Logging curve environment correction method and device

Abstract

The invention discloses a logging curve environment correction method and device, and relates to the technical field of petroleum geological exploration reservoir stratum logging evaluation, and the method comprises the steps: obtaining first logging element data, the first logging element data comprising single-element logging data corresponding to elements contained in a rock debris sample; performing signal decomposition on the well logging curve of the non-expanded well section to obtain full-band well logging signals (including a plurality of first low-frequency well logging signals and high-frequency well logging signals); synthesizing the first low-frequency logging signals to obtain a plurality of second low-frequency logging signals; screening out prediction single-element logging data, and predicting a low-frequency logging signal so as to construct a prediction model; and using the prediction model to predict a third low-frequency logging signal of the expanded interval, and synthesizing the third low-frequency logging signal with a high-frequency logging signal in the full-band logging signal to obtain a corrected logging curve. According to the method, a more accurate logging curve can be obtained at the expanded interval.

Description

Technical Field

[0001] The invention relates to the technical field of well logging evaluation in petroleum geological exploration and reservoir, and in particular to a method and device for environmental correction of a well logging curve. Background Art

[0002] In the field of petroleum geology, logging environment refers to non-stratigraphic factors such as borehole diameter (wellbore diameter), mud density, mud properties, mud invasion, and the gap between the instrument and the formation, which will affect the quality of the logging curve to varying degrees, making it difficult to use directly measured data for accurate formation parameter evaluation. Especially in the drilling process of complex formations such as buried hills, shales and carbonate rocks, stress release, mudstone water absorption and expansion often cause the wellbore wall to collapse to varying degrees, resulting in wellbore expansion and serious distortion of logging information. At present, the methods for environmental correction of logging curves mainly rely on the plate method and computer automatic correction method. Among them, the plate method is that major logging companies draw correction plates under different conditions based on the response values ​​of their own instruments in standard scale wells and the rock physics forward modeling method, and then perform logging curve environmental correction based on the method of manual plate checking. The computer automatic correction method is a method of digitizing the plate, establishing a plate database, and then programming to realize automatic database checking for correction, or using artificial intelligence or digital simulation algorithms for environmental correction.

[0003] For the correction of borehole expansion, the plate correction method determines the correction amount according to the size of the borehole expansion, that is, the larger the expansion, the larger the correction amount. In the logging process, due to the different measurement methods of different instruments in the well, such as the natural gamma instrument adopts the center measurement, and the density logging instrument adopts the measurement close to the wellbore wall, in the well section with large borehole expansion and wide vertical distribution range, the density logging instrument can often fit well with the wellbore wall, the measurement information is accurate, and the required correction amount is small; while when the borehole expansion is not large and the vertical range is small, the density instrument is difficult to fit with the wellbore wall, resulting in serious distortion of the density logging value, and the required correction amount is large. If the plate method is used for correction, the correction result is difficult to match the actual situation. In addition, for the layer section with special expansion, the signal measured by the logging instrument has been completely distorted, and the real logging information cannot be corrected by the plate method. In general, the current method of logging curve environment correction in the expansion layer section is achieved by adjusting the correction amount, which is easy to cause distortion of logging information.

[0004] Therefore, how to obtain more accurate logging curves in the expanded diameter section is a problem that needs to be solved urgently. Summary of the invention

[0005] In order to solve the existing problem that a relatively accurate well logging curve cannot be obtained, the present invention provides a well logging curve environment correction method and device.

[0006] The technical solution adopted by the present invention to solve the technical problem is: to provide a method for environmental correction of a well logging curve, comprising:

[0007] Acquiring first logging element data, wherein the first logging element data includes single element logging data corresponding one-to-one to the elements contained in the cuttings sample;

[0008] Performing signal decomposition on the logging curve of the unexpanded well section to obtain a full-band logging signal, wherein the full-band logging signal includes a plurality of first low-frequency logging signals and a high-frequency logging signal;

[0009] synthesizing the first low-frequency logging signal to obtain a plurality of second low-frequency logging signals;

[0010] Based on the correlation between the first logging element data and the second low-frequency logging signal, predicting single-element logging data is screened out from the single-element logging data, and predicting low-frequency logging signals are screened out from the second low-frequency logging signals;

[0011] Construct a prediction model based on predicting single-element logging data and predicting low-frequency logging signals;

[0012] The prediction model is used to predict the third low-frequency logging signal of the expanded diameter layer section, and the third low-frequency logging signal is synthesized with the high-frequency logging signal to obtain a corrected logging curve.

[0013] In one embodiment, the step of synthesizing the first low-frequency logging signal includes:

[0014] The stacking synthesis is performed in sequence from low to high according to the frequency of the first low-frequency logging signal.

[0015] In one embodiment, based on the correlation between the first logging element data and the second low-frequency logging signal, the steps of screening out predicted single-element logging data from the single-element logging data and screening out predicted low-frequency logging signals from the second low-frequency logging signals include:

[0016] Based on the resolution correlation between the second low-frequency logging signal and the first logging element data, a predicted low-frequency logging signal matching the vertical resolution of the first logging element data is screened out from the second low-frequency logging signal;

[0017] Based on the data correlation between the predicted low-frequency logging signal and the first logging element data, the predicted single-element logging data is screened out from the single-element logging data.

[0018] In one embodiment, based on the resolution correlation between the second low-frequency logging signal and the first logging element data, the step of screening out the predicted low-frequency logging signal that matches the vertical resolution of the first logging element data from the second low-frequency logging signal includes:

[0019] Calculate the resolution correlation coefficient between the second low-frequency logging signal and the single-element logging data;

[0020] The second low-frequency logging signal corresponding to the largest resolution correlation coefficient is selected as the predicted low-frequency logging signal.

[0021] In one embodiment, based on the data correlation between the predicted low-frequency logging signal and the first logging element data, the step of selecting the predicted single-element logging data from all the single-element logging data includes:

[0022] Calculate the correlation coefficient between the predicted low-frequency logging signal and the single-element logging data;

[0023] comparing the correlation coefficient with a preset correlation coefficient threshold;

[0024] The single-element logging data with a correlation coefficient greater than the correlation coefficient threshold are selected as the predicted single-element logging data.

[0025] In one embodiment, the step of obtaining the first logging element data includes:

[0026] In the logging area, cuttings samples are taken at preset intervals;

[0027] Use the measurement tool to obtain single element logging data from rock cuttings samples.

[0028] In one embodiment, the step of performing signal decomposition on the logging curve of the unexpanded well section includes:

[0029] Perform EMD signal decomposition or wavelet decomposition on the logging curve of the unexpanded well section.

[0030] In one embodiment, the step of constructing a prediction model based on predicted single-element logging data and predicted low-frequency logging signals includes:

[0031] The predicted single-element logging data is used as input features;

[0032] The predicted low-frequency logging signal is used as the target output;

[0033] Based on the input features and target output, a predictive model is built using a machine learning algorithm.

[0034] In one embodiment, the step of building a prediction model using a machine learning algorithm includes:

[0035] Use KNN neural network algorithm or BP neural network algorithm to build a prediction model.

[0036] In addition, to achieve the above purpose, the present application also proposes a well logging curve environment correction device, the well logging curve environment correction device comprises:

[0037] A data acquisition module is used to acquire first logging element data, wherein the first logging element data includes single-element logging data corresponding one-to-one to the elements contained in the cuttings sample; perform signal decomposition on the logging curve of the unexpanded well section to acquire a full-band logging signal, wherein the full-band logging signal includes a plurality of first low-frequency logging signals and a high-frequency logging signal; synthesize the first low-frequency logging signal to obtain a plurality of second low-frequency logging signals; based on the correlation between the first logging element data and the second low-frequency logging signal, screen out predicted single-element logging data from the single-element logging data, and screen out predicted low-frequency logging signals from the second low-frequency logging signal;

[0038] A model building module, used to build a prediction model based on predicted single-element logging data and predicted low-frequency logging signals;

[0039] The curve correction module is used to predict the third low-frequency logging signal of the expanded diameter layer section using the prediction model, synthesize the third low-frequency logging signal with the high-frequency logging signal, and obtain a corrected logging curve.

[0040] The present invention provides a method for correcting the environment of a logging curve. The present invention first obtains first logging element data (the first logging element data includes single-element logging data corresponding one-to-one to the elements contained in the cuttings sample); then, the logging curve of the unexpanded well section is subjected to signal decomposition to obtain a full-band logging signal including a plurality of first low-frequency logging signals and a high-frequency logging signal; then, the first low-frequency logging signal is synthesized to obtain a plurality of second low-frequency logging signals; then, based on the correlation between the first logging element data and the second low-frequency logging signal, predicted single-element logging data is screened out from the single-element logging data, and predicted low-frequency logging signals are screened out from the second low-frequency logging signal, wherein the screened predicted single-element logging data and the predicted low-frequency logging signal have a strong correlation; a prediction model is constructed based on the predicted single-element logging data and the predicted low-frequency logging signal; the prediction model is used to predict a third low-frequency logging signal of the expanded layer section, and the third low-frequency logging signal is synthesized with the high-frequency logging signal in the full-band logging signal, and finally, a relatively accurate logging curve after correction is obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0042] Figure 1 It is a schematic diagram of a flow chart provided by the first embodiment of the present invention;

[0043] Figure 2This is a decomposition diagram of the EMD signal of the natural gamma curve of Well B provided by the first embodiment of the present invention;

[0044] Figure 3 is a matching diagram of the natural gamma synthetic low-frequency signal and the vertical resolution of the logging element provided by the second embodiment of the present invention;

[0045] Figure 4 is a correlation diagram between the natural gamma synthetic low-frequency signal and the logging element provided by the third embodiment of the present invention;

[0046] Figure 5 This is a graph of natural gamma synthetic low-frequency signal prediction results for an unexpanded layer segment provided by the fourth embodiment of the present invention;

[0047] Figure 6 It is a result diagram of the correction of the natural gamma curve of the expanded diameter layer provided by the present invention;

[0048] Figure 7 It is a flow chart of the technical steps of the present invention;

[0049] Figure 8 It is a schematic diagram of the module structure of the logging curve environment correction device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] Taking the environmental correction process of the natural gamma logging curve of the buried hill section of well B in oil field A in the Pearl River Mouth Basin as an example, the environmental correction method of the logging curve proposed in the first embodiment of the present invention is described in detail. Please refer to Figure 1 The method comprises steps S10 to S60:

[0052] Step S10, obtaining first logging element data, wherein the first logging element data includes single element logging data corresponding one-to-one to the elements contained in the cuttings sample;

[0053] It should be noted that the first logging element data refers to the content of multiple logging elements in the cuttings sample, and the single element logging data refers to the content of one logging element in the cuttings sample. Acquisition refers to the real-time collection of cuttings samples and the determination of the content of various logging elements in the cuttings by analytical instruments through sampling and measurement methods during the drilling operation. Acquisition not only refers to the physical collection of cuttings samples, but also includes the entire data collection process of analyzing the samples through instruments. The first logging element data refers to the data results of measuring the logging element content in the cuttings using logging technology during drilling operations.

[0054] In a feasible implementation manner, step S10 of acquiring the first logging element data includes steps S11 to S12:

[0055] Step S11, selecting rock cutting samples at preset intervals in the logging area; Step S12, using a measuring tool to obtain single element logging data corresponding to the elements contained in the rock cutting samples.

[0056] It should be noted that the preset interval can be a distance period for selecting rock cuttings samples, which can be set according to the actual distance period required during the drilling mud circulation process. For example, in the logging area, a representative rock cuttings sample is selected at an interval of 1 meter, and the measuring tool can be a rock cuttings element logging analysis instrument, such as the HB-X100 X-ray logging instrument.

[0057] Specifically, in the logging area, rock cutting samples are selected at preset intervals, such as 1-meter intervals, and a rock cutting element logging analysis instrument is used to obtain the corresponding content of each logging element in the rock cutting samples.

[0058] For example, the depth distribution range of the buried hill strata encountered in Well B is 3600 meters to 4200 meters. During the buried hill drilling process, a representative rock cutting sample was selected every one meter, and the HB-X100 X-ray logging instrument was used to measure the contents of 17 elements in the rock cuttings, including 12 major elements silicon (Si), calcium (Ca), iron (Fe), aluminum (Al), sodium (Na), magnesium (Mg), potassium (K), phosphorus (P), sulfur (S), chlorine (Cl), manganese (Mn), titanium (Ti), and 5 trace elements barium (Ba), vanadium (V), nickel (Ni), strontium (Sr) and zirconium (Zr).

[0059] Step S20, performing signal decomposition on the logging curve of the unexpanded well section to obtain a full-band logging signal, wherein the full-band logging signal includes a plurality of first low-frequency logging signals and a high-frequency logging signal;

[0060] It should be noted that the logging curves include resistivity, natural gamma, acoustic time difference, density and other curves. Signal decomposition refers to the processing of the logging curves by mathematical methods to divide the original signals into different frequency bands in order to extract different geological information. In the logging curves of the unexpanded well section, the signal decomposition process is to decompose the original logging signal into logging signals of the full frequency band from high to low.

[0061] In addition, it should be noted that the first low-frequency logging signal is the lower-frequency component in the full-band logging signal, which can reflect the approximate periodic cycle of the geological structure, and the high-frequency logging signal can reflect the fluctuation frequency of the geological structure. The division of the first low-frequency logging signal and the high-frequency logging signal can be carried out by sequentially superimposing and synthesizing the full-band logging signal, because this superposition synthesis method can gradually add more frequency components from the low-frequency signal component, thereby effectively determining the boundary point between low frequency and high frequency. By gradually superimposing and synthesizing the full-band logging signal from low to high, when the superposition reaches a certain degree, the synthesized low-frequency logging signal can still reflect the periodic cycle of the geological structure. When a higher logging signal is superimposed, the fluctuation frequency of the signal becomes larger. At this time, the superimposed signal and the higher-frequency logging signal are the high-frequency logging signal, and the other is the first low-frequency logging signal.

[0062] Optionally, the logging curve of the unexpanded well section may be subjected to EMD signal decomposition or wavelet decomposition to obtain a full-band logging signal.

[0063] It should be noted that EMD (Empirical Mode Decomposition) signal decomposition and wavelet decomposition are both signal processing technologies, which are used to decompose complex signals into different frequency components to extract local features in the signal.

[0064] Exemplarily, the natural gamma logging curve of the non-expanded layer section is subjected to EMD signal decomposition or wavelet decomposition to obtain 7 signals in different frequency bands from high frequency to low frequency, namely IMF1, IMF2, IMF3, IMF4, IMF5, IMF6, and IMF7. The high-frequency logging signals are IMF1 and IMF2, and the first low-frequency logging signals are IMF3, IMF4, IMF5, IMF6, and IMF7.

[0065] Reference Figure 2 , Figure 2This is the EMD signal decomposition diagram of the natural gamma curve of Well B. The first channel in the figure includes the natural gamma curve (GR), the wellbore curve (CAL) and the drill bit diameter curve (BIT); the second channel is the IMF1 signal channel, and the third channel is the IMF2 signal channel, which represent the high-frequency components of the signal; the fourth channel is the IMF3 signal channel, and the fifth channel is the IMF4 signal channel. These two signal channels represent the medium and low-frequency components in the natural gamma curve signal, and usually reflect the larger-scale changes in the signal; the sixth channel is the IMF5 signal channel, the seventh channel is the IMF6 signal channel, and the eighth channel is the IMF7 signal channel. These three signal channels are usually the low-frequency components of the signal, reflecting the overall trend of the signal or the slowly changing part. Figure 2 It can be seen that the curve of the high-frequency signal channel changes more obviously, while the curve of the medium and low-frequency signal channel changes more gently.

[0066] Step S30, synthesizing the first low-frequency logging signal to obtain a plurality of second low-frequency logging signals;

[0067] It should be noted that synthesis refers to reconstructing a new low-frequency logging signal by weighting and superimposing the medium and low frequencies (first low-frequency logging signal) in the decomposed full-band logging signal, especially the low-frequency component. The second low-frequency logging signal represents a more stable formation feature, removes the noise or subtle fluctuations of the high-frequency part, and makes the low-frequency signal clearer and more representative, which is convenient for subsequent analysis.

[0068] Specifically, the first low-frequency logging signal in the full-band logging signal is synthesized by grouping to obtain the second low-frequency logging signal.

[0069] In a feasible implementation manner, the step of synthesizing the first low-frequency logging signal in step S30 includes step S31:

[0070] Step S31, performing stacking synthesis in sequence from low to high according to the frequency of the first low-frequency logging signal.

[0071] Exemplarily, low-frequency logging signals (IMF7, IMF6), (IMF7, IMF6, IMF5), (IMF7, IMF6, IMF5, IMF4), (IMF7, IMF6, IMF5, IMF4, IMF3) are selected for superposition and synthesis in sequence, and the synthesized natural gamma second low-frequency logging signals are recorded as GR_M6, GR_M5, GR_M4, and GR_M3 respectively.

[0072] Step S40, based on the correlation between the first logging element data and the second low-frequency logging signal, screen out predicted single-element logging data from the single-element logging data, and screen out predicted low-frequency logging signals from the second low-frequency logging signals;

[0073] In a feasible implementation manner, step S40 includes steps S41 to S42:

[0074] Step S41, based on the resolution correlation between the second low-frequency logging signal and the first logging element data, screen out the predicted low-frequency logging signal that matches the vertical resolution of the first logging element data from the second low-frequency logging signal; Step S42, based on the data correlation between the predicted low-frequency logging signal and the first logging element data, screen out the predicted single-element logging data from the single-element logging data.

[0075] Specifically, the resolution correlation between the second low-frequency logging signal and the first logging element data is calculated to screen out the predicted low-frequency logging signal that is relatively consistent with the logging element in vertical resolution. The data correlation between the predicted low-frequency logging signal and the first logging element data is calculated to obtain the predicted single-element logging data that is relatively consistent with the predicted low-frequency logging signal in data change.

[0076] Step S50, constructing a prediction model based on the predicted single-element logging data and the predicted low-frequency logging signal;

[0077] Specifically, by analyzing the relationship between the predicted single-element logging data and the predicted low-frequency logging signal, a suitable prediction model (such as a regression model, a neural network, etc.) is constructed. This model can accurately predict the value of the low-frequency logging signal under new measurement data.

[0078] Step S60: Use the prediction model to predict the third low-frequency logging signal of the expanded diameter layer section, synthesize the third low-frequency logging signal with the high-frequency logging signal, and obtain a corrected logging curve.

[0079] It should be noted that the full-band logging signal also includes high-frequency logging signals. Borehole expansion has a greater impact on low-frequency signals in the logging curve, so the predicted third low-frequency logging signal can be synthesized with the high-frequency logging signal to obtain a corrected logging curve.

[0080] In addition, the above steps S20 to S50 are all operations performed on the well section that has not been expanded.

[0081] Specifically, the constructed prediction model was applied to the well section (3845-3865 meters) with serious diameter expansion in Well B, and the low-frequency natural gamma logging signal (that is, the third low-frequency logging signal) of the diameter expansion section was obtained. The third low-frequency logging signal was predicted by using the prediction model. Then, the third low-frequency logging signal was synthesized with the high-frequency logging signal in the full-band logging signal to obtain a corrected logging curve. The synthesized logging curve not only retains the true stratigraphic characteristics of the low-frequency signal, but also fully retains the details of the high-frequency signal. The corrected curve can more accurately reflect the actual geological information of the diameter expansion section and avoid the interference of the diameter expansion effect on the measurement results.

[0082] The present invention provides a method for correcting the environment of a logging curve. The present invention first obtains first logging element data (the first logging element data includes single-element logging data corresponding one-to-one to the elements contained in the cuttings sample); then, the logging curve of the unexpanded well section is subjected to signal decomposition to obtain a full-band logging signal including a plurality of first low-frequency logging signals and a high-frequency logging signal; then, the first low-frequency logging signal is synthesized to obtain a plurality of second low-frequency logging signals; then, based on the correlation between the first logging element data and the second low-frequency logging signal, predicted single-element logging data is screened out from the single-element logging data, and predicted low-frequency logging signals are screened out from the second low-frequency logging signal, wherein the screened predicted single-element logging data and the predicted low-frequency logging signal have a strong correlation; a prediction model is constructed based on the predicted single-element logging data and the predicted low-frequency logging signal; the prediction model is used to predict a third low-frequency logging signal of the expanded well section, and the third low-frequency logging signal is synthesized with the high-frequency logging signal, and finally, a more accurate and corrected logging curve is obtained in the expanded well section.

[0083] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the first embodiment can be referred to the above description, and will not be described in detail later. On this basis, step S41, based on the resolution correlation between the second low-frequency logging signal and the first logging element data, the step of screening out the predicted low-frequency logging signal that matches the vertical resolution of the first logging element data in the second low-frequency logging signal also includes steps S411 to S412:

[0084] Step S411, calculating the resolution correlation coefficient between the second low-frequency logging signal and the single-element logging data; Step S412, selecting the second low-frequency logging signal corresponding to the largest resolution correlation coefficient as the predicted low-frequency logging signal.

[0085] It should be noted that vertical resolution refers to the sampling interval or resolution of logging signals or logging element data in the vertical direction, and is usually used to describe the ability of signals or data to express details in the depth direction of the formation. The higher the vertical resolution, the smaller the depth difference between each sampling point, and the more accurately the formation changes can be captured; lower vertical resolution means that the depth interval between each sampling point is larger, and small-scale formation changes may not be effectively captured. Resolution matching is performed to screen out the second low-frequency logging signal that is more matched with the logging element in the vertical resolution, and to ensure the consistency of the logging element data and the second low-frequency logging signal in spatial scale.

[0086] Specifically, the resolution correlation coefficient between each second low-frequency logging signal and the single-element logging data is calculated, and the second low-frequency logging signal corresponding to the largest resolution correlation coefficient is selected as the predicted low-frequency logging signal.

[0087] For example, the resolution correlation coefficients of the second low-frequency logging signals GR_M6, GR_M5, GR_M4 and GR_M3 with the single-element logging data of K, Mg, Fe and other logging elements are calculated. Among all the logging elements, the single-element logging data of element K has a higher resolution correlation coefficient with all the second low-frequency logging signals. Then, the synthetic logging curve with the highest resolution correlation coefficient with element K is selected from all the second low-frequency logging signals. Figure 3 , the first track in the figure includes the wellbore curve (CAL) and the drill bit diameter curve (BIT); the second track is the natural gamma curve (GR), the third track is the second low-frequency logging signal GR_M6 curve and the single element logging data curve of the logging element K in the first logging element data, the fourth track is the second low-frequency logging signal GR_M5 curve and the data curve of the logging element K in the first logging element data, the fifth track is the second low-frequency logging signal GR_M4 curve and the data curve of the logging element K in the first logging element data, and the sixth track is the second low-frequency logging signal GR_M3 curve and the data curve of the logging element K in the first logging element data. By calculating GR_M6, GR_M5, GR_M4, the resolution correlation coefficients between GR_M3 and the logging potassium (K) element are 0.54, 0.56, 0.57, and 0.64, respectively. The resolution correlation between GR_M3 and the logging potassium (K) element is the best, and it is believed that the vertical resolutions of the two are relatively consistent. Reference Figure 3 By comparing the vertical resolution difference between GR_M6, GR_M5, GR_M4, GR_M3 in the second low-frequency logging signal and the K element data in the first logging element data, it is found that Figure 3 The curve similarity between the last signal GR_M3 and the K element data in the first logging element data is high, indicating that the resolution is relatively consistent. Therefore, GR_M3 can be selected as the second low-frequency logging signal to facilitate the subsequent correlation screening of the second logging element data.

[0088] In this embodiment, by comparing the vertical resolution difference between the second low-frequency logging signal and the first logging element data, the low-frequency logging signal with the most consistent vertical resolution with the logging element data can be screened out. The selected low-frequency logging signal matches the sampling interval or detail performance of the logging element data in the depth direction, ensuring the consistency between the data, thereby improving the accuracy of the logging curve correction.

[0089] Based on the first and second embodiments of the present application, in the third embodiment of the present application, the same or similar contents as those of the first and / or second embodiments can be referred to the above description, and will not be described in detail later. On this basis, step S42, based on the data correlation between the predicted low-frequency logging signal and the first logging element data, selects the predicted single-element logging data from all the single-element logging data, including steps S421 to S423:

[0090] Step S421, calculating the correlation coefficient between the predicted low-frequency logging signal and the single-element logging data;

[0091] It should be noted that the correlation coefficient is a statistic used to quantify the strength of the linear relationship between two variables. Its value ranges from -1 to 1. The closer the value is to 1, the stronger the positive correlation between the two variables. The closer the value is to -1, the stronger the negative correlation is. The value close to 0 indicates that there is almost no linear relationship. By calculating the correlation coefficient between the predicted low-frequency logging signal and the single-element logging data, it can be determined which logging element data is closely related to the change of the logging signal. Specify a correlation coefficient threshold to characterize the correlation between the logging element and the low-frequency logging signal. When the correlation coefficient is greater than the threshold, the correlation between the two is considered to be good.

[0092] Specifically, the correlation coefficient between the predicted low-frequency logging signal and all single-element logging data is calculated.

[0093] Step S422, comparing the correlation coefficient with a preset correlation coefficient threshold; Step S45, selecting the single element logging data with a correlation coefficient greater than the correlation coefficient threshold as the predicted single element logging data.

[0094] It should be noted that the selection of the correlation coefficient threshold is set according to actual needs and data characteristics and is not limited here.

[0095] Specifically, all calculated correlation coefficients are compared with the correlation coefficient threshold, and single-element logging data with correlation coefficients greater than the correlation coefficient threshold are selected as predicted single-element logging data.

[0096] For example, by calculating the correlation coefficients between the predicted low-frequency logging signal GR_M3 and single-element logging data of silicon (Si), calcium (Ca), iron (Fe), aluminum (Al), magnesium (Mg), potassium (K), etc., it is found that GR_M3 has the best correlation with the logging elements iron (Fe) and potassium (K). Figure 4, the first track in the figure includes the wellbore curve (CAL) and the drill bit diameter curve (BIT); the second track is the natural gamma curve (GR), the third track is the predicted low-frequency logging signal GR_M3 curve and the Si content curve, the fourth track is the predicted low-frequency logging signal GR_M3 curve and the Ca content curve, the fifth track is the predicted low-frequency logging signal GR_M3 curve and the Fe content curve, the sixth track is the predicted low-frequency logging signal GR_M3 curve and the Al content curve, the seventh track is the predicted low-frequency logging signal GR_M3 curve and the Mg content curve, and the eighth track is the predicted low-frequency logging signal GR_M3 curve and the K content curve. Figure 4 It can be seen that the similarity between the GR_M3 curve and the logging element K content curve and the logging element Fe content curve is the highest. Therefore, it can be concluded that the correlation between the GR_M3 curve and the logging element K content curve and the logging element Fe content curve is the best. Therefore, the logging elements iron (Fe) and potassium (K) are determined as the input curves for predicting the third low-frequency logging signal.

[0097] In this embodiment, the correlation between the logging elements and the logging signals is calculated to screen out the logging elements with good correlation with the logging signals, and then the logging elements are used as input features for building the model, so as to facilitate more accurate correction of the logging curves in the future.

[0098] Based on the first, second and third embodiments of the present application, in the fourth embodiment of the present application, the same or similar contents as those in the first and / or second and / or third embodiments can be referred to the above description and will not be described in detail later. On this basis, step S50 constructs a prediction model based on predicted single element logging data and predicted low-frequency logging signals, including steps S51 to S53:

[0099] Step S51, using the predicted single-element logging data as input features; Step S52, using the predicted low-frequency logging signal as the target output; Step S53, using a machine learning algorithm to build a prediction model based on the input features and the target output.

[0100] Specifically, the selected predicted single-element logging data is used as the input feature of the machine learning model to train the model to identify the change pattern of the formation. The predicted low-frequency logging signal (such as GR_M3 or other synthetic low-frequency signals) will be used as the target output to compare with the input features to help the model learn the relationship between input and output, thereby predicting the change trend of the low-frequency logging signal. Select a suitable machine learning algorithm to train the input features and target outputs, optimize the model parameters, and minimize the prediction error. Through the trained model, the low-frequency logging signal can be effectively predicted based on the logging element data, and applied to the subsequent logging data correction and analysis.

[0101] Optionally, based on the input features and the target output, a prediction model is constructed using a KNN (K-Nearest Neighbors) neural network algorithm or a BP (Back Propagation Neural Network) neural network algorithm.

[0102] For example, the data of logging element K is used as input features, and the predicted low-frequency logging signal is used as the target output. The KNN neural network method is used to construct a model for predicting low-frequency natural gamma logging signals (GR_M3) using logging element potassium (K). Figure 5 , the first track in the figure includes the wellbore curve (CAL) and the drill bit diameter curve (BIT); the second track is the natural gamma curve (GR); the solid line in the third track is the GR_M3 curve, and the dotted line is the low-frequency natural gamma signal (GR_M3_PRED) predicted by the KNN neural network method; the fourth track is the GR_M3 curve and the logging element K content curve; the fifth track is the GR_M3 curve and the logging element Fe content curve; it can be seen from the figure that the GR_M3 curve has a good correlation with the logging element K and Fe content curves, and the consistency between the GR_M3_Pred curve constructed from this and the GR_M3 curve is also very good, that is, the prediction model established by the KNN neural network method is accurate and reliable.

[0103] Exemplarily, in combination with one or more of the above embodiments, the execution process and effect of the technical solution of the present invention are described as follows:

[0104] Specifically, refer to Figure 7 First, the element logging data (first logging element data) is recorded using a detection instrument, and then the EMD signal is decomposed for the logging curve to be corrected, and then the first low-frequency logging signal obtained by the decomposition is synthesized to obtain multiple second low-frequency logging signals, and a predicted low-frequency logging signal matching the vertical resolution of the first logging element data is screened out from the second low-frequency logging signal, and then the element logging data (predicted single-element logging data) with a good correlation with the predicted low-frequency logging signal is selected, and a prediction model is constructed based on the predicted single-element logging data and the predicted low-frequency logging signal, and the prediction model is used to predict the low-frequency logging signal of the expanded diameter section (third low-frequency logging signal), and the low-frequency logging signal is synthesized with the high-frequency logging signal obtained by signal decomposition to obtain the corrected curve of the expanded diameter section.

[0105] For example, the final corrected logging curve is referenced to Figure 6, the first track in the figure includes the wellbore curve (CAL) and the drill bit diameter curve (BIT); the second track is the high-frequency logging signal IMF1 signal track obtained by decomposition; the third track is the high-frequency logging curve IMF2 signal track obtained by decomposition; the fourth track is the IMF3 signal track, the fifth track is the IMF4 signal track; the sixth track is the IMF5 signal track, the seventh track is the IMF6 signal track, and the eighth track is the IMF7 signal track; the ninth track is the GR_M3 curve and the logging element K content curve, in which the solid line is the GR_M3 curve, and the dotted line is the K element content curve; the tenth track is the GR_M3 curve and the logging element Fe content curve, in which the solid line is the GR_M3 curve, and the dotted line is the Fe element content curve; the eleventh track is the corrected natural gamma curve GR_CORRECT and the original curve GR. The third low-frequency logging signal predicted by the expanded layer section is synthesized with the high-frequency logging signals IMF1 and IMF2 in the decomposed full-band logging signal to obtain the corrected natural gamma curve GR_CORRECT. From Figure 6 It can be seen that at 3845-3850 meters, although the well diameter has been seriously expanded, the curves before and after the natural gamma correction are basically overlapped, indicating that during the logging process in this layer, the natural gamma instrument may not be centered, but synchronized with the expansion direction, so that the measurement signal better reflects the real information of the formation, which also shows that the accuracy of the logging curve predicted by the present invention is good. At 3850-3865 meters, the corrected natural gamma curve GR_CORRECT is significantly larger, indicating that the natural gamma logging instrument should be centered in this layer. Due to the influence of the expansion, the measured value is significantly smaller, and the original measured value can no longer reflect the real information of the formation. At this time, the logging curve predicted by the present invention is more accurate than the original measured value.

[0106] This application also provides a logging curve environment correction device, please refer to Figure 8 , the logging curve environment correction device includes:

[0107] The data acquisition module 10 is used to acquire first logging element data, wherein the first logging element data includes single-element logging data corresponding to the elements contained in the cuttings sample; perform signal decomposition on the logging curve of the unexpanded well section to acquire a full-band logging signal, wherein the full-band logging signal includes a plurality of first low-frequency logging signals; synthesize the first low-frequency logging signal to obtain a plurality of second low-frequency logging signals; based on the correlation between the first logging element data and the second low-frequency logging signal, screen out predicted single-element logging data from the single-element logging data, and screen out predicted low-frequency logging signals from the second low-frequency logging signal;

[0108] A model building module 20, for building a prediction model based on predicted single element logging data and predicted low frequency logging signals;

[0109] The curve correction module 30 is used to predict the third low-frequency logging signal of the expanded diameter layer section using the prediction model, synthesize the third low-frequency logging signal with the high-frequency logging signal in the full-frequency band logging signal, and obtain a corrected logging curve.

[0110] Finally, it should be noted that the steps in the method of the embodiment of the present invention can be adjusted in order, combined, and deleted according to actual needs.

[0111] In addition, a person of ordinary skill in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing the hardware related to the device through a program, and the program may be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0112] In addition, those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by functional modules in the device, which will not be elaborated herein.

[0113] What is disclosed above is only a preferred embodiment of the present invention, which certainly cannot be used to limit the scope of rights of the present invention. A person skilled in the art can understand that all or part of the processes of the above embodiments and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A method for environmental correction of well logging curves, characterized in that: The logging curve environment correction method comprises: Acquire first logging element data, wherein the first logging element data includes single element logging data corresponding one-to-one to elements contained in the cuttings sample; Performing signal decomposition on the logging curve of the unexpanded well section to obtain a full-band logging signal, wherein the full-band logging signal includes a plurality of first low-frequency logging signals and a high-frequency logging signal; synthesizing the first low-frequency logging signals to obtain a plurality of second low-frequency logging signals; Based on the correlation between the first logging element data and the second low-frequency logging signal, predicting single-element logging data is screened out from the single-element logging data, and predicting low-frequency logging signals are screened out from the second low-frequency logging signals; Constructing a prediction model based on the predicted single-element logging data and the predicted low-frequency logging signal; The prediction model is used to predict the third low-frequency logging signal of the expanded diameter layer section, and the third low-frequency logging signal is synthesized with the high-frequency logging signal to obtain a corrected logging curve.

2. The method for well logging curve environmental correction according to claim 1, characterized in that: The step of synthesizing the first low-frequency logging signal comprises: The frequencies of the first low-frequency logging signals are sequentially stacked and synthesized from low to high.

3. The method for well logging curve environmental correction according to claim 1, characterized in that: The step of screening out predicted single-element logging data from the single-element logging data based on the correlation between the first logging element data and the second low-frequency logging signal, and screening out predicted low-frequency logging signals from the second low-frequency logging signals comprises: Based on the resolution correlation between the second low-frequency logging signal and the first logging element data, screening out the predicted low-frequency logging signal that matches the vertical resolution of the first logging element data from the second low-frequency logging signal; Based on the data correlation between the predicted low-frequency logging signal and the first logging element data, the predicted single-element logging data is screened out from the single-element logging data.

4. The method for well logging curve environmental correction according to claim 3, characterized in that: The step of screening out the predicted low-frequency logging signal that matches the vertical resolution of the first logging element data from the second low-frequency logging signal based on the resolution correlation between the second low-frequency logging signal and the first logging element data comprises: Calculating a resolution correlation coefficient between the second low-frequency logging signal and the single-element logging data; The second low-frequency logging signal corresponding to the largest resolution correlation coefficient is selected as the predicted low-frequency logging signal.

5. The method for environmental correction of well logging curves according to claim 3, characterized in that: The step of selecting predicted single-element logging data from all the single-element logging data based on the data correlation between the predicted low-frequency logging signal and the first logging element data comprises: Calculating the correlation coefficient between the predicted low-frequency logging signal and the single-element logging data; comparing the correlation coefficient with a preset correlation coefficient threshold; The single-element logging data whose correlation coefficient is greater than the correlation coefficient threshold is selected as the predicted single-element logging data.

6. The method for well logging curve environmental correction according to claim 1, characterized in that: The step of obtaining the first logging element data comprises: In the logging area, selecting the rock cutting samples at preset intervals; The single element logging data corresponding to the elements contained in the rock cuttings sample is obtained using a measuring tool.

7. The method for well logging curve environmental correction according to claim 1, characterized in that: The step of performing signal decomposition on the logging curve of the unexpanded well section comprises: Perform EMD signal decomposition or wavelet decomposition on the logging curve of the unexpanded well section.

8. The method for well logging curve environmental correction according to claim 1, characterized in that: The step of constructing a prediction model based on the predicted single-element logging data and the predicted low-frequency logging signal comprises: Using the predicted single element logging data as input features; outputting the predicted low-frequency logging signal as a target; Based on the input features and the target output, a prediction model is constructed using a machine learning algorithm.

9. The method for well logging curve environmental correction according to claim 8, characterized in that: The steps of using a machine learning algorithm to build a prediction model include: Use KNN neural network algorithm or BP neural network algorithm to build a prediction model.

10. A logging curve environment correction device, characterized in that: include: A data acquisition module is used to acquire first logging element data, wherein the first logging element data includes single-element logging data corresponding one-to-one to the elements contained in the cuttings sample; perform signal decomposition on the logging curve of the unexpanded well section to acquire a full-band logging signal, wherein the full-band logging signal includes a plurality of first low-frequency logging signals and a high-frequency logging signal; synthesize the first low-frequency logging signal to obtain a plurality of second low-frequency logging signals; based on the correlation between the first logging element data and the second low-frequency logging signal, screen out predicted single-element logging data from the single-element logging data, and screen out predicted low-frequency logging signals from the second low-frequency logging signal; A model building module, used to build a prediction model based on the predicted single-element logging data and the predicted low-frequency logging signal; The curve correction module is used to use the prediction model to predict the third low-frequency logging signal of the expanded diameter layer section, synthesize the third low-frequency logging signal with the high-frequency logging signal, and obtain a corrected logging curve.

Images