Method for calibrating magnetic susceptibility instrument based on standard well

Abstract

The invention relates to the technical field of geophysical logging instrument detection and calibration, particularly provides a method for calibrating a magnetic susceptibility instrument based on a standard well, and aims to solve the problems that a calibration method of a magnetic susceptibility instrument in the prior art has certain limitation, data comparability between different instruments is poor, a standardized calibration system is lacked and the like. In order to achieve the purpose, actually measured magnetic susceptibility of each layer is obtained, and a calibration function is constructed; and constructing a distribution robust optimization problem by taking a minimum weighted error as an optimization target, solving the optimization problem by taking a preset constraint condition as a constraint to obtain an optimal calibration function, and correcting each actually measured magnetic susceptibility by using the optimal calibration function. According to the method, the standard well is constructed, the calibration function is constructed by utilizing the actually measured magnetic susceptibility detected by the magnetic susceptibility instrument layer by layer and the corresponding working conditions, and then the actually measured magnetic susceptibility is corrected by utilizing the calibration function, so that the limitation that the conventional method depends on well site or laboratory small sample comparison is overcome; unified calibration of well logging magnetic susceptibility instruments of different models and different batches is achieved.

Description

Technical Field

[0001] This invention relates to the field of geophysical logging instrument testing and calibration technology, specifically providing a method for calibrating a magnetic susceptibility meter based on a standard well. Background Technology

[0002] Magnetic susceptibility meters are important measuring instruments in geophysical exploration, resource assessment, and environmental geological surveys. They are widely used to obtain the magnetic susceptibility of underground rock formations, aiding in the study of their composition, structure, and genetic characteristics. Magnetic susceptibility, as a crucial physical parameter characterizing the strength of rock magnetism, reflects the content and distribution of ferromagnetic minerals within the rock, and has significant applications in sedimentology, mineral exploration, and environmental geochemistry. Therefore, magnetic susceptibility meters typically require calibration to ensure the accuracy of the detected magnetic susceptibility.

[0003] Currently, magnetic susceptibility meters are typically calibrated using the following methods: 1) Field downhole comparison method: Repeated measurements are performed in different downhole areas, and the calibrator is calibrated by comparing the data with existing data in each area. However, this method is limited by the complex and variable downhole environment, with significant differences in geological conditions at different well sites, resulting in a lack of repeatability and universality in the comparison results. 2) Laboratory sample comparison method: The magnetic susceptibility meter is calibrated using standard laboratory rock samples. Although laboratory conditions are controllable, the volume of standard rock samples is limited, making it impossible to truly simulate the multi-layered rock formations and depth effects in the downhole environment, and thus failing to effectively reflect the instrument's performance in actual well logging. 3) Empirical curve correction method: An empirical curve is established to correlate formation magnetic susceptibility with iron content and mineral composition, and this curve is used to correct the well logging data. However, this method only has certain reference value in specific areas, relies excessively on local geological conditions, and is difficult to extend to different regions and complex formation environments.

[0004] Therefore, existing calibration methods all have certain limitations. They lack standardized calibration platforms and cannot provide repeatable calibration environments for different models and batches of magnetic susceptibility meters. Most existing calibration methods rely on specific laboratory conditions or geological characteristics of specific regions, resulting in poor universality and portability, and failing to meet the unified application needs across regions and devices. They also depend on researchers' experience or ad-hoc comparison methods, lacking a systematic and procedural calibration mechanism, which is detrimental to the formation of industry standards.

[0005] Accordingly, a new technical solution is needed in this field to solve the above problems. Summary of the Invention

[0006] The present invention aims to solve the above-mentioned technical problems, namely, to solve the problems that existing magnetic susceptibility meter calibration methods have certain limitations, poor data comparability between different instruments, and lack of standardized calibration system.

[0007] This invention provides a method for calibrating a magnetic susceptibility meter based on a standard well. The standard well includes a housing, within which a measurement channel is provided to allow the magnetic susceptibility meter to pass. Multiple layers of receiving spaces are provided within the housing along the extension direction of the measurement channel, and a standard dielectric material having a standard magnetic susceptibility can be placed within these receiving spaces.

[0008] The method includes:

[0009] The magnetic susceptibility of the standard dielectric material in each layer's containment space was measured using the aforementioned magnetic susceptibility meter.

[0010] The measured magnetic susceptibility in multiple ranges is used as discrete calibration samples and associated with their corresponding operating conditions to construct a calibration function for the magnetic susceptibility meter. The operating conditions include at least one of the following: ambient temperature, the descent speed of the magnetic susceptibility meter along the measurement channel, and the eccentricity of the geometric axis of the magnetic susceptibility meter relative to the central axis of the measurement channel.

[0011] With minimizing the weighted error as the optimization objective, a partial Brussels bar optimization problem with respect to the calibration function parameters is constructed, and the preset constraints are used as constraints for the optimization problem. The constrained partial Brussels bar optimization problem is solved to obtain the optimal calibration function.

[0012] The measured magnetic susceptibility is corrected using the optimal calibration function.

[0013] In the preferred embodiment of the above method, the calibration function is constructed as follows:

[0014] The establishment form is The mapping function, where K m f is the measured magnetic susceptibility. θ Let θ be a piecewise rational spline function with adjustable parameters, θ be the spline parameter set, and c be the operating conditions.

[0015] In the preferred embodiment of the above method, the preset constraint conditions are specifically shown in Equation 1 below:

[0016] ;

[0017] Among them, K m f is the measured magnetic susceptibility. θ It is a piecewise rational spline function with adjustable parameters.

[0018] In the preferred embodiment of the above method, the optimization problem of the split-bar is shown in Equation 2:

[0019] ;

[0020] Among them, w i For sample weights, For a robust loss function, Let δ be the Wasserstein distance, δ be the robust radius, and Q be the value that satisfies... The adversarial distribution within the distribution set, f is a weighted empirical distribution constructed based on multi-layer samples from the standard wells. θ Let K be a piecewise rational spline function with adjustable parameters, where θ is the set of spline parameters, and K is the set of parameters. mi The measured magnetic susceptibility of layer i is... For the operating conditions of layer i, K si Let be the standard magnetic susceptibility of layer i.

[0021] In the preferred embodiment of the above method, the sample weights The overall uncertainty of the synthesis based on layer i is determined as shown in Equation 3 below:

[0022] ;in, s i It is the standard deviation of repeated measures in the i-th layer, n i It is the number of repeated measurements in the i-th layer, u s,i The uncertainty is pre-calibrated for the i-th layer.

[0023] In a preferred embodiment of the above method, the method further includes:

[0024] Before correcting the measured magnetic susceptibility using the optimal calibration function, the difference between the measured magnetic susceptibility of each layer and the corresponding standard magnetic susceptibility is calculated.

[0025] Determine whether the absolute value of each difference is greater than the threshold;

[0026] When the absolute value of the difference is greater than the threshold, the measured magnetic susceptibility corresponding to the difference is corrected using the optimal calibration function.

[0027] In a preferred embodiment of the above method, the method further includes:

[0028] When applying the calibration function for correction, a confidence interval for each calibration value is provided simultaneously, wherein the confidence interval is obtained by quantifying the prediction result of the calibration function using a bootstrap method or variational inference method.

[0029] In the above technical solution, the standard well of this invention has multiple layers of standard medium material with standard magnetic susceptibility placed along the extension direction of the detection channel. First, the measured magnetic susceptibility of the standard medium material in each containment space is detected using a magnetic susceptibility meter, obtaining the measured magnetic susceptibility corresponding to each containment space. Then, these measured magnetic susceptibility ranges are used as discrete calibration samples and their corresponding operating conditions to construct a calibration function. A robust optimization problem is constructed with minimizing the weighted error as the optimization objective, making the calibration function curve insensitive to noise and random outliers, thus improving the stability of the algorithm. During the optimization process, the derivative of the function is constrained based on the monotonic physical relationship between magnetic susceptibility and magnetic response to avoid the non-physical phenomenon of the measured value increasing while the calibration value decreases. At the same time, the slope of the calibration function curve is constrained to ensure a smooth transition of the calibration function between layers. Then, the optimal calibration function is obtained by solving for it. This optimal calibration function is used to correct the measured magnetic susceptibility detected by the magnetic susceptibility meter, realizing the standardization of the magnetic susceptibility meter measurement data.

[0030] In a preferred embodiment of the above method, standard dielectric materials covering different magnetic susceptibility ranges are respectively disposed in multiple accommodating spaces, and the magnetic susceptibility of the standard dielectric materials disposed in each accommodating space is distributed in a stepwise manner, either increasing or decreasing along the axial direction of the measurement channel; and / or

[0031] The standard medium material is detachably disposed within the accommodating space.

[0032] In a preferred embodiment of the above method, the measuring channel is coaxially arranged with the housing; and / or

[0033] The standard well is also equipped with a constant temperature control module, which is configured to maintain a constant temperature in the standard well.

[0034] In a preferred embodiment of the above method, the outer shell is made of a non-magnetic material; and / or

[0035] An isolation layer is provided between two adjacent accommodating spaces, and the isolation layer is made of a non-magnetic material.

[0036] By employing the above technical solution, the standard medium material is detachably placed within the accommodating space. This allows for the replacement of the standard medium material in each accommodating space as needed, satisfying the accuracy correction requirements across the entire measurement range of different magnetic susceptibility meters and improving engineering practicality. By using non-magnetic materials for the outer shell and isolation layer, interference from external magnetic fields can be avoided, and mutual interference between the standard medium materials in each accommodating space within the shell can also be prevented. This ensures that even with long-term use, environmental interference will not affect the stability of the detection.

[0037] The technical effects that can be obtained by this invention are as follows:

[0038] (1) This invention provides a reusable standardized testing environment under surface conditions through the layered design of standard wells, overcoming the limitations of traditional methods that rely on complex downhole formations or small laboratory samples for comparison, and realizing unified calibration of logging magnetic susceptibility meters of different models and batches.

[0039] (2) The standard well of the present invention has multiple layers of accommodating spaces for placing standard medium materials along the extension direction of the measurement channel, and the magnetic susceptibility of the standard medium materials placed in each accommodating space increases progressively along the length of the measurement channel. This can simultaneously cover different ranges of weak, medium, and strong magnetic fields, and can satisfy the verification and correction of the measurement accuracy of the magnetic susceptibility meter across the entire range. Compared with the method of calibration under only a single magnetic susceptibility condition, the present invention can obtain more comprehensive performance evaluation results.

[0040] (3) The standard well of the present invention has a simple overall structure and low manufacturing cost. The outer shell and the isolation layers between each layer are made of non-magnetic materials to avoid magnetic interference between the external or internal layers. In addition, the filling material in each layer is a stable solid standard medium material that has been precisely calibrated in the laboratory and is not affected by the environment during long-term use. During testing, the instrument only needs to be lowered into the well body, avoiding cumbersome laboratory operations and complex mathematical correction processes.

[0041] (4) The number of layers and magnetic susceptibility range of the standard wells of this invention can be flexibly adjusted as needed, which can adapt to different regions and different research needs. This invention is not only applicable to instrument testing under laboratory conditions, but can also be extended to various application scenarios such as field exploration, scientific research and teaching, and instrument factory quality inspection, and has broad engineering application value.

[0042] (5) This invention uses the measured magnetic susceptibility and the corresponding standard magnetic susceptibility in different magnetic susceptibility ranges in the standard well detected by the magnetic susceptibility meter as calibration samples, and introduces monotonicity constraints, uncertainty weighting and split-blob bar optimization in the process of constructing the calibration function. This enables the obtained optimal calibration function to maintain continuity, consistency and physical rationality in the full magnetic susceptibility range. It can uniformly characterize the overall measurement response characteristics of the magnetic susceptibility meter in different magnetic susceptibility ranges, realize reliable evaluation and standardized calibration of instrument performance, and avoid the problems of mixed error sources and unstable calibration caused by repeated error correction in the prior art.

[0043] (6) The magnetic susceptibility meter calibrated based on the standard well can significantly reduce systematic errors and ensure that the data from different equipment and different logging environments have better comparability and consistency, which is conducive to promoting the formation of a unified standardized calibration system in the industry. Attached Figure Description

[0044] The preferred embodiments of the present invention are described below with reference to the accompanying drawings, in which:

[0045] Figure 1 This is a schematic diagram of the structure of a standard well according to an embodiment of the present invention.

[0046] Figure label:

[0047] 1. Outer shell; 11. Guide structure; 111. Through hole; 2. Measurement channel; 3. Support base; 4. Isolation layer; 5. Accommodation space; 6. Constant temperature control module. Detailed Implementation

[0048] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0049] It should be noted that in the description of this invention, terms such as "upper" and "lower" indicate directions or positional relationships based on the directions or positional relationships shown in the accompanying drawings. This is merely for ease of description and does not indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0050] Currently, the calibration methods for magnetic susceptibility meters have certain limitations. Most rely on specific laboratory conditions or geological characteristics of specific areas, as well as the experience of researchers or temporary comparison methods. They lack standardized calibration mechanisms, have poor universality and portability, and cannot meet the calibration needs of different models and batches of magnetic susceptibility meters.

[0051] Therefore, this invention provides a method for calibrating a magnetic susceptibility meter based on a standard well. The standard well includes a housing with a measurement channel within it allowing the magnetic susceptibility meter to pass through. Multiple layers of receiving spaces are arranged within the housing along the extension direction of the measurement channel, where a standard dielectric material with a standard magnetic susceptibility can be placed. The standard dielectric material placed in each receiving space within the standard well has at least a portion of different standard magnetic susceptibilities. Those skilled in the art can select appropriate standard dielectric materials to place in the corresponding receiving spaces based on the type of magnetic susceptibility meter and the magnetic susceptibility of the actual field to be measured. This allows the standard well to simulate the actual situation of the field to be measured and also covers the full magnetic susceptibility range of various types of magnetic susceptibility meters, thereby enabling more accurate calibration of the magnetic susceptibility meter. The specific structure of the standard well is detailed below and will not be repeated here.

[0052] The method of the present invention includes:

[0053] Step 1: Use a magnetic susceptibility meter to detect the measured magnetic susceptibility of the standard dielectric material in each layer's containment space.

[0054] Specifically, a magnetic susceptibility meter is placed in the measurement channel and moved along the channel from top to bottom or bottom to top. The magnetic susceptibility of the standard dielectric material placed in each layer of the space is measured using the magnetic susceptibility meter, and this is the measured magnetic susceptibility.

[0055] It should be noted that, for accurate testing, the magnetic susceptibility of the standard dielectric material within each containment space is typically measured multiple times. The representative value of the magnetic susceptibility for that layer is then determined by calculating the median or a robust weighted average, and this representative value is used as the measured magnetic susceptibility for that layer. For example, assuming each layer is measured three times, and there are five containment spaces in total, the test result for the first layer is... The detection results of the second layer are The detection results of the third layer are The detection results of the fourth layer are The detection result of the fifth layer is The median was used to determine the representative values ​​for each layer: the first layer was... The second layer is The third layer is The fourth layer is The fifth floor is .

[0056] Step 2: Multiple measured magnetic susceptibility values ​​in different magnetic susceptibility ranges are used as discrete calibration samples and associated with their corresponding operating conditions to construct a calibration function for the magnetic susceptibility meter. The operating conditions include ambient temperature, the descent speed of the magnetic susceptibility meter along the measurement channel, and the eccentricity of the geometric axis of the magnetic susceptibility meter relative to the central axis of the measurement channel.

[0057] It should be noted that the measured magnetic susceptibility in the discrete calibration sample is the measured magnetic susceptibility of the standard dielectric material in each layer of the containment space detected in step 1. The standard magnetic susceptibility of the standard dielectric material placed in each containment space is in different magnetic susceptibility ranges. Of course, the standard magnetic susceptibility values ​​of each layer of containment space may be partially in the same magnetic susceptibility range and partially in different magnetic susceptibility ranges.

[0058] It should be noted that the operating conditions may also include only one or two of the following: ambient temperature, descent speed, and eccentricity.

[0059] As a specific embodiment of the present invention, the calibration function is constructed as follows: The form is... The mapping function, where K m f is the measured magnetic susceptibility. θLet be a piecewise rational spline function with adjustable parameters, θ be the spline parameter set, represent the set of parameter state vectors at the current time / current iteration, and c be the operating conditions.

[0060] Specifically, the measured magnetic susceptibility of each layer detected in step 1, along with the operating conditions, are used as inputs to construct a mapping function f. θ To output the target variable.

[0061] It should be noted that f θ The calibration function is a piecewise rational spline function with adjustable parameters. This allows the calibration function to meet the physical consistency requirements between the measured and calibrated magnetic susceptibility in magnetic susceptibility measurements, making it suitable for constructing a full-range calibration function under multi-level standard sample constraints. θ It can be a non-uniform rational spline function, which can be a non-uniform rational B-spline (NURBS) function. In this case, the calibration function f θ It can be represented as a rational spline function determined by a set of control points, weight parameters, and node vectors. Its parameter set θ is used to adjust the curve shape of the function in different magnetic susceptibility ranges. By adjusting the parameter set θ, the calibration function can remain piecewise continuous, overall smooth, and monotonically increasing across the entire magnetic susceptibility range, thus meeting the physical consistency requirements between the measured and calibrated magnetic susceptibility in magnetic susceptibility measurements. Furthermore, the NURBS function has good local adjustability; adjusting the control points in one range does not significantly affect other magnetic susceptibility ranges, making it suitable for constructing a calibration function covering the entire range under multi-level standard sample constraints.

[0062] Of course, the calibration function f θ It can also be other parameterized function forms that satisfy monotonicity constraints, such as piecewise polynomial functions, rational trigonometric spline functions, cubic spline functions, monotone rational functions, exponential functions, logarithmic functions, or regression models subject to monotonicity constraints. Without departing from the basic principles of this application, as long as f θ Any magnetic susceptibility mapping relationship that can be constructed under the constraints of multi-layer discrete calibration samples should be considered to fall within the protection scope of this invention.

[0063] Step 3: With minimizing the weighted error as the optimization objective, construct a partial Brussels bar optimization problem with respect to the calibration function parameters, and use preset constraints as constraints for the optimization problem to solve the constrained partial Brussels bar optimization problem to obtain the optimal calibration function.

[0064] The weighted error refers to the error term obtained by weighting the loss terms corresponding to each discrete calibration sample according to the sample weights determined based on uncertainty. The sample weights are determined by the total uncertainty obtained by combining the intra-layer repeated measurement uncertainty and the standard layer calibration uncertainty within each layer's containment space. By minimizing the weighted error as the optimization objective, calibration samples with lower uncertainty and higher reliability can occupy a higher weight in the optimization objective.

[0065] In this way, by optimizing the calibration function parameters and applying physical constraints, the resulting optimal calibration function can maintain good continuity and smoothness across the entire magnetic susceptibility range, avoiding abrupt changes, oscillations, or discontinuities at different magnetic susceptibility intervals or level boundaries, thereby improving the stability and repeatability of the calibration curve. Simultaneously, by applying monotonicity constraints during the optimization process, it can be ensured that the calibrated magnetic susceptibility changes monotonically with the measured magnetic susceptibility. Furthermore, uncertainty weighting and partial Bruker bar optimization ensure that the calibration function remains insensitive to changes in sample distribution even in the presence of random measurement noise, occasional abnormal samples, and disturbances in operating conditions, reducing the risk of overfitting and local bias. Therefore, the calibrated magnetic susceptibility obtained based on this optimal calibration function can maintain higher consistency with the standard magnetic susceptibility across different magnetic susceptibility intervals. Even with different models and batches of magnetic susceptibility meters and different measurement conditions, more stable and comparable calibration results can be obtained, better achieving unified calibration and standardized output of magnetic susceptibility meters.

[0066] As a specific embodiment of the present invention, the preset constraint conditions are as follows:

[0067] ;

[0068] Among them, K m f is the measured magnetic susceptibility. θ It is a piecewise rational spline function with adjustable parameters.

[0069] There is a natural monotonic physical relationship between magnetic susceptibility and magnetic response. By imposing constraints on the derivative of the calibration function with the aforementioned monotonicity constraints during the optimization process, it can be ensured that the calibration magnetic susceptibility changes monotonically with the measured magnetic susceptibility. This avoids the non-physical mapping where the measured magnetic susceptibility increases but the calibration result changes in the opposite direction, thus improving the physical rationality of the calibration results. Simultaneously, derivative continuity constraints or slope range constraints can be applied to the piecewise rational spline function at the boundaries of adjacent segments, enabling the calibration function to achieve a smooth transition and maintain continuity across the entire range. This avoids abrupt changes or oscillations at segment boundaries, ensuring that the calibration results are consistent and stable across different magnetic susceptibility ranges.

[0070] As a specific embodiment of the present invention, the Bruker bar optimization problem is shown in Equation 2 below:

[0071] ;

[0072] Among them, w i For sample weights;

[0073] This is a robust loss function used to measure the deviation between the calibration output and the standard magnetic susceptibility, preferably Huber loss or L1 loss. Taking L1 loss as an example, L1 loss is the mean absolute error loss, and its calculation method is as follows: ,in, It is the actual value. It is a predicted value;

[0074] Wasserstein distance, used to measure distribution. With experience distribution Differences;

[0075] δ is the robust radius, used to limit the range of distribution deviation. It can be determined based on the uncertainty of repeated measurements in each layer and the calibration uncertainty of the standard layer, or through methods such as cross-validation.

[0076] Q is satisfied The adversarial distribution within the distribution set is used to characterize the distribution uncertainty caused by noise, operating condition disturbances and outliers, thereby improving the robustness of the calibration function;

[0077] The weighted empirical distribution constructed based on multi-layer samples from standard wells is preferably represented as follows: ,in Can be determined by sample weights Normalization yields the result.

[0078] f θ It is a piecewise rational spline function with adjustable parameters;

[0079] θ is the set of spline parameters;

[0080] K mi Let be the measured magnetic susceptibility of the i-th layer;

[0081] Let i be the operating conditions of the i-th layer;

[0082] K si Let be the standard magnetic susceptibility of the i-th layer.

[0083] As a specific embodiment of the present invention, sample weight The overall uncertainty of the synthesis based on layer i is determined as shown in Equation 3 below:

[0084] ;

[0085] in, s i It is the standard deviation of repeated measures in the i-th layer, n i It is the number of repeated measurements in the i-th layer, u s,i The uncertainty introduced by pre-calibrating the i-th layer.

[0086] in, The magnetic susceptibility of this layer was calculated based on multiple measured magnetic susceptibility results, as shown in Equations 4 and 5 below:

[0087] ;

[0088] ;

[0089] in, For the magnetic susceptibility meter in the first Layer The measured magnetic susceptibility value obtained from repeated measurements.

[0090] Among them, u s,i It is an inherent property of the i-th layer standard dielectric material itself, specifically the standard magnetic susceptibility uncertainty determined by pre-calibration when the i-th layer standard dielectric material is used as a magnetic susceptibility standard.

[0091] Taking a space with 5 floors and 10 measurements per floor (n=10) as an example, the first floor has the following standard deviation: s i =2. Calibration uncertainty u s,i =1, Second layer: Standard deviation s i =5. Calibration uncertainty u s,i =1, Third layer: Standard deviation s i =2. Calibration uncertainty u s,i =3, .

[0092] As a specific embodiment of the present invention, numerical optimization methods can be used to solve the constrained sub-Brussels bar optimization problem, such as combining dual transformation with gradient-based methods, or using interior-point methods. Under the condition that the selected Wasserstein distance form and corresponding loss function conditions are satisfied, the dual theory of Wasserstein sub-Brussels bar optimization can be used to transform the worst-case optimization of the inner layer with respect to the distribution into an equivalent or upper-bound regularized empirical risk minimization problem; and further, combined with numerical optimization methods such as stochastic gradient descent, projected gradient method, augmented Lagrange method, or interior-point method, the optimal calibration function parameters Θ can be obtained under the constraints. Here, Θ represents the calibration function parameter space or parameter population.

[0093] Of course, an iterative approach combining inner-layer adversarial perturbation with outer-layer constraint optimization can also be used to solve the sub-Bruker optimization problem. Specifically, the th... Layer samples are represented as input vectors The output of the calibration function is the predicted calibration magnetic susceptibility. and with the corresponding standard magnetic susceptibility As a supervisory reference value, a constraint function is constructed on a preset set of magnetic susceptibility sampling points. This constraint function is used to impose physical consistency constraints on the parameters of the calibration function, where the monotonicity constraint can be achieved by imposing a non-negative constraint on the derivative of the calibration function with respect to the measured magnetic susceptibility.

[0094] In the inner-layer optimization, a perturbation that maximizes the loss function value is found for each sample within a preset perturbation radius ρ. This perturbation is obtained iteratively using a projected gradient ascent method to obtain an approximate worst-case perturbation. The perturbation radius ρ can correspond to or be derived from the robust radius δ in the sub-Bruker optimization. For example, the perturbation radius ρ and the robust radius δ in the sub-Bruker optimization satisfy the following correspondence: ,in, Let θ be the sample weight corresponding to the i-th discrete calibration sample, and the sample weight is determined by the combined total uncertainty. In the outer layer optimization, the constraints are incorporated into the calibration function in the form of a penalty function, and the spline parameters θ are updated using gradient descent or stochastic gradient descent methods.

[0095] By repeating the above iterative process of inner and outer layers until the calibration function converges or reaches the preset number of iterations, the optimal calibration function that satisfies the constraints is finally obtained. The adversarial training method of alternating inner-layer adversarial and outer-layer optimization to solve the bibru bar optimization problem can obtain a stable, smooth, and physically consistent calibration function even in the presence of measurement noise, anomalous samples, and operating condition disturbances.

[0096] Step 4: Correct the measured magnetic susceptibility of each layer using the optimal calibration function.

[0097] Specifically, the measured magnetic susceptibility of each layer will be... The corresponding operating conditions c=(T,v,r,…) are taken as input and substituted into the optimal calibration function obtained in step 3 to obtain the calibrated magnetic susceptibility. Where T is the ambient temperature, v is the descent speed of the magnetic susceptibility meter along the measurement channel, and r is the eccentricity of the geometric axis of the magnetic susceptibility meter relative to the central axis of the measurement channel. The final calibrated magnetic susceptibility is then obtained. The distribution of magnetic susceptibility with well depth z can serve as a standard input for subsequent geological interpretation, lithological comparison, and quantitative analysis. In actual well logging, the magnetic susceptibility meter continuously collects measured magnetic susceptibility data along the well depth direction, obtaining a measured magnetic susceptibility curve that varies with well depth. Simultaneously record the operating conditions at the corresponding well depth. .Will Substitute the corresponding operating conditions point by point into the optimal calibration function obtained in step 3. The calibrated magnetic susceptibility curve can then be obtained. Because the calibration function remains continuous, smooth, and physically consistent across the entire magnetic susceptibility range, and exhibits strong stability against operating condition disturbances, the obtained... It can more realistically reflect the variation characteristics of formation magnetism with well depth and can be used as a standard input for subsequent geological interpretation, lithological comparison and quantitative analysis.

[0098] When using the optimal calibration function for correction, a confidence interval for each calibration value is provided simultaneously. This confidence interval is determined by either a bootstrap method or variational inference of the calibration function f. θ The prediction results are quantified. The confidence interval is output along with the calibration value, providing users with a quantitative basis for judging the reliability of the data and significantly improving the interpretability of the calibration value.

[0099] The bootstrap method can be used to assess the prediction uncertainty of the calibration magnetic susceptibility and construct the corresponding confidence interval. Specifically, during the calibration function training phase, multiple bootstrap sample sets are constructed from the original calibration sample set through resampling with replacement. Each bootstrap sample set is used to independently train and obtain a set of calibration function parameters. Thus, a set of calibration functions is obtained. .

[0100] Given a new input Substituting these values ​​into the aforementioned calibration functions yields a corresponding set of predicted calibration magnetic susceptibility values. The distribution of the predicted value set reflects the prediction uncertainty caused by the finiteness of the sample and the uncertainty of the model parameters. Based on the empirical quantiles of this distribution, we take... and quantiles can be used to construct the corresponding confidence level. The calibration magnetic susceptibility confidence interval.

[0101] In actual well logging operations, online fine-tuning can also be performed: adjusting the calibration function parameters... Treating the standard magnetic susceptibility at the anchor point as a state variable, and using Kalman filtering as an observation, the parameters are slightly corrected. Specifically, while keeping the calibration function structure unchanged, the predicted output of the calibration function is calculated based on the measured magnetic susceptibility at the anchor point and the operating conditions. This predicted output is then compared with the corresponding standard magnetic susceptibility to construct the observation residual. This observation residual is then used to update the parameters through Kalman filtering. This allows for local correction of the calibration function. By configuring a smaller process noise covariance and a relatively larger observation noise covariance, the gain matrix of the Kalman filter is kept at a low level, enabling active limitation of the parameter update magnitude. This allows the calibration function to maintain global stability while improving matching accuracy under the current operating conditions or hierarchical conditions.

[0102] In one specific embodiment of the present invention, before correcting the measured magnetic susceptibility using the optimal calibration function, the difference between the measured magnetic susceptibility of each layer and the corresponding standard magnetic susceptibility is calculated, and it is determined whether the absolute value of each difference is greater than a threshold, for example, the threshold is... When the absolute value of the difference exceeds a threshold, the measured magnetic susceptibility corresponding to that difference is corrected using the optimal calibration function. In other words, correction is not applied to all measured magnetic susceptibility values, but only to those detected by the susceptibility meter when a significant deviation occurs. This reduces the amount of correction and increases calibration speed.

[0103] Taking a space with a total of 5 floors as an example, the test results for the first floor are as follows: Standard magnetic susceptibility is The error is The detection results of the second layer are Standard magnetic susceptibility is The error is The detection results of the third layer are Standard magnetic susceptibility is The error is The detection results of the fourth layer are Standard magnetic susceptibility is The error is The detection results of the fifth layer are Standard magnetic susceptibility is The error is It can be seen that the magnetic susceptibility meter has a smaller detection error in the weak magnetic regions of the first and second layers, but the error increases significantly in the strong magnetic regions of the third to fifth layers. This indicates that the response value of the magnetic susceptibility meter becomes higher in the high magnetic region, resulting in a higher detection error, which needs to be corrected.

[0104] The following is combined with Figure 1 To further illustrate possible implementations of the standard well of the present invention.

[0105] like Figure 1As shown, the magnetic susceptibility standard well of the present invention includes a housing 1 and a support base 3. The housing 1 is mounted on the support base 3, which is generally a plate-like structure. The support base 3 allows the magnetic susceptibility standard well to be stably placed in a laboratory or field location, enabling it to be moved and used in both environments. This meets the needs of research institutes, instrument manufacturers, and others for rapid deployment and repeated testing in different locations. The housing 1 is generally a vertically oriented cylindrical structure, located approximately in the center of the support base 3. The measurement channel 2 is coaxial with the housing 1 and extends vertically. A guide structure 11 is provided at the top of the housing 1. A through hole 111 communicating with the measurement channel 2 is provided in the center of the guide structure 11. The axis of the through hole 111 is coaxial with the axis of the measurement channel 2, and the magnetic susceptibility meter enters and exits the measurement channel 2 through the through hole 111.

[0106] Multiple isolation layers 4 are arranged inside the outer casing 1 along the axial direction of the measurement channel 2. These isolation layers 4 divide the space between the outer casing 1 and the measurement channel 2 into multiple independent annular receiving spaces 5, each of which constitutes a standard layer. The isolation layers 4 not only define the spatial range of the multiple standard layers but also reduce magnetic or electromagnetic interference between adjacent standard layers. The multiple receiving spaces 5 are distributed sequentially along the extension direction of the measurement channel 2, and standard dielectric materials can be placed within these spaces to form standard layers. The standard dielectric material has a standard magnetic susceptibility, which is calibrated under laboratory conditions using a precise magnetic susceptibility instrument (e.g., Kappabridge). This standard dielectric material is a solid standard dielectric, made from artificially synthesized materials or compacted natural rock powder. The artificially synthesized material can be a block formed from a mixture of ferrite powder and a non-magnetic binder. Those skilled in the art can flexibly choose the specific preparation method of the standard dielectric material according to the specific application scenario, as long as the magnetic susceptibility requirements for calibration are met and the magnetic susceptibility value remains stable over a long period.

[0107] In one possible implementation, the standard medium material is detachably placed in the receiving space 5, so that the user can replace the standard medium material in each receiving space 5 with a medium material of different magnetic susceptibility range as needed, so as to meet the measurement accuracy correction requirements of different models of magnetic susceptibility meters across the entire range, with comprehensive range coverage and wide application range.

[0108] Continue to refer to Figure 1 Along the axial direction of measuring channel 2 (approximately) Figure 1 Observing vertically, the magnetic susceptibility of each standard layer increases sequentially from top to bottom. Taking a shell 1 with five accommodating spaces 5 as an example, the first, second, third, fourth, and fifth standard layers are arranged sequentially from top to bottom. Among them, the magnetic susceptibility of the first standard layer is... Between, the magnetic susceptibility of the second standard layer is The magnetic susceptibility of the third standard layer between Between, the magnetic susceptibility of the fourth standard layer is Between, the magnetic susceptibility of the fifth standard layer is In this way, the standard well simultaneously covers the weak magnetic, medium magnetic, and strong magnetic ranges, which can intuitively quantify the measurement error of the magnetic susceptibility meter in different magnetic susceptibility ranges. Moreover, the measurement accuracy of various types of magnetic susceptibility meters across the entire range can be verified and corrected, thereby enabling a more accurate and comprehensive evaluation of the performance of the magnetic susceptibility meter.

[0109] It should be noted that the magnetic susceptibility of each standard layer can also be within other ranges. Of course, the outer casing 1 can also have 2, 3, or 4 or fewer other accommodating spaces 5, or 6, 7, 8 or more other accommodating spaces 5.

[0110] It should be noted that the magnetic susceptibility of the standard dielectric material filling each accommodating space 5 can be distributed along the measurement channel direction in a preset manner, such as decreasing sequentially from top to bottom. In other embodiments, a non-monotonic arrangement can also be used as needed, so that the standard dielectric materials of different magnetic susceptibility ranges are arranged crosswise in the axial direction to cover the target magnetic susceptibility range.

[0111] In one possible implementation, both the outer shell 1 and the insulating layer 4 are made of non-magnetic materials, which can be one or more of PVC, glass fiber reinforced plastic, and non-magnetic alloys. This non-magnetic outer shell 1 isolates the interior from the exterior, preventing external magnetic fields from interfering with the magnetic susceptibility meter's measurements. The non-magnetic insulating layer 4 not only separates the various accommodating spaces 5, preventing the standard media materials filled in each space from mixing, but also prevents interference between standard media materials with different magnetic susceptibilities, thus avoiding impact on the measurement results.

[0112] In summary, the standard well structure of this application is simple and has low manufacturing cost. The use of a completely non-magnetic outer shell 1 and an isolation layer 4 ensures long-term use unaffected by environmental interference. Calibration only requires moving the device downwards, avoiding cumbersome laboratory operations and complex mathematical correction processes.

[0113] In one possible implementation, the outer casing 1 has an opening (not shown) corresponding to each receiving space 5, and a cover (not shown) is hinged to the opening, through which the opening can be opened or closed. When the cover is open, the opening is open, allowing the standard medium material to be added or replaced to adapt to the calibration requirements of different magnetic susceptibility ranges and different models of magnetic susceptibility meters. When the cover is closed, the opening is closed, forming a sealed receiving space 5 to prevent the magnetic field in the external environment from affecting the measurement results of the magnetic susceptibility meter. This avoids uncertainties introduced by material replacement or external interference, ensuring that the geometric dimensions and filling state of the standard medium material in each receiving space remain stable during the measurement process. The calibration benchmark remains its pre-calibrated standard magnetic susceptibility value, ensuring the calibration accuracy of the magnetic susceptibility meter. At the same time, the standard medium material with different magnetic susceptibility can be replaced as needed through the opening, which can match various downhole environments and is widely used for the calibration of various models of magnetic susceptibility meters with different magnetic susceptibility ranges.

[0114] In one possible implementation, the outer wall of the housing 1 is provided with a mark (not shown) at the position corresponding to each standard layer. The mark may be different in texture or grayscale of the housing 1. The mark can help the user to clearly distinguish each standard layer and facilitate the filling or replacement of standard media materials.

[0115] In one possible implementation, the standard well is further equipped with a constant temperature control module 6, which includes a temperature control unit and a temperature regulation unit and a temperature detection unit connected thereto. The temperature regulation unit is used to heat and / or cool the interior of the standard well, and the temperature detection unit is used to collect temperature information of the well's interior or exterior. The temperature control unit performs closed-loop control of the temperature regulation unit based on the temperature information collected by the temperature detection unit, ensuring that the internal temperature of the standard well remains stable within a preset range. This constant temperature control module 6 enables the standard well to maintain a constant temperature, preventing fluctuations in ambient temperature from affecting the measurement results of the magnetic susceptibility meter, thereby better ensuring the accuracy of calibration.

[0116] It should be noted that the temperature regulation unit includes a resistance heating element, a heating band, a semiconductor cooling element, or a combination thereof. Taking a resistance heating element as the temperature regulation unit, a resistance heating element is provided at the location corresponding to each containment space, and each containment space is heated and / or cooled by the resistance heating element. The temperature detection unit includes a thermistor, a platinum resistance thermometer, or a digital temperature sensor. Taking a thermistor as the temperature detection unit, a thermistor is provided at each containment space, and each thermistor detects the temperature in each containment space and transmits it to the temperature control unit, so that the control unit can control the resistance heating element to heat or cool the corresponding area in a timely manner to ensure a constant temperature inside the standard well. Of course, a resistance heating element and a thermistor can also be provided at the location of two adjacent containment spaces, or the corresponding number of resistance heating elements and thermistors can be provided in other possible ways. Without departing from the basic principles of this application, those skilled in the art can flexibly choose the specific type and specific setting method of the temperature regulation unit and the temperature detection unit according to the specific application scenario, as long as the overall constant temperature of the standard well can be ensured.

[0117] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for calibrating a magnetic susceptibility meter based on a standard well, characterized in that, The standard well includes a housing with a measurement channel inside allowing a magnetic susceptibility meter to pass through. Multiple layers of receiving spaces are provided within the housing along the extension direction of the measurement channel, where a standard dielectric material with a standard magnetic susceptibility can be placed. The method includes: The magnetic susceptibility of the standard dielectric material in each layer's containment space was measured using the aforementioned magnetic susceptibility meter. The measured magnetic susceptibility in multiple ranges is used as discrete calibration samples and associated with their corresponding operating conditions to construct a calibration function for the magnetic susceptibility meter. The operating conditions include at least one of the following: ambient temperature, the descent speed of the magnetic susceptibility meter along the measurement channel, and the eccentricity of the geometric axis of the magnetic susceptibility meter relative to the central axis of the measurement channel. With minimizing the weighted error as the optimization objective, a partial Brussels bar optimization problem with respect to the calibration function parameters is constructed, and the preset constraints are used as constraints for the optimization problem. The constrained partial Brussels bar optimization problem is solved to obtain the optimal calibration function. The measured magnetic susceptibility of each layer is corrected using the optimal calibration function.

2. The method according to claim 1, characterized in that, The calibration function is constructed as follows: The establishment form is The mapping function, where K m f is the measured magnetic susceptibility. θ Let θ be a piecewise rational spline function with adjustable parameters, θ be the spline parameter set, and c be the operating conditions.

3. The method according to claim 1, characterized in that, The preset constraints are shown in Equation 1 below: Among them, K m f is the measured magnetic susceptibility. θ It is a piecewise rational spline function with adjustable parameters.

4. The method according to claim 1, characterized in that, The optimization problem of the sub-Bruker bar is shown in Equation 2 below: Among them, w i For sample weights, For a robust loss function, Let δ be the Wasserstein distance, δ be the robust radius, and Q be the value that satisfies... The adversarial distribution within the distribution set, f is a weighted empirical distribution constructed based on multi-layer samples from the standard wells. θ Let K be a piecewise rational spline function with adjustable parameters, where θ is the set of spline parameters, and K is the set of parameters. mi Let be the measured magnetic susceptibility of the i-th layer. Let K be the operating condition of the i-th layer. si Let be the standard magnetic susceptibility of the i-th layer.

5. The method according to claim 4, characterized in that, The sample weight The total uncertainty of the synthesis based on the i-th layer is determined as shown in Equation 3 below: in, s i It is the standard deviation of repeated measures in the i-th layer, n i It is the number of repeated measurements in the i-th layer, u s,i The uncertainty is pre-calibrated for the i-th layer.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Before correcting the measured magnetic susceptibility using the optimal calibration function, the difference between the measured magnetic susceptibility of each layer and the corresponding standard magnetic susceptibility is calculated. Determine whether the absolute value of each difference is greater than the threshold; When the absolute value of the difference is greater than the threshold, the measured magnetization corresponding to the difference is corrected using the optimal calibration function.

7. The method according to any one of claims 1 to 5, characterized in that, The method further includes: When applying the calibration function for correction, a confidence interval for each calibration value is provided simultaneously, wherein the confidence interval is obtained by quantifying the prediction result of the calibration function using a bootstrap method or variational inference method.

8. The method according to any one of claims 1 to 5, characterized in that, Each of the multiple accommodating spaces contains a standard dielectric material covering a different magnetic susceptibility range, and the magnetic susceptibility of the standard dielectric material in each accommodating space increases or decreases in a stepwise manner along the axial direction of the measurement channel; and / or The standard medium material is detachably disposed within the accommodating space.

9. The method according to any one of claims 1 to 5, characterized in that, The measuring channel is coaxially arranged with the housing; and / or The standard well is also equipped with a constant temperature control module, which is configured to maintain a constant temperature in the standard well.

10. The method according to any one of claims 1 to 5, characterized in that, The outer shell is made of a non-magnetic material; and / or An isolation layer is provided between two adjacent accommodating spaces, and the isolation layer is made of a non-magnetic material.

Images