Method, device and equipment for determining formation resistivity of post-casing transient electromagnetic logging curve and medium

Abstract

The invention discloses a formation resistivity determination method and device for a post-casing transient electromagnetic logging curve, equipment and a medium, and relates to the technical field of oil exploration and development, and the formation resistivity determination method comprises the following steps: obtaining a post-casing electromagnetic logging curve, and carrying out time sequence conversion and data preprocessing to obtain time sequence data; converting the data into frequency domain data by adopting wavelet transform, and performing multi-scale refinement splitting to obtain low-frequency features and high-frequency features; fitting the low-frequency features and the high-frequency features by using a symbol regression method to obtain a nonlinear high-dimensional symbol regression formula for the low-frequency data and a nonlinear high-dimensional symbol regression formula for the high-frequency data, and generating a post-casing electromagnetic logging curve nonlinear fitting formula; the mapping relation between the post-casing transient electromagnetic curve and the formation resistivity is determined through the post-casing electromagnetic logging curve nonlinear fitting formula, the formation resistivity is obtained, the uncertainty of manual selection of model parameters is avoided, the capacity of processing nonlinear problems is improved, and the accuracy and reliability of high-dimensional relation mapping and nonlinear fitting are improved.

Description

Technical Field

[0001] The invention relates to the technical field of petroleum exploration and development, and in particular to a method, device, equipment and medium for determining formation resistivity of a post-casing transient electromagnetic logging curve. Background Art

[0002] Post-casing logging is a geophysical logging method for the formation in the well after the drilling, casing and cementing operations are completed. Due to the complexity and variability of the formation structure, the formation resistivity often shows a highly nonlinear characteristic. Conventional logging curve fitting methods include linear fitting, polynomial fitting, spline fitting, exponential fitting, logarithmic fitting, Gaussian fitting, neural network fitting and support vector machine fitting. However, traditional linear fitting methods have great limitations when dealing with such nonlinear data, such as: (1) the structure of the model needs to be artificially selected, which may cause the model to be too simple to capture the complex characteristics of the data, or too complex to cause overfitting; (2) Linear fitting and polynomial fitting have limited ability to capture strong nonlinear relationships; (3) It is difficult to achieve high-dimensional relationship mapping and nonlinear fitting, or although high-dimensional relationship mapping and nonlinear fitting can be performed, the interpretable formula structure cannot be fitted; (4) The accuracy of formula fitting is insufficient, making it difficult to accurately characterize the relationship between post-casing transient electromagnetic logging curves and formation resistivity.

[0003] It can be seen from the above that how to avoid the uncertainty of artificially selected model parameters, improve the ability to deal with nonlinear problems and the accuracy and reliability of high-dimensional relationship mapping and nonlinear fitting, and accurately characterize the relationship between the post-casing transient electromagnetic logging curve and the formation resistivity are problems to be solved in this field. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method, device, equipment and medium for determining the formation resistivity of the post-casing transient electromagnetic logging curve, which can avoid the uncertainty of artificially selected model parameters, improve the ability to handle nonlinear problems and the accuracy and reliability of high-dimensional relationship mapping and nonlinear fitting, and accurately characterize the relationship between the post-casing transient electromagnetic logging curve and the formation resistivity. The specific scheme is as follows:

[0005] In a first aspect, the present application discloses a method for determining formation resistivity of a post-casing transient electromagnetic logging curve, comprising:

[0006] Acquire the post-casing electromagnetic logging curve of the oil well, and perform time series conversion and data preprocessing on the post-casing electromagnetic logging curve to obtain preprocessed time series data;

[0007] The time domain data is converted into frequency domain data by wavelet transform, and multi-scale refinement and splitting is performed to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions;

[0008] The low-frequency feature and the high-frequency feature are fitted by using a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and a nonlinear fitting formula for a post-casing electromagnetic logging curve is generated based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data;

[0009] The casing transient electromagnetic curve of the oil well to be tested is obtained, and the mapping relationship between the casing transient electromagnetic curve and the formation resistivity of the oil well to be tested is determined by using the nonlinear fitting formula of the casing electromagnetic logging curve to obtain the formation resistivity.

[0010] Optionally, performing time series conversion and data preprocessing on the post-casing electromagnetic logging curve to obtain preprocessed time series data includes:

[0011] Performing time series conversion on the data in the post-casing electromagnetic logging curve to obtain converted time series data;

[0012] The converted time series data is normalized, feature screened, and cleaned to obtain the preprocessed time series data.

[0013] Optionally, the using of wavelet transform to transform the time domain data into frequency domain data and performing multi-scale refinement and splitting includes:

[0014] The time domain data is transformed into frequency domain data by wavelet transform and using a preset data transformation formula, and multi-scale refinement and splitting is performed; the preset data transformation formula is:

[0015] ;

[0016] ;

[0017] in, is the result of continuous wavelet transform, a is the scaling parameter, s(t) is the input data of a time series with a variable of time t, for The complex conjugate function of , b is the location parameter, is the wavelet function, is a constant of the center frequency, and i is an imaginary number that determines the information in the time domain or space domain in the transformation result.

[0018] Optionally, the using of a symbolic regression method to fit the low-frequency features and the high-frequency features to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data includes:

[0019] Initializing the population of the low-frequency features and the high-frequency features by using a symbolic regression method, randomly generating a group of symbolic expressions as an initial population, screening the initial population to obtain a screened population, and performing crossover and mutation on the screened population to obtain a progeny population;

[0020] Record the current number of iterations, and repeat the step of randomly generating a set of symbolic expressions as the initial population until the current number of iterations is equal to a preset iteration setting value, so as to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data.

[0021] Optionally, the using of a symbolic regression method to initialize the population of the low-frequency features and the high-frequency features, randomly generating a group of symbolic expressions as an initial population, and screening the initial population to obtain a screened population includes:

[0022] Using a symbolic regression method to initialize the population, encoding the symbolic expressions in the low-frequency features and the high-frequency features into chromosomes in a genetic algorithm, and randomly generating a group of symbolic expressions as an initial population;

[0023] The preset fitness value calculation formula is used to calculate each individual in the initial population to obtain each fitness value, and the initial population is screened based on each fitness value in a roulette manner to obtain a screened population.

[0024] Optionally, the fitness value calculation formula is:

[0025] ;

[0026] Among them, f is the individual fitting formula, is the true value when the depth is i, is the predicted value when the depth is i according to the formula of the individual, is the average of all predicted values.

[0027] Optionally, the nonlinear fitting formula for the post-casing electromagnetic logging curve is generated based on the nonlinear high-dimensional symbolic regression formula for the low-frequency data and the nonlinear high-dimensional symbolic regression formula for the high-frequency data, including:

[0028] The preset first weight coefficient and the second weight coefficient are respectively combined with the nonlinear high-dimensional symbol regression formula for low-frequency data and the nonlinear high-dimensional symbol regression formula for high-frequency data, and the combined nonlinear high-dimensional symbol regression formula for low-frequency data and the nonlinear high-dimensional symbol regression formula for high-frequency data are added to obtain a nonlinear fitting formula for the post-casing electromagnetic logging curve.

[0029] In a second aspect, the present application discloses a formation resistivity determination device for a post-casing transient electromagnetic logging curve, comprising:

[0030] A data processing module is used to obtain the casing electromagnetic logging curve of the oil well, and perform time series conversion and data preprocessing on the casing electromagnetic logging curve to obtain preprocessed time series data;

[0031] A multi-scale refinement and splitting module is used to transform the time domain data into frequency domain data by wavelet transform, and perform multi-scale refinement and splitting to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions;

[0032] A fitting formula generation module is used to fit the low-frequency feature and the high-frequency feature using a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and to generate a nonlinear fitting formula for a post-casing electromagnetic logging curve based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data;

[0033] The formation resistivity determination module is used to obtain the post-casing transient electromagnetic curve of the oil well to be tested, and use the nonlinear fitting formula of the post-casing electromagnetic logging curve to determine the mapping relationship between the post-casing transient electromagnetic curve and the formation resistivity of the oil well to be tested, so as to obtain the formation resistivity.

[0034] In a third aspect, the present application discloses an electronic device, comprising:

[0035] Memory, used to store computer programs;

[0036] The processor is used to execute the computer program to implement the above-mentioned method for determining the formation resistivity of the post-casing transient electromagnetic logging curve.

[0037] In a fourth aspect, the present application discloses a computer storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the steps of the method for determining formation resistivity of the post-casing transient electromagnetic logging curve disclosed above are implemented.

[0038] It can be seen that the present application provides a method for determining the formation resistivity of a post-casing transient electromagnetic logging curve, including obtaining a post-casing electromagnetic logging curve of an oil well, and performing time series conversion and data preprocessing on the post-casing electromagnetic logging curve to obtain preprocessed time series data; using wavelet transform to convert the time domain data into frequency domain data, and performing multi-scale refinement and splitting to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions; using a symbolic regression method to fit the low-frequency features and the high-frequency features to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and generating a nonlinear fitting formula for the post-casing electromagnetic logging curve based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data; obtaining a post-casing transient electromagnetic curve of the oil well to be tested, and using the nonlinear fitting formula for the post-casing electromagnetic logging curve to determine the mapping relationship between the post-casing transient electromagnetic curve and the formation resistivity of the oil well to be tested to obtain the formation resistivity. The present application adopts wavelet transform to split the multi-scale time domain data into low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions, and uses symbolic regression method for fitting. The use of wavelet transform can enable the symbolic regression method to effectively process the nonlinear characteristics of data in different frequency domains, so that the fitting is more accurate. The symbolic regression method is introduced for fitting, so that it is possible to automatically discover and construct mathematical expressions suitable for data without presetting a specific function form, so as to find a more accurate and concise relationship in a complex model space, avoid the uncertainty of manually selecting model parameters, obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, improve the ability to handle nonlinear problems and the accuracy and reliability of high-dimensional relationship mapping and nonlinear fitting, generate a nonlinear fitting formula for post-casing electromagnetic logging curves, and use the nonlinear fitting formula for post-casing electromagnetic logging curves to determine the mapping relationship between post-casing transient electromagnetic curves and formation resistivity, and obtain formation resistivity, so it has stronger generalization ability and interpretability, and can accurately characterize the relationship between post-casing transient electromagnetic logging curves and formation resistivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] 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 embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0040] Figure 1 This is a flow chart of a method for determining formation resistivity of a post-casing transient electromagnetic logging curve disclosed in the present application;

[0041] Figure 2 An example diagram of formation resistivity of an original signal disclosed in this application;

[0042] Figure 3 This is an example diagram of a low-frequency flow (low-frequency component) of formation resistivity disclosed in the present application;

[0043] Figure 4 This is an example diagram of a high-frequency flow (high-frequency component) of formation resistivity disclosed in the present application;

[0044] Figure 5 This is a flow chart of a method for determining formation resistivity of a post-casing transient electromagnetic logging curve disclosed in the present application;

[0045] Figure 6 A schematic diagram of a crossover and mutation process disclosed in this application;

[0046] Figure 7 A detailed flow chart of formation resistivity determination disclosed in the present application;

[0047] Figure 8 This is a schematic diagram of the structure of a device for determining formation resistivity of a post-casing transient electromagnetic logging curve disclosed in the present application;

[0048] Fig. 9 A structural diagram of an electronic device provided for this application. DETAILED DESCRIPTION

[0049] 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.

[0050] Post-casing logging is a geophysical logging method for the formation in the well after the drilling, casing and cementing operations are completed. Due to the complexity and variability of the formation structure, the formation resistivity often shows a highly nonlinear characteristic. Conventional logging curve fitting methods include linear fitting, polynomial fitting, spline fitting, exponential fitting, logarithmic fitting, Gaussian fitting, neural network fitting and support vector machine fitting. However, traditional linear fitting methods have great limitations when dealing with such nonlinear data, such as: (1) the structure of the model needs to be artificially selected, which may cause the model to be too simple to capture the complex characteristics of the data, or too complex to cause overfitting; (2) Linear fitting and polynomial fitting have limited ability to capture strong nonlinear relationships; (3) It is difficult to achieve high-dimensional relationship mapping and nonlinear fitting, or although high-dimensional relationship mapping and nonlinear fitting can be performed, the interpretable formula structure cannot be fitted; (4) The accuracy of formula fitting is insufficient, making it difficult to accurately characterize the relationship between post-casing transient electromagnetic logging curves and formation resistivity. It can be seen from the above that how to avoid the uncertainty of artificially selected model parameters, improve the ability to deal with nonlinear problems and the accuracy and reliability of high-dimensional relationship mapping and nonlinear fitting, and accurately characterize the relationship between the post-casing transient electromagnetic logging curve and the formation resistivity are problems to be solved in this field.

[0051] See also Figure 1 As shown, the embodiment of the present invention discloses a method for determining formation resistivity of a post-casing transient electromagnetic logging curve, which may specifically include:

[0052] Step S11: Acquire the casing electromagnetic logging curve of the oil well, and perform time series conversion and data preprocessing on the casing electromagnetic logging curve to obtain preprocessed time series data.

[0053] In this embodiment, the post-casing electromagnetic logging curve of the oil well is obtained, the data in the post-casing electromagnetic logging curve is converted into time series to obtain converted time series data, and the converted time series data is normalized, feature screened and data cleaned to obtain the pre-processed time series data.

[0054] Specifically, in the data preprocessing stage, the post-casing electromagnetic logging curve data is preprocessed to convert the data into time series data, and the normalization and feature screening are completed. Due to the existence of singular points and noise points, data cleaning is also required to regularize the data to eliminate the dimensional differences between features and reduce the scale differences between features. Among them, the time series conversion is as follows: (1) each post-casing electromagnetic logging curve is arranged in depth order, and the time series length is determined to be 1m according to the reservoir heterogeneity, that is, 40 sample lengths (when each sample interval is 0.025m); (2) according to the time series length, the logging curve is converted from (depth length, feature number) to (depth length-time series length+1, time series length, feature number) through the time window method; normalization processing is: using a unified normalization method, using a standardization method to convert the data into a standard normal distribution with a mean of 0 and a standard deviation of 1; data cleaning is as follows: (1) removing outliers, by using the Z Score to detect and remove data points that deviate from the normal range; (2) processing missing values, deleting data samples containing missing values; (3) smoothing data, using the moving average technique to smooth the curve and reduce the impact of noise on data analysis.

[0055] Step S12: using wavelet transform to transform the time domain data into frequency domain data, and performing multi-scale refinement and splitting to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions.

[0056] In this embodiment, wavelet transform is adopted and a preset data transformation formula is used to transform the time domain data into frequency domain data, and multi-scale refinement and splitting are performed to obtain low-frequency features that meet the preset low-frequency conditions and high-frequency features that meet the preset high-frequency conditions; the preset data transformation formula is:

[0057] ;

[0058] ;

[0059] in, is the result of continuous wavelet transform, a is the scaling parameter, s(t) is the input data of a time series with a variable of time t, for The complex conjugate function of , b is the location parameter, is the wavelet function, is a constant of the center frequency, and i is an imaginary number that determines the information in the time domain or space domain in the transformation result.

[0060] Wavelet transform (WT) is a transform analysis method that inherits and develops the localization idea of ​​short-time Fourier transform, while overcoming the shortcomings of window size not changing with frequency. It can provide a "time-frequency" window that changes with frequency, and is an ideal tool for signal time-frequency analysis and processing. Wavelet transform can fully highlight the frequency domain characteristics of the post-casing logging curve and realize localized analysis of time and frequency characteristics. This method uses wavelet transform to convert input data from time domain data to frequency domain data in advance, thereby realizing frequency splitting and multi-scale refinement, and subdividing time domain data into low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions.

[0061] Step S13: Use a symbolic regression method to fit the low-frequency features and the high-frequency features to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and generate a nonlinear fitting formula for the post-casing electromagnetic logging curve based on the nonlinear high-dimensional symbolic regression formula for the low-frequency data and the nonlinear high-dimensional symbolic regression formula for the high-frequency data.

[0062] In this embodiment, after obtaining the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data, the preset first weight coefficient and the second weight coefficient are respectively combined with the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data, and the combined nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data are added to obtain the nonlinear fitting formula for the post-casing electromagnetic logging curve.

[0063] Step S14: obtaining the casing transient electromagnetic curve of the oil well to be tested, and determining the mapping relationship between the casing transient electromagnetic curve and the formation resistivity of the oil well to be tested by using the nonlinear fitting formula of the casing electromagnetic logging curve to obtain the formation resistivity.

[0064] The purpose of this invention is to introduce wavelet transformation and symbolic regression methods to carry out the nonlinear formula fitting process of the post-casing electromagnetic logging curve. Therefore, taking 30 post-casing electromagnetic logging curves of an oil field in the eastern alluvial plain as an example, the specific steps of feature dimensionality reduction are as follows:

[0065] Step 1: Data preprocessing, performing time series conversion, normalization, feature screening and data cleaning on the 30-dimensional post-casing electromagnetic logging curve features of the post-casing electromagnetic logging curve one by one;

[0066] Step 2: Perform wavelet decomposition to separate high-frequency flow from low-frequency flow. The formation resistivity signal curve of the original signal is as follows: Figure 2As shown in the figure, the low-frequency flow (low-frequency component) of the formation resistivity signal curve of the low-frequency component is as follows Figure 3 As shown in the figure, the high-frequency flow (high-frequency component) of the high-frequency component of the formation resistivity signal curve is as follows Figure 4 As shown;

[0067] Step 3: Perform symbolic regression: For example arrive They are STEMUB01 to STEMUB017 of the electromagnetic logging curves after casing, and the progeny population is screened out by symbolic regression, and the nonlinear high-dimensional symbolic regression formula for low-frequency data is obtained as follows:

[0068] ;

[0069] The nonlinear high-dimensional symbolic regression formula for high-frequency data is obtained as follows:

[0070] ;

[0071] The nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data are combined to generate the nonlinear fitting formula for the post-casing electromagnetic logging curve; the nonlinear fitting formula for the post-casing electromagnetic logging curve is:.

[0072] ;

[0073] Among them, a and b are weight coefficients.

[0074] It is worth noting that here, 30 post-casing electromagnetic logging curves are taken as an example, and the symbolic regression method is used to generate the above-mentioned nonlinear high-dimensional symbolic regression formula for low-frequency data, the nonlinear high-dimensional symbolic regression formula for high-frequency data, and the nonlinear fitting formula for post-casing electromagnetic logging curves.

[0075] Symbolic regression is a powerful data-driven model building method. Unlike general parameter regression methods, symbolic regression does not require the initial structure of the model or formula to be specified in advance. It can simultaneously search for the form of mathematical analytical expressions and the values ​​of internal parameters through methods such as genetic programming. Therefore, it does not rely on existing knowledge to a large extent, but can generate a series of formulas with reference value in understanding physical systems. For physical systems where the relationship between variables is vague and the overall change is not understood deeply, symbolic regression becomes a powerful tool for establishing symbolic models between parameters. Among them, symbolic models refer to function expressions represented by letters and operators. Symbolic regression is different from common regression methods. It does not require the form of regression formula to be set in advance, but establishes a regression model through the selection, crossover, mutation and other steps of genetic algorithms. Based on the wavelet transform method, high-frequency features and low-frequency features are separated, and symbolic regression of high-frequency and low-frequency data is performed simultaneously, so as to better realize the nonlinear high-dimensional relationship mapping in different frequency domains and obtain a more accurate nonlinear fitting formula for post-casing electromagnetic logging curves.

[0076] Therefore, it can be seen that the above formula is only one of the nonlinear high-dimensional symbolic regression formula for low-frequency data, the nonlinear high-dimensional symbolic regression formula for high-frequency data, and the nonlinear fitting formula for the post-casing electromagnetic logging curve. It does not mean that the above formula is the nonlinear high-dimensional symbolic regression formula for low-frequency data, the nonlinear high-dimensional symbolic regression formula for high-frequency data, and the nonlinear fitting formula for the post-casing electromagnetic logging curve.

[0077] In this embodiment, a post-casing electromagnetic logging curve of an oil well is obtained, and the post-casing electromagnetic logging curve is converted into a time series and data preprocessed to obtain preprocessed time series data; the time domain data is converted into frequency domain data by wavelet transform, and multi-scale refinement and splitting are performed to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions; the low-frequency features and the high-frequency features are fitted by a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and a nonlinear fitting formula for the post-casing electromagnetic logging curve is generated based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data; the post-casing transient electromagnetic curve of the oil well to be tested is obtained, and the mapping relationship between the post-casing transient electromagnetic curve and the formation resistivity of the oil well to be tested is determined by using the nonlinear fitting formula for the post-casing electromagnetic logging curve to obtain the formation resistivity. The present application adopts wavelet transform to split the multi-scale time domain data into low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions, and uses symbolic regression method for fitting. The use of wavelet transform can enable the symbolic regression method to effectively process the nonlinear characteristics of data in different frequency domains, so that the fitting is more accurate. The symbolic regression method is introduced for fitting, so that it is possible to automatically discover and construct mathematical expressions suitable for data without presetting a specific function form, so as to find a more accurate and concise relationship in a complex model space, avoid the uncertainty of manually selecting model parameters, obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, improve the ability to handle nonlinear problems and the accuracy and reliability of high-dimensional relationship mapping and nonlinear fitting, generate a nonlinear fitting formula for post-casing electromagnetic logging curves, and use the nonlinear fitting formula for post-casing electromagnetic logging curves to determine the mapping relationship between post-casing transient electromagnetic curves and formation resistivity, and obtain formation resistivity, so it has stronger generalization ability and interpretability, and can accurately characterize the relationship between post-casing transient electromagnetic logging curves and formation resistivity.

[0078] See also Figure 5 As shown, the embodiment of the present invention discloses a method for determining formation resistivity of a post-casing transient electromagnetic logging curve, which may specifically include:

[0079] Step S21: Acquire the casing electromagnetic logging curve of the oil well, and perform time series conversion and data preprocessing on the casing electromagnetic logging curve to obtain preprocessed time series data.

[0080] Step S22: transforming the time domain data into frequency domain data by wavelet transform, and performing multi-scale refinement and splitting to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions.

[0081] Step S23: using a symbolic regression method to initialize the population of the low-frequency features and the high-frequency features, randomly generating a group of symbolic expressions as an initial population, screening the initial population to obtain a screened population, and performing crossover and mutation on the screened population to obtain a progeny population.

[0082] In this embodiment, a symbolic regression method is used to initialize the population, the symbolic expressions in the low-frequency features and the high-frequency features are encoded as chromosomes in a genetic algorithm, a group of symbolic expressions are randomly generated as an initial population, and a preset fitness value calculation formula is used to calculate each individual in the initial population to obtain each fitness value. A roulette method is used to screen the initial population based on each fitness value to obtain a screened population, and the screened population is crossovered and mutated to obtain an offspring population.

[0083] The fitness value calculation formula is:

[0084] ;

[0085] Among them, f is the individual fitting formula, is the true value when the depth is i, is the predicted value when the depth is i according to the formula of the individual, is the average of all predicted values.

[0086] This application selects RMSE (Root Mean Squared Error) as the evaluation standard of the fitness function. The formula tree with the best fitness function performance will undergo crossover and mutation to form a new offspring. The process includes: (1) randomly generating initial expressions and establishing an initial population; (2) selecting some expressions based on the fitness function performance for crossover and mutation to create a new offspring population; (3) when the iteration reaches the set number of generations, the program is terminated and the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data with the best fitness function are output; (4) finally, a nonlinear fitting formula for the post-casing electromagnetic logging curve is generated. The crossover and mutation process in the program is as follows: Figure 6 As shown. By means of symbolic regression, the high-frequency flow and the low-frequency flow are fitted to obtain their respective symbolic regression formulas, and the sum of them is used to obtain the final nonlinear fitting result of the post-casing electromagnetic logging curve to the formation resistivity curve.

[0087] Step S24: record the current number of iterations, and repeat the step of randomly generating a set of symbolic expressions as the initial population until the current number of iterations is equal to the preset iteration setting value, so as to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data.

[0088] Step S25: generating a nonlinear fitting formula for the post-casing electromagnetic logging curve based on the nonlinear high-dimensional symbolic regression formula for the low-frequency data and the nonlinear high-dimensional symbolic regression formula for the high-frequency data.

[0089] Step S26: obtaining the casing transient electromagnetic curve of the oil well to be tested, and determining the mapping relationship between the casing transient electromagnetic curve and the formation resistivity of the oil well to be tested using the nonlinear fitting formula of the casing electromagnetic logging curve to obtain the formation resistivity.

[0090] The following difficulties exist in the process of fitting formation resistivity of post-casing transient electromagnetic logging curves: a. Data complexity: TEM (Transient Electromagnetic Method) data usually contains rich formation information, which is nonlinear in time and space and is affected by many factors, such as formation electrical properties, rock structure, borehole environment, etc. Nonlinear relationships mean that traditional linear fitting methods cannot accurately capture the complex features in the data and require more advanced nonlinear models to describe them. b. Multi-parameter coupling: The shape of the logging curve is affected by multiple parameters, and there may be coupling relationships between these parameters, that is, the change of one parameter may affect multiple logging responses at the same time. Decoupling these parameters and establishing an accurate relationship between them and the logging curve is a difficulty in the study. c. Noise and outliers: Logging data often contains noise and outliers, which may have a serious impact on nonlinear fitting. The presence of noise and outliers requires the model to have good robustness, which is a challenge for nonlinear fitting methods.

[0091] In view of the above difficulties, the present invention realizes the separation of high-frequency flow and low-frequency flow through wavelet transform, and then performs fitting through symbolic regression method, regresses the nonlinear fitting formula results from the high-frequency flow and the low-frequency time series flow respectively, and finally adds the wavelet transform results to obtain the characteristics of the original value, and adds the regression formulas of the high-frequency flow and the low-frequency flow to obtain the final nonlinear fitting formula of the post-casing electromagnetic logging curve. The process is as follows: Figure 7 As shown:

[0092] Step 1: Data preprocessing: time series conversion and data preprocessing are performed on the post-casing electromagnetic logging curves one by one, including normalization, feature screening and data cleaning, so that the data is regularized to eliminate the dimension differences between features and reduce the scale differences between features, and the preprocessed time series data is obtained;

[0093] Step 2: transform the time domain data into frequency domain data through wavelet transform, and perform multi-scale refinement and splitting to obtain low-frequency features that meet the preset low-frequency conditions and high-frequency features that meet the preset high-frequency conditions;

[0094] Step 3: Perform symbolic regression on both high-frequency and low-frequency features: First, initialize the population, encode the symbolic expression as a chromosome in the genetic algorithm, randomly generate a set of symbolic expressions as the initial population, define MSE (Mean Squared Error) to evaluate the performance of each expression, and the symbolic preset operators included are as follows: "+", "-", "*", " / ", "cos", "exp", "sin", "1 / x", and then calculate the fitness value for each individual (symbolic expression) in the population. The higher the fitness value of an individual, the stronger its corresponding expression prediction ability is, and the larger the fitness value is, the worse the individual's fitting effect is, and the higher the fitness is. Then use roulette wheel selection to select excellent individuals from the current population for reproduction (individuals that replicate themselves), eliminate the worst 50% of individuals, randomly select crossover points from the parent individuals of the selected initial population, exchange the parent's gene fragments at the crossover point, generate new offspring individuals in the selected population, and randomly change the genes of the offspring individuals to increase the diversity of the population. The mutation probability (mutation rate) determines the possibility of mutation of the individual's genes. Repeat the above steps until the current number of iterations is equal to the preset iteration setting value, and obtain the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data.

[0095] Step 4: Combining the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data, a nonlinear fitting formula for the post-casing electromagnetic logging curve is generated.

[0096] The key points of this application are: (1) Multi-scale feature fitting: The post-casing transient electromagnetic logging (TEM) data is separated into high-frequency stream and low-frequency stream by wavelet transform. The high-frequency characteristic stream and low-frequency characteristic stream output by wavelet transform enable the symbolic regression method to effectively process the nonlinear characteristics of the data in different frequency domains, making the subsequent fitting more accurate. This separation helps to capture the characteristics of different time scales and spatial scales in the data. (2) Introducing the symbolic regression method: The symbolic regression method is used to perform nonlinear fitting on the separated high-frequency stream and low-frequency stream respectively. Symbolic regression can automatically discover and construct mathematical expressions suitable for the data without presetting the specific function form. This method helps to discover the coupling relationship between parameters and establish an accurate relationship between them and the logging curve. (3) Construction of robust fitting formula: By regressing the nonlinear fitting formula from the high-frequency stream and the low-frequency stream, and using the characteristics of wavelet transform to add the two results to restore the original data, this method improves the robustness of the model to noise and outliers, because the noise and outliers are effectively separated and processed during the transformation process. The final fitting formula not only takes into account the nonlinear characteristics of the data, but also reduces the influence of noise and outliers, thereby improving the accuracy and reliability of the fitting.

[0097] This application adopts a symbolic regression algorithm to overcome the limitations of traditional post-casing electromagnetic logging curve fitting. By introducing the symbolic regression algorithm to the fitting of post-casing electromagnetic logging curves, the fitting of the formula retaining the interpretability is achieved and the nonlinear fitting process is completed. The nonlinear fitting formula of the high-dimensional mapping relationship can be provided as a result, which has stronger generalization ability and interpretability. This method provides geological engineers with a more accurate and comprehensive post-casing electromagnetic logging curve fitting tool.

[0098] In this embodiment, a post-casing electromagnetic logging curve of an oil well is obtained, and the post-casing electromagnetic logging curve is converted into a time series and data preprocessed to obtain preprocessed time series data; the time domain data is converted into frequency domain data by wavelet transform, and multi-scale refinement and splitting are performed to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions; the low-frequency features and the high-frequency features are fitted by a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and a nonlinear fitting formula for the post-casing electromagnetic logging curve is generated based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data; the post-casing transient electromagnetic curve of the oil well to be tested is obtained, and the mapping relationship between the post-casing transient electromagnetic curve and the formation resistivity of the oil well to be tested is determined by using the nonlinear fitting formula for the post-casing electromagnetic logging curve to obtain the formation resistivity. The present application adopts wavelet transform to split the multi-scale time domain data into low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions, and uses symbolic regression method for fitting. The use of wavelet transform can enable the symbolic regression method to effectively process the nonlinear characteristics of data in different frequency domains, so that the fitting is more accurate. The symbolic regression method is introduced for fitting, so that it is possible to automatically discover and construct mathematical expressions suitable for data without presetting a specific function form, so as to find a more accurate and concise relationship in a complex model space, avoid the uncertainty of manually selecting model parameters, obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, improve the ability to handle nonlinear problems and the accuracy and reliability of high-dimensional relationship mapping and nonlinear fitting, generate a nonlinear fitting formula for post-casing electromagnetic logging curves, and use the nonlinear fitting formula for post-casing electromagnetic logging curves to determine the mapping relationship between post-casing transient electromagnetic curves and formation resistivity, and obtain formation resistivity, so it has stronger generalization ability and interpretability, and can accurately characterize the relationship between post-casing transient electromagnetic logging curves and formation resistivity.

[0099] See also Figure 8 As shown, the embodiment of the present invention discloses a formation resistivity determination device of a post-casing transient electromagnetic logging curve, which may specifically include:

[0100] The data processing module 11 is used to obtain the casing electromagnetic logging curve of the oil well, and perform time series conversion and data preprocessing on the casing electromagnetic logging curve to obtain preprocessed time series data;

[0101] A multi-scale refinement and splitting module 12 is used to transform the time domain data into frequency domain data by wavelet transform, and perform multi-scale refinement and splitting to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions;

[0102] A fitting formula generation module 13 is used to fit the low-frequency feature and the high-frequency feature using a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and to generate a nonlinear fitting formula for a post-casing electromagnetic logging curve based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data;

[0103] The formation resistivity determination module 14 is used to obtain the casing transient electromagnetic curve of the oil well to be tested, and use the nonlinear fitting formula of the casing electromagnetic logging curve to determine the mapping relationship between the casing transient electromagnetic curve and the formation resistivity of the oil well to be tested, so as to obtain the formation resistivity.

[0104] In this embodiment, a post-casing electromagnetic logging curve of an oil well is obtained, and the post-casing electromagnetic logging curve is converted into a time series and data preprocessed to obtain preprocessed time series data; the time domain data is converted into frequency domain data by wavelet transform, and multi-scale refinement and splitting are performed to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions; the low-frequency features and the high-frequency features are fitted by a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and a nonlinear fitting formula for the post-casing electromagnetic logging curve is generated based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data; the post-casing transient electromagnetic curve of the oil well to be tested is obtained, and the mapping relationship between the post-casing transient electromagnetic curve and the formation resistivity of the oil well to be tested is determined by using the nonlinear fitting formula for the post-casing electromagnetic logging curve to obtain the formation resistivity. The present application adopts wavelet transform to split the multi-scale time domain data into low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions, and uses symbolic regression method for fitting. The use of wavelet transform can enable the symbolic regression method to effectively process the nonlinear characteristics of data in different frequency domains, so that the fitting is more accurate. The symbolic regression method is introduced for fitting, so that it is possible to automatically discover and construct mathematical expressions suitable for data without presetting a specific function form, so as to find a more accurate and concise relationship in a complex model space, avoid the uncertainty of manually selecting model parameters, obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, improve the ability to handle nonlinear problems and the accuracy and reliability of high-dimensional relationship mapping and nonlinear fitting, generate a nonlinear fitting formula for post-casing electromagnetic logging curves, and use the nonlinear fitting formula for post-casing electromagnetic logging curves to determine the mapping relationship between post-casing transient electromagnetic curves and formation resistivity, and obtain formation resistivity, so it has stronger generalization ability and interpretability, and can accurately characterize the relationship between post-casing transient electromagnetic logging curves and formation resistivity.

[0105] In some specific embodiments, the data processing module 11 may specifically include:

[0106] A time series conversion module is used to perform time series conversion on the data in the post-casing electromagnetic logging curve to obtain converted time series data;

[0107] The preprocessing module is used to perform normalization, feature screening and data cleaning on the converted time series data to obtain the preprocessed time series data.

[0108] In some specific embodiments, the multi-scale refinement and splitting module 12 may specifically include:

[0109] The multi-scale refinement splitting module is used to transform the time domain data into frequency domain data by wavelet transform and using a preset data transformation formula, and perform multi-scale refinement splitting; the preset data transformation formula is:

[0110] ;

[0111] ;

[0112] in, is the result of continuous wavelet transform, a is the scaling parameter, s(t) is the input data of a time series with a variable of time t, for The complex conjugate function of , b is the location parameter, is the wavelet function, is a constant of the center frequency, and i is an imaginary number that determines the information in the time domain or space domain in the transformation result.

[0113] In some specific embodiments, the fitting formula generating module 13 may specifically include:

[0114] A crossover and mutation module, used for initializing the population of the low-frequency features and the high-frequency features by using a symbolic regression method, randomly generating a group of symbolic expressions as an initial population, screening the initial population to obtain a screened population, and performing crossover and mutation on the screened population to obtain a progeny population;

[0115] The symbolic regression formula generation module is used to record the current number of iterations and repeat the step of randomly generating a set of symbolic expressions as the initial population until the current number of iterations is equal to a preset iteration setting value, so as to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data.

[0116] In some specific embodiments, the fitting formula generating module 13 may specifically include:

[0117] An initial population generation module is used to initialize the population using a symbolic regression method, encode the symbolic expressions in the low-frequency features and the high-frequency features as chromosomes in a genetic algorithm, and randomly generate a group of symbolic expressions as an initial population;

[0118] The screening module is used to calculate each individual in the initial population using a preset fitness value calculation formula to obtain each fitness value, and screen the initial population based on each fitness value in a roulette manner to obtain a screened population.

[0119] In some specific embodiments, the fitness value calculation formula is:

[0120] ;

[0121] Among them, f is the individual fitting formula, is the true value when the depth is i, is the predicted value when the depth is i according to the formula of the individual, is the average of all predicted values.

[0122] In some specific embodiments, the fitting formula generating module 13 may specifically include:

[0123] A nonlinear fitting formula generation module is used to combine a preset first weight coefficient and a second weight coefficient with a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, respectively, and add the combined nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data to obtain a nonlinear fitting formula for a post-casing electromagnetic logging curve.

[0124] Fig. 9 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the method for determining the formation resistivity of the post-casing transient electromagnetic logging curve performed by the electronic device disclosed in any of the aforementioned embodiments.

[0125] In this embodiment, the power supply 23 is used to provide working voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0126] In addition, the memory 22, as a carrier for storing resources, can be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon include an operating system 221, a computer program 222 and data 223, etc. The storage method can be temporary storage or permanent storage.

[0127] The operating system 221 is used to manage and control the hardware devices and computer program 222 on the electronic device 20, so as to realize the operation and processing of the data 223 in the memory 22 by the processor 21, which can be Windows, Unix, Linux, etc. In addition to including a computer program that can be used to complete the formation resistivity determination method of the post-casing transient electromagnetic logging curve performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can further include a computer program that can be used to complete other specific tasks. In addition to including data transmitted from an external device received by the formation resistivity determination device of the post-casing transient electromagnetic logging curve, the data 223 can also include data collected by its own input and output interface 25, etc.

[0128] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0129] Furthermore, an embodiment of the present application also discloses a computer-readable storage medium, in which a computer program is stored. When the computer program is loaded and executed by a processor, the steps of the method for determining the formation resistivity of the post-casing transient electromagnetic logging curve disclosed in any of the aforementioned embodiments are implemented.

[0130] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0131] The above is a detailed introduction to the method, device, equipment and storage medium for determining the formation resistivity of the post-casing transient electromagnetic logging curve provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for determining formation resistivity from a post-casing transient electromagnetic logging curve, characterized in that: include: Acquire the post-casing electromagnetic logging curve of the oil well, and perform time series conversion and data preprocessing on the post-casing electromagnetic logging curve to obtain preprocessed time series data; The time domain data is converted into frequency domain data by wavelet transform, and multi-scale refinement and splitting is performed to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions; The low-frequency feature and the high-frequency feature are fitted by using a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and a nonlinear fitting formula for a post-casing electromagnetic logging curve is generated based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data; The casing transient electromagnetic curve of the oil well to be tested is obtained, and the mapping relationship between the casing transient electromagnetic curve and the formation resistivity of the oil well to be tested is determined using the nonlinear fitting formula of the casing electromagnetic logging curve to obtain the formation resistivity.

2. The method for determining formation resistivity from a post-casing transient electromagnetic logging curve according to claim 1, characterized in that: The time series conversion and data preprocessing of the post-casing electromagnetic logging curve to obtain preprocessed time series data includes: Performing time series conversion on the data in the post-casing electromagnetic logging curve to obtain converted time series data; The converted time series data is normalized, feature screened, and cleaned to obtain the preprocessed time series data.

3. The method for determining formation resistivity from a post-casing transient electromagnetic logging curve according to claim 1, characterized in that: The method of converting the time domain data into frequency domain data by wavelet transform and performing multi-scale refinement and splitting includes: The time domain data is transformed into frequency domain data by wavelet transform and using a preset data transformation formula, and multi-scale refinement and splitting is performed; the preset data transformation formula is: ； ； in, is the result of continuous wavelet transform, a is the scaling parameter, s(t) is the input data of a time series with a variable of time t, for The complex conjugate function of , b is the location parameter, is the wavelet function, is a constant of the center frequency, and i is an imaginary number that determines the information in the time domain or space domain in the transformation result.

4. The method for determining formation resistivity from a post-casing transient electromagnetic logging curve according to claim 1, characterized in that: The low-frequency features and the high-frequency features are fitted by a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, including: Initializing the population of the low-frequency features and the high-frequency features by using a symbolic regression method, randomly generating a group of symbolic expressions as an initial population, screening the initial population to obtain a screened population, and performing crossover and mutation on the screened population to obtain a progeny population; Record the current number of iterations, and repeat the step of randomly generating a set of symbolic expressions as the initial population until the current number of iterations is equal to a preset iteration setting value, so as to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data.

5. The method for determining formation resistivity from a post-casing transient electromagnetic logging curve according to claim 4, characterized in that: The method of using the symbolic regression method to initialize the population of the low-frequency features and the high-frequency features, randomly generating a group of symbolic expressions as the initial population, and screening the initial population to obtain a screened population includes: Using a symbolic regression method to initialize the population, encoding the symbolic expressions in the low-frequency features and the high-frequency features as chromosomes in a genetic algorithm, and randomly generating a group of symbolic expressions as an initial population; The preset fitness value calculation formula is used to calculate each individual in the initial population to obtain each fitness value, and the initial population is screened based on each fitness value in a roulette manner to obtain a screened population.

6. The method for determining formation resistivity from a post-casing transient electromagnetic logging curve according to claim 5, characterized in that: The fitness value calculation formula is: ； Among them, f is the individual fitting formula, is the true value when the depth is i, is the predicted value when the depth is i according to the formula of the individual, is the average of all predicted values.

7. The method for determining formation resistivity from a post-casing transient electromagnetic logging curve according to any one of claims 1 to 6, characterized in that: The nonlinear fitting formula for the post-casing electromagnetic logging curve is generated based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data, including: The preset first weight coefficient and the second weight coefficient are respectively combined with the nonlinear high-dimensional symbol regression formula for low-frequency data and the nonlinear high-dimensional symbol regression formula for high-frequency data, and the combined nonlinear high-dimensional symbol regression formula for low-frequency data and the nonlinear high-dimensional symbol regression formula for high-frequency data are added to obtain a nonlinear fitting formula for the post-casing electromagnetic logging curve.

8. A device for determining formation resistivity of a post-casing transient electromagnetic logging curve, characterized in that: include: A data processing module is used to obtain the casing electromagnetic logging curve of the oil well, and perform time series conversion and data preprocessing on the casing electromagnetic logging curve to obtain preprocessed time series data; A multi-scale refinement and splitting module is used to transform the time domain data into frequency domain data by wavelet transform, and perform multi-scale refinement and splitting to obtain low-frequency features that meet preset low-frequency conditions and high-frequency features that meet preset high-frequency conditions; A fitting formula generation module is used to fit the low-frequency feature and the high-frequency feature using a symbolic regression method to obtain a nonlinear high-dimensional symbolic regression formula for low-frequency data and a nonlinear high-dimensional symbolic regression formula for high-frequency data, and to generate a nonlinear fitting formula for a post-casing electromagnetic logging curve based on the nonlinear high-dimensional symbolic regression formula for low-frequency data and the nonlinear high-dimensional symbolic regression formula for high-frequency data; The formation resistivity determination module is used to obtain the post-casing transient electromagnetic curve of the oil well to be tested, and use the nonlinear fitting formula of the post-casing electromagnetic logging curve to determine the mapping relationship between the post-casing transient electromagnetic curve and the formation resistivity of the oil well to be tested, so as to obtain the formation resistivity.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor is used to execute the computer program to implement the method for determining formation resistivity of a post-casing transient electromagnetic logging curve as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store computer programs; wherein, when the computer program is executed by a processor, the method for determining formation resistivity of a post-casing transient electromagnetic logging curve as described in any one of claims 1 to 7 is implemented.

Images