Three-dimensional geo-stress inversion method based on strain gauge reliability and related equipment

Abstract

The application discloses a three-dimensional ground stress inversion method based on strain gauge reliability and related equipment, and the method comprises the following steps: obtaining the geometric and elastic mechanics parameters of surrounding rock and hollow package strain gauge to construct a forward model; determining the theoretical strain of each strain measurement channel according to the forward model; setting the channel state parameters corresponding to the normal and abnormal channel states for each strain measurement channel respectively; constructing the probability observation models corresponding to the normal and abnormal channel states for each strain measurement channel based on the residual error of the measured and theoretical strains to construct a global probability observation model; calculating the initial joint posterior distribution, updating each probability observation model and its weight to obtain a target global probability observation model; performing ground stress inversion based on the target global probability observation model and outputting the result; and the application can reduce the influence of abnormal strain gauges on the ground stress inversion, improve the accuracy and reliability of the ground stress inversion, and can be widely applied to the field of in-situ rock stress measurement.

Description

Technical Field

[0001] This application relates to the field of original rock stress measurement technology, and in particular to a three-dimensional geostress inversion method and related equipment based on strain gauge reliability. Background Technology

[0002] In-situ stress, existing within the Earth's crust, is the force per unit area within the medium caused by rock deformation. It is a crucial fundamental parameter for the design and stability evaluation of underground caverns, tunnels, mines, slopes, underground laboratories, and underground energy projects. Currently, the hollow-inclusion strain gauge method is commonly used in engineering to obtain directly observable strain data. Specifically, hollow-inclusion strain gauges with multiple strain plates in different directions are attached to the inner wall of a borehole. The rock core containing these strain gauges is then subjected to in-situ stress relief to measure strain changes in different directions before and after relief. Finally, the three-dimensional in-situ stress at the measuring point is inverted based on the rock's elastic parameters and the stress-strain conversion relationship.

[0003] However, during the stress relief process, the measurement quality of different strain gauges is inconsistent. For example, individual strain gauges may be affected by factors such as local debonding or insufficient bonding, rock mass fissures, local heterogeneity, drilling vibration, conductor contact, zero-point drift, temperature changes, instrument noise, strain gauge installation angle errors, and local fractures in the stress relief core, leading to anomalies. The final release value of abnormal strain gauges may deviate from the true response, and their relief curves may exhibit sudden jumps, continuous drift, asynchronous changes, or instability, thus significantly impacting the geostress inversion results. Currently, fixed residual thresholds or manual experience are generally used to remove abnormal data. However, the selection of thresholds relies on experience, leading to strong subjectivity and a tendency to incorrectly delete or omit data, thereby affecting the accuracy and reliability of the geostress inversion results. Summary of the Invention

[0004] The main objective of this application is to propose a three-dimensional geostress inversion method and related equipment based on strain gauge reliability, thereby reducing the impact of abnormal strain gauges on geostress inversion results and improving the accuracy and reliability of geostress inversion.

[0005] To achieve the above objectives, one aspect of this application proposes a three-dimensional geostress inversion method based on strain gauge reliability, comprising: Obtain the geometric parameters and elastic mechanical parameters of the surrounding rock and the hollow inclusion strain gauge, and construct the forward model of the hollow inclusion strain gauge based on the geometric parameters and the elastic mechanical parameters; Obtain the current values ​​of the geostress parameters to be inverted, and determine the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge based on the forward model and the current values. Each strain measurement channel is assigned a channel status parameter, which includes at least two values, corresponding to the normal channel status and the abnormal channel status, respectively. The measured strain of each strain measurement channel is obtained, and based on the residual between the measured strain and the theoretical strain of each strain measurement channel, a probabilistic observation model corresponding to the normal channel state and the abnormal channel state is constructed for each strain measurement channel. An initial global probabilistic observation model is constructed based on each of the probabilistic observation models, and the initial joint posterior distribution of the geostress parameters to be inverted and the channel state parameters is calculated based on the initial global probabilistic observation models. Each of the probability observation models is updated according to the initial joint posterior distribution, and the reliability weight of each of the probability observation models in the initial global probability observation model is updated to obtain the target global probability observation model. Based on the global probability observation model of the target, geostress inversion is performed, and the geostress inversion results are output.

[0006] In some embodiments, the step of constructing an initial global probabilistic observation model based on each of the probabilistic observation models, and calculating the initial joint posterior distribution of the geostress parameter to be inverted and the channel state parameter based on the initial global probabilistic observation model specifically includes: Obtain the strain relief curve for each strain measurement channel; For each strain measurement channel, extract at least one curve quality feature based on the corresponding strain relief curve; Based on a preset logic function and at least one of the curve quality characteristics, the anomalous prior probability of each strain measurement channel is determined, and the initial reliability weight of each strain measurement channel is determined based on the anomalous prior probability. An initial global probability observation model is constructed based on the initial reliability weights and each of the probability observation models. The geostress parameters to be inverted and the channel state parameters are taken as parameters to be solved and incorporated into the initial global probability observation model. The initial joint posterior distribution is obtained by combining the prior distributions corresponding to the geostress parameters to be inverted and the channel state parameters.

[0007] In some embodiments, updating each of the probability observation models according to the initial joint posterior distribution and updating the reliability weights of each of the probability observation models in the initial global probability observation model to obtain the target global probability observation model specifically includes: The initial joint posterior distribution is solved using a preset probability inference algorithm to obtain posterior samples of the geostress parameters to be inverted and the channel state parameters; The ratio of the channel state parameter to the sample corresponding to the abnormal channel state in the posterior sample of each strain measurement channel is determined as the abnormal posterior probability of each strain measurement channel. The reliability weight of each probability observation model is determined based on the anomalous posterior probability. The effective observation variance of each probability observation model is determined based on the reliability weight, and each probability observation model is updated based on the effective observation variance to obtain the updated probability observation model. Based on each of the reliability weights and each of the updated probabilistic observation models, the updated global probabilistic observation model is determined; The geostress parameters to be inverted and the channel state parameters are incorporated into the updated global probability observation model, and the updated joint posterior distribution is obtained by combining the current values ​​and the prior distributions corresponding to the channel state parameters. The updated joint posterior distribution is solved using the preset probability inference algorithm to obtain the updated posterior samples of the geostress parameters to be inverted and the channel state parameters; The ratio of the channel state parameter to the sample corresponding to the abnormal channel state in the updated posterior sample corresponding to each strain measurement channel is determined as the updated abnormal posterior probability of each strain measurement channel. When the change in the mean of the joint posterior distribution, the change in the objective function value corresponding to the joint posterior distribution, or the change in the abnormal posterior probability satisfies a preset convergence condition, the currently updated global probability observation model is determined as the target global probability observation model. Otherwise, based on the updated abnormal posterior probability, the step of determining the reliability weight of each probability observation model based on the abnormal posterior probability is returned to be executed until the change in the mean of the joint posterior distribution, the change in the objective function value corresponding to the joint posterior distribution, or the change in the abnormal posterior probability satisfies the preset convergence condition.

[0008] In some embodiments, after determining the reliability weight of each of the probabilistic observation models based on the anomalous posterior probability, the method further includes: The current effective observation matrix is ​​determined based on the anomalous posterior probability of each strain measurement channel and the reliability weight of each probabilistic observation model; The identifiability of the current geostress results is determined based on the current valid observation matrix. If the current geostress result is determined to be identifiable, then the steps of determining the effective observation variance of each probability observation model according to the reliability weight, and updating each probability observation model according to the effective observation variance, are continued to obtain the updated probability observation model. If the current geostress result is determined to be unrecognizable, the iteration is terminated and a preset quality prompt is output. The preset quality prompt is used to indicate that the current geostress inversion result is unrecognizable.

[0009] In some embodiments, determining the current effective observation matrix based on the anomalous posterior probability of each strain measurement channel and the reliability weight of each probabilistic observation model specifically includes: Strain measurement channels with anomaly posterior probabilities higher than a preset range are identified as low-effective channels. The current effective observation matrix is ​​determined based on the reliability weights of the strain measurement channels other than the low-efficiency channels.

[0010] In some embodiments, the determination of the identifiability of the current geostress results based on the current valid observation matrix specifically includes: If the rank, minimum singular value, and condition number of the current effective observation matrix satisfy the preset identifiable conditions, then the current geostress result is determined to be identifiable. If the rank, minimum singular value, and condition number of the effective observation matrix do not meet the preset identifiable conditions, then the current geostress result is determined to be unidentifiable.

[0011] In some embodiments, another aspect of this application provides a three-dimensional geostress inversion device based on strain gauge reliability, comprising: The first construction module is used to obtain the geometric parameters and elastic mechanical parameters of the surrounding rock and the hollow inclusion strain gauge, and to construct the forward model of the hollow inclusion strain gauge based on the geometric parameters and the elastic mechanical parameters. The determination module is used to obtain the current value of the geostress parameter to be inverted, and to determine the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge according to the forward model and the current value. The setting module is used to set channel status parameters for each strain measurement channel. The channel status parameters include at least two values, which correspond to normal channel status and abnormal channel status, respectively. The second construction module is used to obtain the measured strain of each strain measurement channel, and based on the residual between the measured strain and the theoretical strain of each strain measurement channel, to construct a probabilistic observation model for each strain measurement channel corresponding to the normal channel state and the abnormal channel state respectively. The third construction module is used to construct an initial global probabilistic observation model based on each of the probabilistic observation models, and to calculate the initial joint posterior distribution of the geostress parameters to be inverted and the channel state parameters based on the initial global probabilistic observation models. An update module is used to update each of the probability observation models according to the initial joint posterior distribution, and to update the reliability weight of each of the probability observation models in the initial global probability observation model, so as to obtain the target global probability observation model. The inversion module is used to perform geostress inversion based on the target global probability observation model and output the geostress inversion results.

[0012] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0013] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the methods described above.

[0014] This application also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium and execute the computer instructions, causing the computer device to perform the aforementioned method.

[0015] The embodiments of this application include at least the following beneficial effects: This application provides a three-dimensional geostress inversion method and related equipment based on strain gauge reliability. This method obtains the geometric parameters and elastic mechanical parameters of the surrounding rock and the hollow inclusion strain gauge, and constructs a forward model of the hollow inclusion strain gauge based on the geometric and elastic mechanical parameters; obtains the current values ​​of the geostress parameters to be inverted, and determines the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge based on the forward model and the current values; sets channel state parameters for each strain measurement channel, with each channel state parameter including at least two values ​​corresponding to the normal channel state and the abnormal channel state, respectively; obtains the theoretical strain of each strain measurement channel... The measured strain of each strain measurement channel is collected, and based on the residuals between the measured and theoretical strains of each channel, a probabilistic observation model corresponding to the normal and abnormal channel states is constructed for each channel. An initial global probabilistic observation model is constructed based on each probabilistic observation model, and an initial joint posterior distribution of the stress parameters to be inverted and the channel state parameters is calculated based on the initial global probabilistic observation model. Each probabilistic observation model is updated based on the initial joint posterior distribution, and the reliability weight of each probabilistic observation model in the initial global probabilistic observation model is also updated to obtain the target global probabilistic observation model. Stress inversion is performed based on the target global probabilistic observation model, and the stress inversion results are output. This embodiment of the application identifies abnormal strain measurement channels by setting independent channel state parameters for each strain measurement channel, and can adaptively adjust the weights of multiple channels, thereby reducing the impact of abnormal strain measurement channels on the stress inversion results and improving the accuracy and reliability of the inversion results. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the three-dimensional geostress inversion method based on strain gauge reliability provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of the hollow inclusion strain gauge provided in the embodiments of this application; Figure 3 This is a schematic diagram of the strain relief curve of the twelve-channel strain gauge provided in the embodiments of this application; Figure 4 This is a probability density function graph of the probability observation model provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of the three-dimensional geostress inversion device based on strain gauge reliability provided in the embodiments of this application; Figure 6 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0018] It is understood that the terms "first," "second," "third," "fourth," etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0019] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0021] Before providing a detailed description of the embodiments of this application, some related technologies involved in the embodiments of this application will be described first, as follows: In-situ stress is a crucial fundamental parameter for the design and stability evaluation of underground caverns, tunnels, mines, slopes, underground laboratories, and underground energy projects. The hollow-inclusion strain gauge method involves attaching hollow-inclusion strain gauges with multiple strain cells in different directions to the inner wall of a borehole, then removing the strain gauges from the core sample by inserting a strain gauge. The strain changes in different directions before and after removal are measured, and the three-dimensional in-situ stress at the measurement point is inverted based on the rock's elastic parameters and the stress-strain conversion relationship. Commonly used hollow-inclusion strain gauges typically contain 9, 12, or other strain measurement channels. To obtain a unique solution from the three-dimensional stress tensor with six independent components, an overdetermined observation equation formed by multi-strain gauge measurements is usually used, and the ordinary least squares method or weighted least squares method is employed for solution. Before calculation, this type of method generally assumes that all effective strain gauges conform to the same or preset measurement error, and technicians decide, based on experience, whether to delete outlier data.

[0022] However, existing methods typically have the following shortcomings: First, they treat all strain gauges as equally reliable, failing to consider the quality differences between different measurement channels. Individual anomalous channels may have an excessive impact on the inversion results of the six geostress components. Second, they use fixed residual thresholds or manual experience to remove anomalous data. The threshold selection lacks objective basis, easily leading to the deletion of valid data or the omission of anomalous data. Third, the sequential processing method of identifying anomalous data first and then calculating geostress ignores the coupling relationship between theoretical geostress and anomaly judgment—that is, the interdependence between geostress results influencing anomaly judgment, and anomaly judgment in turn influencing geostress results. Fourth, they only output a deterministic geostress result, failing to reflect the propagation of factors such as strain gauge measurement errors, rock elastic parameter errors, and installation angle errors to the geostress result. Fifth, when some strain gauges fail or data is missing, there is a lack of identifiable judgment on whether the remaining observations can still uniquely and stably identify the three-dimensional geostress. Sixth, they fail to fully utilize the information contained in the complete stress relief curve, such as stability, drift, abrupt changes, and inter-channel coordination, to comprehensively evaluate the reliability of each strain gauge.

[0023] Furthermore, although existing Bayesian borehole stress studies mainly focus on estimating the overall regression error and the posterior interval of stress, they cannot directly solve practical engineering problems such as "identifying which strain gauge is abnormal, what the probability of the abnormality is, how to continuously reduce the role of abnormal channels in the inversion, and whether the stress results can still be identified after some channels fail."

[0024] In view of this, this application provides a three-dimensional geostress inversion method and related equipment based on strain gauge reliability. This scheme sets channel state parameters for each strain measurement channel to represent the channel state of each channel. Based on these channel state parameters, probabilistic observation models are constructed corresponding to normal and abnormal channel states, respectively, enabling the identification of corresponding abnormal strain measurement channels. Then, the reliability weight of each strain measurement channel can be determined according to the probabilistic observation model, achieving strain gauge-by-strain reliability evaluation. Data from abnormal strain gauges is not simply retained or deleted entirely, but its contribution is reduced, thereby reducing the impact of abnormal strain measurement channels on the geostress inversion results, and thus improving the accuracy and reliability of the inversion results.

[0025] Please see Figure 1 , Figure 1 This is an optional flowchart of a three-dimensional geostress inversion method based on strain gauge reliability provided in this application embodiment. Figure 1 The method may include, but is not limited to, steps S101 to S107.

[0026] S101. Obtain the geometric parameters and elastic parameters of the surrounding rock and the hollow inclusion strain gauge, and construct the forward model of the hollow inclusion strain gauge based on the geometric parameters and elastic parameters.

[0027] Specifically, please refer to Figure 2 A hollow inclusion strain gauge is typically a thin-walled epoxy resin cylinder pre-filled with multiple sets of resistance strain gauges. Common hollow inclusion strain gauges generally contain 9, 12, or other numbers of strain gauges. These strain gauges are typically divided into three groups, arranged at 120° intervals around the circumference of the cylinder, corresponding to three different orientations of the borehole wall. The collection of strain gauges located at the same circumferential position but with different sensitive directions constitutes a strain rosette. For example... Figure 2 As shown, a common 9-element strain gauge typically consists of three strain gauges in each strain rosette: an axial strain gauge (0°) parallel to the borehole axis, a circumferential strain gauge (90°) perpendicular to the borehole axis, and an oblique strain gauge at a 45° or 135° angle to the axis. A common 12-element strain gauge, compared to the 9-element strain gauge, includes two oblique strain gauges in each strain rosette, in addition to the axial and circumferential strain gauges, and these two oblique strain gauges are perpendicular to each other.

[0028] Specifically, geometric parameters mainly include the geometric parameters of the borehole, the geometric parameters of the hollow inclusion strain gauge, and the geometric parameters of the strain gauge arrangement within the hollow inclusion strain gauge. Elastic parameters mainly include the elastic parameters of the rock, the elastic parameters of the hollow inclusion, and the elastic parameters of the coupling agent. Specifically, the geometric parameters of the borehole can include the borehole inner diameter, borehole depth, and orientation; the geometric parameters of the hollow inclusion can include the inner and outer diameters, wall thickness, and strain rosette embedding diameter; the geometric parameters of the strain gauge arrangement can include the number of strain rosette groups, the number and angle of each strain rosette group, the polar angle and axial strain rosette deflection angle of each strain rosette, and the position of the strain rosette along the axial direction of the inclusion.

[0029] In this embodiment, firstly, the hollow-encased strain gauge is installed inside the borehole to be measured, and the bonding, curing, and initial reading recording are completed. During the stress relief process of the borehole, the strain-time series of each strain gauge is continuously acquired, and key parameters such as strain gauge number, spatial orientation, measurement direction, strain gauge installation angle, borehole azimuth and dip angle, rock elastic modulus and Poisson's ratio, and sampling time are recorded. For each strain measurement channel, a complete strain relief curve is obtained, including the stable section before stress relief, the stress release section, and the stable section after stress relief. Figure 3 As shown, Figure 3 This is a schematic diagram of the strain release curves of the twelve-channel strain gauge provided in this application embodiment. The horizontal axis represents the drilling distance for casing release (unit: mm), and the vertical axis represents the measured strain value of each strain measurement channel (unit: με). The figure contains twelve sets of strain curves that vary with the release distance, corresponding to the twelve strain measurement channels. The number of each channel curve, its corresponding strain rosette group, and the measurement direction are indicated in the legend.

[0030] Overall, the curves of each channel exhibit a typical "three-stage" characteristic: in the initial stage when the release distance is small, the strain value remains relatively stable, which is the pre-release stable stage; as the release distance increases, the drilling approaches and crosses the strain gauge attachment position, and the strain value changes rapidly, which is the stress release stage; when the release distance exceeds a certain depth beyond the strain gauge position, the strain value gradually tends to a new stable state, which is the post-release stable stage. Finally, the difference between the strain value in the post-release stable stage and the strain value in the pre-release stable stage for each channel is the strain release amount for that channel.

[0031] In this embodiment, the geometric parameters of the hollow inclusion strain gauge mainly include the spatial orientation of the strain gauge, the measurement direction, the strain gauge installation angle, and the borehole azimuth and dip angle. The elastic mechanical parameters mainly include the rock elastic modulus and Poisson's ratio. Specifically, a forward model between three-dimensional geostress and the theoretical strain of each strain gauge can be established based on the geometric parameters and the rock elastic mechanical parameters. For example, a traditional hollow inclusion geostress calculation model or a finite element model can be used.

[0032] In this embodiment, the three-dimensional geostress is represented as six independent components: Then, based on the geometric parameters of the hollow inclusion strain gauge and the rock elasticity parameters, the unit response coefficients of the channel to the six geostress components are calculated, and the stress-strain transformation matrix is ​​obtained based on the stress response coefficient vectors corresponding to each strain measurement channel. Finally, the forward model is determined based on this stress-strain transformation matrix.

[0033] It is easy to understand that, in addition to the data that together constitute the basic constraints for geostress inversion, the acquired data may further include one or more of the following auxiliary data: a complete stress relief time series or strain curves varying with borehole footage; acquisition process parameters such as sampling time, drilling footage, drilling status, and reading stabilization time; temperature compensation data such as strain gauge temperature, borehole water temperature, or ambient temperature; surrounding rock mechanical parameters such as rock elastic modulus, Poisson's ratio, and its experimental dispersion; contact medium parameters such as cement layer thickness, colloidal elastic parameters, and hollow inclusion material parameters; and sensor and acquisition system parameters such as strain gauge sensitivity coefficient, initial resistance, channel calibration coefficient, and acquisition instrument range.

[0034] S102. Obtain the current values ​​of the geostress parameters to be inverted, and determine the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge based on the forward model and the current values.

[0035] In this process, the current values ​​of the inverted geostress parameters can be set to zero for all geostress components, and then gradually approximated to the true values ​​in subsequent iterations. Alternatively, the self-weight stress field can be used as the initial value, which is estimated based on the burial depth of the measuring point, the rock unit weight, and Poisson's ratio. Or, after constructing the forward model, a set of geostresses can be calculated using the least squares method. It's easy to understand that using this calculated geostress as the current value, as the starting point for iteration, can significantly accelerate the convergence speed.

[0036] Specifically, according to the forward model, for the th The theoretical strain of a strain gauge can be expressed as: .

[0037] in, For the first The theoretical strain of each strain gauge This is the stress response coefficient vector corresponding to the strain gauge. Auxiliary parameters that affect the response coefficient include rock elastic modulus, Poisson's ratio, strain gauge orientation angle, and strain gauge installation orientation.

[0038] Taking a twelve-channel strain gauge as an example, combining the twelve channels yields a matrix form: .in, It is a 12-row, 6-column stress-strain transformation matrix.

[0039] S103. Set channel status parameters for each strain measurement channel. The channel status parameters shall include at least two values, which correspond to the normal channel status and the abnormal channel status, respectively.

[0040] Specifically, a channel status parameter can be set for each strain measurement channel to represent the channel status of the corresponding strain measurement channel. In this embodiment, two values ​​are set for the channel status, representing whether the corresponding strain measurement channel is in a normal or abnormal state. It is worth noting that multiple values ​​can also be set for the channel status to represent the degree of abnormality of the corresponding strain measurement channel.

[0041] In this embodiment, a channel state variable is set for each strain measurement channel. .when When, it indicates the first Each strain gauge (strain measurement channel) is in normal channel condition; when When, it indicates the first One strain gauge is in an abnormal channel state.

[0042] S104. Obtain the measured strain of each strain measurement channel, and based on the residual between the measured strain and the theoretical strain of each strain measurement channel, construct a probabilistic observation model for each strain measurement channel corresponding to the normal channel state and the abnormal channel state, respectively.

[0043] Specifically, the measured strain can be obtained from the strain release curve. First, the strain release curves of each channel are preprocessed, which mainly includes zero-point correction, temperature compensation, sampling synchronization, preliminary identification of abnormal peaks, drift correction, and unit conversion. Then, for each strain measurement channel, the difference between the average strain of the stable segment after release and the average strain of the stable segment before release is taken as the final strain release amount of that channel.

[0044] In this embodiment, different probabilistic observation models are used for the normal channel state and the abnormal channel state. Under the normal channel state, the measured strain exhibits a small variance distribution around the theoretical strain; under the abnormal channel state, the measured strain is allowed to have a larger residual range. Its probabilistic observation model can be expressed as:

[0045] in, For the first The theoretical strain of each strain gauge For the first The measured strain of each strain gauge For channel offset parameters, This is the error scale under normal conditions. The abnormal error amplification factor is greater than 1. This indicates a normal distribution, a wide-tailed distribution, or other probability distributions with large variance that can accommodate abnormal residuals.

[0046] Under normal channel conditions, i.e., when the state variable is equal to 0, the measured strain obeys... The normal distribution. Under abnormal channel conditions, i.e., with the state variable equal to 1, the measured strain follows a normal distribution. The variance is greater than that of a normal distribution, a wide-tailed distribution, or other probability distributions that can accommodate abnormal residuals.

[0047] Using the hybrid observation model corresponding to the two channel states described above, strain measurement channels with large residuals between theoretical and measured strain are not immediately deleted, but have a higher probability of being classified as anomalous states, thus automatically reducing their impact on the posterior results of geostress. Compared to fixed threshold rejection, the hybrid observation model provided in this embodiment can reflect the degree of uncertainty of strain measurement channels being in anomalous states.

[0048] Specifically, please refer to Figure 4 , Figure 4 This diagram illustrates the probability density function of the probabilistic observation model provided in this application embodiment. The horizontal axis represents the residual between the measured strain value and the theoretical strain value, and the vertical axis represents the probability density. The solid line represents an example of the probability density curve for the normal measurement channel, where the probability density is concentrated near the zero residual, indicating that the measured strain and theoretical strain of this channel match well, and the measurement data is highly reliable. The dashed line represents an example of the probability density curve for the abnormal measurement channel, where the curve is flatter, and the probability density is lower over a wider residual range, indicating that the measurement uncertainty of this channel is large and its reliability is low.

[0049] S105. Construct an initial global probabilistic observation model based on each probabilistic observation model, and calculate the initial joint posterior distribution of the geostress parameters and channel state parameters to be inverted based on the initial global probabilistic observation model.

[0050] It is easy to understand that the three-dimensional geostress tensor usually contains six independent stress components, while each strain measurement channel only provides strain observation in one direction. After completing the construction of the probabilistic observation model for each strain measurement channel, it is necessary to combine the local probabilistic observation models of all channels to integrate the observation information in all directions and form a complete constraint on the six geostress components, so as to uniquely and stably invert the three-dimensional geostress.

[0051] Specifically, step S105 may include: acquiring the strain relief curve of each strain measurement channel; extracting at least one curve quality feature for each strain measurement channel based on the corresponding strain relief curve; determining the anomalous prior probability of each strain measurement channel based on a preset logic function and at least one curve quality feature, and determining the initial reliability weight of each strain measurement channel based on the anomalous prior probability; constructing an initial global probability observation model based on the initial reliability weight and each probability observation model; incorporating the geostress parameters to be inverted and the channel state parameters as parameters to be solved into the initial global probability observation model, and obtaining the initial joint posterior distribution by combining the prior distributions corresponding to the geostress parameters to be inverted and the channel state parameters.

[0052] To further improve the accuracy and robustness of abnormal strain measurement channel identification, this embodiment can extract one or more curve quality feature parameters from the stress relief curve to characterize data quality. These include, but are not limited to: the standard deviation of noise before relief, the standard deviation of the stable segment after relief, the linear drift rate, the number of jumps, the maximum single-step change, the effective sampling ratio, the correlation coefficient with the theoretical relief trend, the synchronization index with the same group of strain gauges, whether stability is restored after relief, and the degree of curve hysteresis. In specific implementations, these curve quality feature parameters can be used to calculate the prior probability of abnormal states in the strain measurement channel, or as auxiliary observation variables in the probabilistic observation model, thereby providing richer discrimination criteria for the adaptive update of channel reliability weights during iterative inversion.

[0053] Specifically, for the first Each channel is used to compose a vector from its curve quality characteristics. The prior probability that the channel is in an abnormal state is calculated using a logical function, i.e., the abnormal prior probability:

[0054] therefore:

[0055] in, For the first The prior probability that a strain gauge is in an anomalous state before the observation data is updated, i.e., the anomalous prior probability, β0 and β are parameters that are pre-calibrated, learned from historical data, or jointly estimated in the current inversion. This setting allows strain gauges with high noise before removal, continuous drift after removal, multiple abrupt jumps, or asynchronous with other channels in the same group to have a relatively high anomalous prior probability at the start of the inversion. However, the final anomaly judgment is still updated jointly by all strain data and the geostress forward modeling relationship.

[0056] Specifically, based on this prior probability of anomaly, the initial reliability weight for each strain measurement can be determined; for example, the initial reliability weight can be equal to... Specifically, the initial reliability weights reflect a preliminary assessment of the reliability of each channel based solely on the morphological characteristics of the release curve before obtaining measured strain data, providing reasonable initial weight values ​​for the construction of the initial global probability observation model.

[0057] In this embodiment, the prior state parameters of each strain measurement channel can be expressed as:

[0058] in, It can be set as a uniform constant, or it can be adaptively determined based on the quality characteristics of the stress relief curve.

[0059] This embodiment makes full use of the complete dynamic information of the stress relief curve to comprehensively evaluate the abnormal channel, avoiding the one-sidedness of judging the abnormal channel based solely on the final strain release value, and effectively improving the accuracy and stability of abnormal channel identification.

[0060] Specifically, the prior knowledge of geostress, i.e., the prior distribution of the geostress parameters to be inverted, can be set based on burial depth, self-weight stress, regional tectonic stress data, or results from nearby measuring points. The prior knowledge of rock and installation parameters can be determined based on laboratory tests and field installation records.

[0061] Specifically, the prior distributions of the channel bias parameters and the error scale parameters are used to characterize the initial understanding of the systematic bias and random noise level of the strain measurement channel before acquiring the measured data for this stress relief. The prior distribution of the error scale parameter can be determined based on the nominal accuracy of the strain acquisition equipment and the noise standard deviation of the stable segment data before stress relief. The prior distribution of the channel bias parameters can be determined based on the factory calibration data of the hollow inclusion strain gauge and the systematic bias range determined by historical installation experience.

[0062] Subsequently, the geostress parameters to be inverted will be... Rock and installation parameters Channel state variables Channel offset parameters and error scale parameters The initial joint posterior distribution is obtained by uniformly incorporating the parameters into the initial global probability observation model and combining the prior distribution of the above parameters: .

[0063] in, is the likelihood function of the probabilistic observation model, representing the probability of the current measured strain occurring given the inverted geostress parameters, rock and installation parameters, channel state parameters, channel bias parameters, and error scale parameters. For the prior art of inverting the geostress parameters, For the rock and installation parameters as a priori, As a priori for channel state parameters, The channel offset parameters are priors. The error scale parameter is a prior. According to Bayes' theorem, this initial joint posterior distribution is proportional to the product of the likelihood function and the prior distribution.

[0064] S106. Update each probability observation model according to the initial joint posterior distribution, and update the reliability weight of each probability observation model in the initial global probability observation model to obtain the target global probability observation model.

[0065] It is easy to understand that when constructing the initial joint posterior distribution, the measured strain data from all strain measurement channels, the prior distribution of the geostress parameters to be inverted, and the prior distribution of the channel state parameters are integrated together. This is equivalent to gathering the observation information and prior knowledge from all strain measurement channels, making a comprehensive global judgment on the current geostress state and the measurement quality of each strain measurement channel. Based on this global judgment, it is possible to assess whether the measurement data of each strain measurement channel is consistent with the overall trend and whether there are any anomalies, thus providing an objective basis for updating the probabilistic observation model and its reliability weights for that channel.

[0066] In some embodiments, step S106 may specifically include: solving the initial joint posterior distribution using a preset probabilistic inference algorithm to obtain posterior samples of the geostress parameters and channel state parameters to be inverted; determining the anomalous posterior probability of each strain measurement channel by the ratio of the channel state parameters to the samples corresponding to the anomalous channel states in the posterior samples corresponding to each strain measurement channel; determining the reliability weight of each probabilistic observation model based on the anomalous posterior probability; determining the effective observation variance of each probabilistic observation model based on the reliability weight, and updating each probabilistic observation model based on the effective observation variance to obtain the updated probabilistic observation model; determining the updated global probabilistic observation model based on each reliability weight and each updated probabilistic observation model; incorporating the geostress parameters and channel state parameters to be inverted into the updated global probabilistic observation model, and combining it with the prior distributions corresponding to the geostress parameters and channel state parameters to obtain... The updated joint posterior distribution is solved using a preset probabilistic inference algorithm to obtain updated posterior samples of the geostress parameters and channel state parameters to be inverted. The ratio of the channel state parameters to the samples corresponding to the abnormal channel states in the updated posterior samples corresponding to each strain measurement channel is determined as the updated abnormal posterior probability for each strain measurement channel. When the change in the mean of the joint posterior distribution, the change in the objective function value corresponding to the joint posterior distribution, or the change in the abnormal posterior probability meets the preset convergence condition, the currently updated global probabilistic observation model is determined as the target global probabilistic observation model. Otherwise, based on the updated abnormal posterior probability, the step of determining the reliability weight of each probabilistic observation model based on the abnormal posterior probability is returned until the change in the mean of the joint posterior distribution, the change in the objective function value corresponding to the joint posterior distribution, or the change in the abnormal posterior probability meets the preset convergence condition.

[0067] Specifically, a probabilistic inference algorithm can be used to solve the aforementioned initial joint posterior distribution to obtain posterior samples of the inverted geostress parameters and state parameters of each channel. It is easy to understand that the posterior samples are a large number of parameter value combinations randomly selected from the posterior distribution. Each set of samples represents a "potentially real geostress state," and the overall distribution pattern of the samples is completely consistent with the posterior distribution. Specifically, the preset probabilistic inference algorithm can be Markov chain Monte Carlo, Hamiltonian Monte Carlo, backtracking sampling, variational inference, Laplace approximation, sequential Monte Carlo, or other probabilistic inference algorithms.

[0068] In this embodiment, for the first Each strain gauge can be used to statistically analyze the posterior sample. The proportion is used to obtain the anomalous posterior probability of this strain measurement channel: .

[0069] Then, continuous reliability weights can be constructed based on the anomalous posterior probability: .

[0070] in, The value ranges from 0 to 1. The larger the value, the more reliable the strain gauge; The smaller the value, the lower the contribution of the channel to the geostress inversion results. Compared to existing methods that only have two extremes, "0" and "1," when removing outlier data—that is, setting the weight to 1 when the strain measurement channel is judged to be normal and setting the weight to 0 when the strain measurement channel is judged to be abnormal—this embodiment does not simply delete strain measurement channels in abnormal states completely, but continuously reduces their influence based on the probability of abnormality.

[0071] Specifically, after obtaining the reliability weight, the effective observation variance of the corresponding strain measurement channel can be determined based on this reliability weight. Specifically, the effective observation variance can be expressed as:

[0072] in, For the first Effective observation variance for each channel The original variance of the probabilistic observation model. The minimum weight is set to prevent numerical divergence. Specifically, this effective observation variance means that strain measurement channels with lower reliability weights have larger effective observation variances, thus weakening their impact in subsequent inversion processes.

[0073] In this embodiment, the stress parameters to be inverted and the channel state parameters are jointly updated iteratively. Specifically, after determining the reliability weight of each probabilistic observation model based on the anomaly posterior probability, the effective observation variance of each probabilistic observation model is determined based on the reliability weight. The original variance of the corresponding probabilistic observation model is replaced with the effective observation variance to obtain the updated probabilistic observation model. Then, based on each reliability weight and each updated probabilistic observation model, an updated global probabilistic observation model is determined. The stress parameters to be inverted and the channel state parameters are then used as the stress parameters to be inverted and incorporated into the updated global probabilistic observation model to obtain a new joint posterior distribution. Then, a preset probabilistic inference algorithm is used to solve the updated joint posterior distribution to obtain the stress parameters to be inverted and the updated posterior samples of each parameter. Based on the sample ratio of the channel state parameters and the anomaly channel states in the posterior samples, the updated anomaly posterior probability of the strain measurement channel is determined. If the change in the mean of the joint posterior distribution, the change in the objective function value corresponding to the joint posterior distribution, or the change in the abnormal posterior probability satisfies the preset convergence condition, then the iteration stops, and the currently updated global probability observation model is determined as the target global probability observation model. Otherwise, the steps of determining the reliability weight of each probability observation model based on the abnormal posterior probability, determining the effective observation variance of each probability observation model based on the reliability weight, and updating each probability observation model based on the effective observation variance are repeated to obtain the updated probability observation model.

[0074] It is easy to understand that the joint update achieves mutual constraints between channel anomaly detection and geostress calculation. Whether a strain measurement channel is anomaly depends not only on its absolute residual, but also on the geostress results supported by other strain measurement channels. The geostress results are no longer overly influenced by a single anomaly channel.

[0075] In some embodiments, after the step "determine the reliability weight of each probabilistic observation model based on the anomalous posterior probability", the method may further include: determining the current effective observation matrix based on the anomalous posterior probability of each strain measurement channel and the reliability weight of each probabilistic observation model; judging the identifiability of the current geostress result based on the current effective observation matrix; if the current geostress result is determined to be identifiable, then the method continues to determine the effective observation variance of each probabilistic observation model based on the reliability weight, and updates each probabilistic observation model based on the effective observation variance to obtain the updated probabilistic observation model; if the current geostress result is determined to be unidentifiable, then the iteration is terminated, and a preset quality prompt is output, which is used to indicate that the current geostress inversion result is unidentifiable.

[0076] It's easy to understand that some hollow inclusion strain gauges have fewer measurement channels, such as nine-channel hollow inclusion strain gauges. A nine-channel strain gauge only has nine equations to solve for six unknown geostress components, resulting in fewer redundant observations. Therefore, when channels become abnormal or data is missing, the observation matrix is ​​more likely to become ill-conditioned. Therefore, in each iteration, after determining the reliability weights of each strain measurement channel based on the anomaly posterior probability, before updating the probabilistic observation model based on these reliability weights, the effective observation matrix can be calculated first. The identifiability of the current geostress result can then be determined based on this effective observation matrix. This avoids continuing ineffective iterations when the current geostress result is no longer identifiable. It's easy to understand that the identifiability of the current identifiable result refers to whether, under the current anomaly posterior probability and reliability weights, the geostress result can be uniquely determined by a reliable measurement channel.

[0077] Specifically, the preset quality prompts can be "unidentifiable", "only some components are identifiable", or "remeasurement is recommended".

[0078] In some embodiments, the step "determine the current effective observation matrix based on the anomalous posterior probability of each strain measurement channel and the reliability weight of each probabilistic observation model" specifically includes: identifying strain measurement channels with anomalous posterior probabilities higher than a preset range as low-effective channels; and determining the current effective observation matrix based on the reliability weights corresponding to strain measurement channels other than low-effective channels.

[0079] Specifically, the anomaly posterior probability can be understood as the probability that the strain measurement channel is in an abnormal state. In this embodiment, the preset range can be a judgment standard pre-set by humans or the system. When the anomaly posterior probability is higher than this judgment standard, the data provided by the measurement channel can be considered unreliable. Therefore, it can be marked as a low-efficiency channel, and when constructing the effective observation matrix, it can be constructed based on strain measurement channels other than the low-efficiency channels. Specifically, the current effective observation matrix can be calculated according to the following formula: .

[0080] in, This is a diagonal matrix composed of the reliability weights of each strain measurement channel. This is the stress-strain transformation matrix.

[0081] Specifically, after identifying strain measurement channels with abnormal posterior probabilities exceeding a preset range as low-efficiency channels, if the current geostress result is determined to be identifiable and iteration can continue, the use of measured strain from low-efficiency channels to update the joint posterior distribution is stopped in subsequent iterations. Specifically, the reliability weight corresponding to low-efficiency channels can be set to 0, or they can be directly discarded, preventing them from participating in subsequent inversion calculations. This effectively avoids the continuous interference of abnormal or low-quality strain data on the inversion process, preventing the repeated introduction of "bad data" during iterations, thereby significantly improving the accuracy of geostress inversion results. Simultaneously, eliminating unreliable channels while ensuring identifiability reduces computational complexity and improves iteration convergence speed while retaining sufficient effective information.

[0082] In some embodiments, the step "determining the identifiability of the current geostress result based on the current valid observation matrix" may specifically include: if the rank, minimum singular value, and condition number of the current valid observation matrix meet the preset identifiability conditions, then the current geostress result is determined to be identifiable; if the rank, minimum singular value, and condition number of the current valid observation matrix do not meet the preset conditions, then the current geostress result is determined to be unidentifiable.

[0083] In this embodiment, identifiability can be determined based on the rank, minimum singular value, and condition number of the current effective observation matrix. The rank of the effective observation matrix is ​​used to determine whether the basic requirement of independent solution for the six geostress components is met. The minimum singular value is used to determine whether the observation sensitivity corresponding to each stress component is sufficient. The condition number is used to determine the numerical stability of geostress solution based on the effective observation matrix; the smaller the condition number, the weaker the amplification effect of small perturbations in the observation data on the geostress solution results, and the more stable the solution results.

[0084] In this embodiment, an effective observation matrix is ​​formed based on the effective strain measurement channels (strain measurement channels other than the low-effective channels). For a typical three-dimensional symmetric stress tensor, the geostress parameters to be inverted have six independent components, therefore it is necessary to determine: Secondly, calculation The condition number and minimum singular value of the weighted equations are used to determine whether an effective strain measurement channel can stably distinguish the six stress components. The preset identifiable condition is used to identify whether an effective strain measurement channel can stably distinguish the six stress components. When the preset identifiable condition is not met, this embodiment does not forcibly output the geostress inversion result, but instead outputs a preset quality prompt indicating that the current geostress inversion result is unidentifiable. For example, it can output quality prompts such as "unidentifiable," "only some components are identifiable," or "re-measurement recommended." Therefore, this embodiment can not only continue geostress inversion when some channels fail, but also avoid forcibly providing unreliable stress results when the observation direction is insufficient.

[0085] In some embodiments, when acquiring the measured strain of each strain measurement channel, if measured strain data cannot be obtained, the corresponding strain measurement channel can be directly marked as a low-effective channel and excluded from subsequent calculations. Alternatively, the reliability weight of the corresponding strain measurement channel can be directly set to zero. For other strain measurement channels where valid data can be acquired, subsequent calculation steps continue. For example, if some strain gauges cannot provide valid strain release due to acquisition failure, wire damage, signal saturation, or data loss, the corresponding strain measurement channel will not participate in subsequent calculations, or its reliability weight can be directly set to zero.

[0086] S107. Perform geostress inversion based on the target global probabilistic observation model and output the geostress inversion results.

[0087] The straightforward process involves iteratively converging to obtain the target global probabilistic observation model. Then, combining this with the prior distribution of the geostress parameters to be inverted, the final geostress inversion is performed, calculating the final posterior distribution of these parameters. The final geostress inversion result can then be determined and output based on this final posterior distribution. For example, the posterior mean can be extracted from the posterior distribution as the final geostress inversion result.

[0088] In some embodiments, after obtaining the geostress inversion results, the inverted stress tensor can be transformed to the engineering coordinate system to calculate the magnitude and direction of the maximum principal stress, intermediate principal stress, and minimum principal stress. The output geostress inversion results may include: posterior statistical information of the geostress components in the borehole coordinate system, including the posterior mean, median, standard deviation, and confidence interval; detailed evaluation information for each strain measurement channel, including anomaly posterior probability, reliability weight, residuals, and anomaly cause indications; and overall system diagnostic information, including the number of effective channels, the rank and condition number of the effective observation matrix, posterior prediction error, and measurement quality level.

[0089] The measurement quality level can be determined comprehensively based on factors such as the number of effective channels, the maximum anomaly probability, the posterior relative standard deviation of ground stress, the condition number of the observation matrix, and the posterior prediction error. For example, it can be divided into four levels: high confidence, reliable, use with caution, and recommended retesting. Specific grading thresholds can be determined by equipment calibration, historical engineering data, or industry requirements.

[0090] It is easy to understand that posterior statistical information can quantify the uncertainty of the geostress inversion results, providing richer reference for subsequent engineering design. Detailed evaluation information from each strain measurement channel can assist technicians in quickly locating abnormal measurement channels and analyzing the causes of anomalies. Through the output of data at various levels, this embodiment not only provides definitive geostress results but also provides posterior uncertainty quantification for each component, detailed reliability assessments of each strain measurement channel, and overall system diagnostic information, providing complete and traceable data support for engineering decisions.

[0091] In summary, the three-dimensional geostress inversion method based on strain gauge reliability provided in this application obtains the geometric and elastic parameters of the surrounding rock and the hollow inclusion strain gauge. A forward model of the hollow inclusion strain gauge is constructed based on these parameters. Then, the current values ​​of the geostress parameters to be inverted are obtained. Based on the forward model and the current values, the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge is determined. Next, channel state parameters are set for each strain measurement channel, including at least two values ​​corresponding to the normal and abnormal channel states, respectively. Finally, the measured strain of each strain measurement channel is obtained, and a model is constructed for each strain measurement channel based on the residual between the measured and theoretical strain. Probabilistic observation models corresponding to normal and abnormal channel states are generated. Then, an initial global probabilistic observation model is constructed based on each probabilistic observation model. The initial joint posterior distribution of the geostress parameters to be inverted and the channel state parameters is calculated based on the initial global probabilistic observation model. Then, each probabilistic observation model is updated based on the initial joint posterior distribution, and the reliability weight of each probabilistic observation model in the initial global probabilistic observation model is updated to obtain the target global probabilistic observation model. Finally, geostress inversion is performed based on the target global probabilistic observation model, and the geostress inversion results are output. This method can identify abnormal strain measurement channels and adaptively adjust the weights of multiple channels, thereby reducing the impact of abnormal strain measurement channels on geostress inversion results and improving the accuracy and reliability of inversion results.

[0092] Please see Figure 5 This application also provides a three-dimensional geostress inversion device based on strain gauge reliability, which can implement the above method. The device includes the following modules: The first construction module 11 is used to obtain the geometric parameters and elastic mechanical parameters of the surrounding rock and the hollow inclusion strain gauge, and to construct the forward model of the hollow inclusion strain gauge based on the geometric parameters and elastic mechanical parameters. The determination module 12 is used to obtain the current values ​​of the geostress parameters to be inverted, and to determine the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge based on the forward model and the current values. Setting module 13 is used to set channel status parameters for each strain measurement channel. The channel status parameters include at least two values, which correspond to the normal channel status and the abnormal channel status, respectively. The second construction module 14 is used to obtain the measured strain of each strain measurement channel, and based on the residual between the measured strain and the theoretical strain of each strain measurement channel, to construct a probabilistic observation model corresponding to the normal channel state and the abnormal channel state for each strain measurement channel. The third construction module 15 is used to construct an initial global probabilistic observation model based on each probabilistic observation model, and to calculate the initial joint posterior distribution of the geostress parameters and channel state parameters to be inverted based on the initial global probabilistic observation model. The update module 16 is used to update each probability observation model according to the initial joint posterior distribution and update the reliability weight of each probability observation model in the initial global probability observation model to obtain the target global probability observation model. Inversion module 17 is used to perform geostress inversion based on the target global probability observation model and output the geostress inversion results.

[0093] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0094] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described three-dimensional geostress inversion method based on strain gauge reliability. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0095] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0096] Please see Figure 6 , Figure 6 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 to implement the three-dimensional geostress inversion method based on strain gauge reliability of the embodiments of this application. The 903 input / output interface is used to implement information input and output. The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904); The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0097] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described three-dimensional geostress inversion method based on strain gauge reliability.

[0098] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0099] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0100] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0101] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0102] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0103] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0104] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0105] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0106] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0107] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0108] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A three-dimensional geostress inversion method based on strain gauge reliability, characterized in that, include: Obtain the geometric parameters and elastic mechanical parameters of the surrounding rock and the hollow inclusion strain gauge, and construct the forward model of the hollow inclusion strain gauge based on the geometric parameters and the elastic mechanical parameters; Obtain the current values ​​of the geostress parameters to be inverted, and determine the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge based on the forward model and the current values. Each strain measurement channel is assigned a channel status parameter, which includes at least two values, corresponding to the normal channel status and the abnormal channel status, respectively. The measured strain of each strain measurement channel is obtained, and based on the residual between the measured strain and the theoretical strain of each strain measurement channel, a probabilistic observation model corresponding to the normal channel state and the abnormal channel state is constructed for each strain measurement channel. An initial global probabilistic observation model is constructed based on each of the probabilistic observation models, and the initial joint posterior distribution of the geostress parameters to be inverted and the channel state parameters is calculated based on the initial global probabilistic observation models. Each of the probability observation models is updated according to the initial joint posterior distribution, and the reliability weight of each of the probability observation models in the initial global probability observation model is updated to obtain the target global probability observation model. Based on the global probability observation model of the target, geostress inversion is performed, and the geostress inversion results are output.

2. The three-dimensional geostress inversion method based on strain gauge reliability according to claim 1, characterized in that, The step of constructing an initial global probabilistic observation model based on each of the probabilistic observation models, and calculating the initial joint posterior distribution of the geostress parameters to be inverted and the channel state parameters based on the initial global probabilistic observation models, specifically includes: Obtain the strain relief curve for each strain measurement channel; For each strain measurement channel, extract at least one curve quality feature based on the corresponding strain relief curve; Based on a preset logic function and at least one of the curve quality characteristics, the anomalous prior probability of each strain measurement channel is determined, and the initial reliability weight of each strain measurement channel is determined based on the anomalous prior probability. An initial global probability observation model is constructed based on the initial reliability weights and each of the probability observation models. The geostress parameters to be inverted and the channel state parameters are taken as parameters to be solved and incorporated into the initial global probability observation model. The initial joint posterior distribution is obtained by combining the prior distributions corresponding to the geostress parameters to be inverted and the channel state parameters.

3. The three-dimensional geostress inversion method based on strain gauge reliability according to claim 2, characterized in that, The step of updating each of the probability observation models according to the initial joint posterior distribution, and updating the reliability weights of each of the probability observation models in the initial global probability observation model to obtain the target global probability observation model, specifically includes: The initial joint posterior distribution is solved using a preset probability inference algorithm to obtain posterior samples of the geostress parameters to be inverted and the channel state parameters; The ratio of the channel state parameter to the sample corresponding to the abnormal channel state in the posterior sample of each strain measurement channel is determined as the abnormal posterior probability of each strain measurement channel. The reliability weight of each probability observation model is determined based on the anomalous posterior probability. The effective observation variance of each probability observation model is determined based on the reliability weight, and each probability observation model is updated based on the effective observation variance to obtain the updated probability observation model. Based on each of the reliability weights and each of the updated probabilistic observation models, the updated global probabilistic observation model is determined; The geostress parameters to be inverted and the channel state parameters are incorporated into the updated global probability observation model, and the updated joint posterior distribution is obtained by combining the prior distributions corresponding to the geostress parameters to be inverted and the channel state parameters. The updated joint posterior distribution is solved using the preset probability inference algorithm to obtain the updated posterior samples of the geostress parameters to be inverted and the channel state parameters; The ratio of the channel state parameter to the sample corresponding to the abnormal channel state in the updated posterior sample corresponding to each strain measurement channel is determined as the updated abnormal posterior probability of each strain measurement channel. When the change in the mean of the joint posterior distribution, the change in the objective function value corresponding to the joint posterior distribution, or the change in the abnormal posterior probability satisfies a preset convergence condition, the currently updated global probability observation model is determined as the target global probability observation model. Otherwise, based on the updated abnormal posterior probability, the step of determining the reliability weight of each probability observation model based on the abnormal posterior probability is returned to be executed until the change in the mean of the joint posterior distribution, the change in the objective function value corresponding to the joint posterior distribution, or the change in the abnormal posterior probability satisfies the preset convergence condition.

4. The three-dimensional geostress inversion method based on strain gauge reliability according to claim 3, characterized in that, After determining the reliability weight of each probability observation model based on the anomalous posterior probability, the method further includes: The current effective observation matrix is ​​determined based on the anomalous posterior probability of each strain measurement channel and the reliability weight of each probabilistic observation model; The identifiability of the current geostress results is determined based on the current valid observation matrix. If the current geostress result is determined to be identifiable, then the steps of determining the effective observation variance of each probability observation model according to the reliability weight, and updating each probability observation model according to the effective observation variance, are continued to obtain the updated probability observation model. If the current geostress result is determined to be unrecognizable, the iteration is terminated and a preset quality prompt is output. The preset quality prompt is used to indicate that the current geostress inversion result is unrecognizable.

5. The three-dimensional geostress inversion method based on strain gauge reliability according to claim 4, characterized in that, The step of determining the current effective observation matrix based on the anomalous posterior probability of each strain measurement channel and the reliability weight of each probabilistic observation model specifically includes: Strain measurement channels with anomaly posterior probabilities higher than a preset range are identified as low-effective channels. The current effective observation matrix is ​​determined based on the reliability weights of the strain measurement channels other than the low-efficiency channels.

6. The three-dimensional geostress inversion method based on strain gauge reliability according to claim 4, characterized in that, The step of determining the identifiability of the current geostress results based on the current effective observation matrix specifically includes: If the rank, minimum singular value, and condition number of the current effective observation matrix satisfy the preset identifiable conditions, then the current geostress result is determined to be identifiable. If the rank, minimum singular value, and condition number of the current valid observation matrix do not meet the preset conditions, then the current geostress result is determined to be unidentifiable.

7. A three-dimensional geostress inversion device based on strain gauge reliability, characterized in that, include: The first construction module is used to obtain the geometric parameters and elastic mechanical parameters of the surrounding rock and the hollow inclusion strain gauge, and to construct the forward model of the hollow inclusion strain gauge based on the geometric parameters and the elastic mechanical parameters. The determination module is used to obtain the current value of the geostress parameter to be inverted, and to determine the theoretical strain of multiple strain measurement channels in the hollow inclusion strain gauge according to the forward model and the current value. The setting module is used to set channel status parameters for each strain measurement channel. The channel status parameters include at least two values, which correspond to normal channel status and abnormal channel status, respectively. The second construction module is used to obtain the measured strain of each strain measurement channel, and based on the residual between the measured strain and the theoretical strain of each strain measurement channel, to construct a probabilistic observation model for each strain measurement channel corresponding to the normal channel state and the abnormal channel state respectively. The third construction module is used to construct an initial global probabilistic observation model based on each of the probabilistic observation models, and to calculate the initial joint posterior distribution of the geostress parameters to be inverted and the channel state parameters based on the initial global probabilistic observation models. An update module is used to update each of the probability observation models according to the initial joint posterior distribution, and to update the reliability weight of each of the probability observation models in the initial global probability observation model, so as to obtain the target global probability observation model. The inversion module is used to perform geostress inversion based on the target global probability observation model and output the geostress inversion results.

8. An electronic device, characterized in that, It includes a processor and a memory, the memory being used to store a program, and the processor executing the program to implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

Images