The method of claim 1, wherein, The numerical simulation processing of the non-linear constitutive model and the non-linear constitutive calculation specifically adopt a finite difference method. The specific content of the calculated boundary condition coefficient and the current boundary condition coefficient includes X-direction extrusion, Y-direction extrusion, Z-direction extrusion, XY plane shear, YZ plane shear, XZ plane shear and self weight, and the corresponding boundary condition application type includes uniform type and triangular type.

7. A device for nonlinear stress field inversion based on physical data neural networks, characterized in that, The device comprises: A first processing unit is configured to preliminarily construct a geostress field inversion model based on engineering geological condition data and measured results of geostress of a research area, and to preliminarily determine initial boundary conditions of the geostress field inversion model by continuing a traditional elastic constitutive model multiple linear regression process and using calculated results of the multiple linear regression geostress field; A second processing unit is configured to perform nonlinear constitutive model numerical simulation processing on each boundary condition based on an orthogonal test table constructed by the initial boundary conditions, and to configure a neural network data set after normalizing corresponding nonlinear constitutive model numerical simulation results, wherein initial data of the orthogonal test table includes three content items of model measurement point geostress inversion value, boundary condition regression coefficient, root mean square error of measurement point inversion result and measured result, the content items are unchangeable, and specific recorded contents in the content items are updated by subsequent cyclic processing; A third processing unit is configured to train a back propagation neural network using the neural network data set as a training sample to obtain a trained neural network meeting a loss function requirement; A fourth processing unit is configured to perform nonlinear constitutive calculation on a corresponding nonlinear constitutive model by nonlinear constitutive model numerical simulation processing using calculated boundary condition coefficients of the trained neural network to invert each measured point stress value and calculate a corresponding measurement point root mean square error, and to add corresponding data to the content item in the orthogonal test table if the measurement point root mean square error is not lower than a threshold value, and to perform the cyclic processing until all measurement point errors meet the requirement of not being lower than the threshold value, output current boundary condition coefficients, and obtain a regional geostress field by calculation.

8. A processing device, characterized by The device comprises a processor and a memory, the memory stores a computer program, and the processor executes the computer program in the memory to perform the method in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a plurality of instructions adapted to be loaded by the processor to execute the method in any one of claims 1 to 6.

Images