CN117538937A - Feature extraction method of underground transient electromagnetic signals - Google Patents
Feature extraction method of underground transient electromagnetic signals Download PDFInfo
- Publication number
- CN117538937A CN117538937A CN202311400966.XA CN202311400966A CN117538937A CN 117538937 A CN117538937 A CN 117538937A CN 202311400966 A CN202311400966 A CN 202311400966A CN 117538937 A CN117538937 A CN 117538937A
- Authority
- CN
- China
- Prior art keywords
- transient electromagnetic
- preset
- feature extraction
- characteristic
- well
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000001052 transient effect Effects 0.000 title claims abstract description 112
- 238000000605 extraction Methods 0.000 title claims abstract description 90
- 230000004044 response Effects 0.000 claims abstract description 80
- 230000009545 invasion Effects 0.000 claims abstract description 48
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims abstract description 48
- 238000005259 measurement Methods 0.000 claims abstract description 19
- 238000000034 method Methods 0.000 claims description 46
- 238000004422 calculation algorithm Methods 0.000 claims description 19
- 238000007781 pre-processing Methods 0.000 claims description 17
- 238000012545 processing Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 9
- 238000010219 correlation analysis Methods 0.000 claims description 8
- 238000013499 data model Methods 0.000 claims description 8
- 238000007637 random forest analysis Methods 0.000 claims description 8
- 238000012216 screening Methods 0.000 claims description 7
- 230000010354 integration Effects 0.000 claims description 6
- 238000011156 evaluation Methods 0.000 abstract description 6
- 230000015572 biosynthetic process Effects 0.000 abstract description 4
- 238000004364 calculation method Methods 0.000 description 11
- 230000002159 abnormal effect Effects 0.000 description 8
- 238000001914 filtration Methods 0.000 description 7
- 230000008569 process Effects 0.000 description 7
- 230000003044 adaptive effect Effects 0.000 description 4
- 238000004445 quantitative analysis Methods 0.000 description 4
- 230000009467 reduction Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 3
- 238000009499 grossing Methods 0.000 description 3
- 238000001514 detection method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000010220 Pearson correlation analysis Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 239000004568 cement Substances 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000005674 electromagnetic induction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/10—Pre-processing; Data cleansing
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/08—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with magnetic or electric fields produced or modified by objects or geological structures or by detecting devices
- G01V3/081—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with magnetic or electric fields produced or modified by objects or geological structures or by detecting devices the magnetic field is produced by the objects or geological structures
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/08—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with magnetic or electric fields produced or modified by objects or geological structures or by detecting devices
- G01V3/083—Controlled source electromagnetic [CSEM] surveying
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/18—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation specially adapted for well-logging
- G01V3/26—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation specially adapted for well-logging operating with magnetic or electric fields produced or modified either by the surrounding earth formation or by the detecting device
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/38—Processing data, e.g. for analysis, for interpretation, for correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Remote Sensing (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Electromagnetism (AREA)
- Artificial Intelligence (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention provides a feature extraction method of an underground transient electromagnetic signal, which comprises the following steps: acquiring a preset feature extraction model; obtaining actual measurement transient electromagnetic response signals of different depths in a well to be measured, and obtaining target characteristic parameters of the corresponding depths according to the actual measurement transient electromagnetic response signals, wherein the target characteristic parameters are characteristic parameters with correlation degree larger than preset correlation degree with water invasion degree of different depths; substituting the target characteristic parameters into a preset characteristic extraction model to obtain the topographic characteristics of the well to be measured corresponding to the depth of the well. The feature extraction method of the underground transient electromagnetic signal provided by the invention enables the extracted features to be comprehensive, and further can effectively identify formation related information at different well depths, so that the evaluation of residual oil gas in the well is accurate.
Description
Technical Field
The invention relates to the technical field of oil and gas field logging, in particular to a feature extraction method of an underground transient electromagnetic signal.
Background
In the development process of oil and gas fields, the condition and distribution of residual oil and gas in an underground oil and gas reservoir are the basis and key of targets and important development decisions pursued by oil field development research institute. To determine the condition and distribution of the residual oil gas, the monitoring and evaluation of the residual oil gas need to be accurately carried out in time. In the prior art, a transient electromagnetic induction logging method is adopted to monitor residual oil gas, and the transient electromagnetic logging method is used for measuring secondary induced electromotive force excited in a stratum, but is easily influenced by factors such as casing electrical parameters, cement rings, invasion zones and the like in the logging process, and a measured transient electromagnetic response signal passing through the casing deviates from actual stratum information. To accurately implement formation resistivity calculations, efficient feature extraction must be performed on the recorded transient electromagnetic signals.
At present, the characteristic extraction method of the transient electromagnetic signals is single, the extracted characteristics are not comprehensive, and therefore the obtained stratum related information and residual oil gas evaluation are not very accurate.
Disclosure of Invention
In view of the above problems, the present invention provides a feature extraction method for a downhole transient electromagnetic signal, which overcomes the above problems or at least partially solves the above problems, and can solve the problems that the existing method for extracting transient electromagnetic features is too single, the extracted features are not comprehensive, and the obtained formation related information and the residual oil gas evaluation are inaccurate, so as to achieve the technical effect of better distinguishing the residual oil gas features of the formation.
Specifically, the invention provides a feature extraction method of a downhole transient electromagnetic signal, which is characterized by comprising the following steps:
acquiring a preset feature extraction model;
obtaining actual measurement transient electromagnetic response signals of different depths in a well to be measured, and obtaining target characteristic parameters of the corresponding depths according to the actual measurement transient electromagnetic response signals, wherein the target characteristic parameters are characteristic parameters with correlation degree larger than preset correlation degree with water invasion degree at different depths;
substituting the target characteristic parameters into the preset characteristic extraction model to obtain the topographic characteristics of the well to be detected corresponding to the depth of the well.
Optionally, the method for obtaining the preset feature extraction model includes:
obtaining standard transient electromagnetic response signals of different depths in a standard well;
calculating characteristic parameters of multiple dimensions of the standard transient electromagnetic response signal;
acquiring the correlation degree of each characteristic parameter and the water invasion degree at different well depths, and screening out the characteristic parameter with the correlation degree of the water invasion degree larger than the preset correlation degree as a target characteristic parameter;
and establishing the preset feature extraction model according to the target feature parameters.
Optionally, before calculating the characteristic parameters of the plurality of dimensions of the standard transient electromagnetic response signal, the method for obtaining the preset characteristic extraction model further includes:
and carrying out data preprocessing on the standard transient electromagnetic response signal.
Optionally, the data preprocessing of the standard transient electromagnetic response signal includes: data deduplication, missing value processing, outlier processing, and/or data denoising.
Optionally, the plurality of dimensions includes a statistical dimension, a time-frequency domain dimension, a shape metric dimension, and/or a hysteresis dimension.
Optionally, the feature values of the statistical dimension include a maximum value, a minimum value, a mean value, a variance, a standard deviation, a peak value, a kurtosis, and/or a skewness;
the characteristic values of the time-frequency domain dimension comprise center-of-gravity frequency, average frequency, frequency root mean square and/or frequency variance;
the characteristic values of the shape measurement dimension comprise shannon entropy, a Hi Gu Qi fractal dimension, a Katz fractal dimension and/or a generalized Hersteter index;
the characteristic value of the hysteresis dimension includes calculating an autocorrelation characteristic value of the time series.
Optionally, the obtaining the correlation between each of the characteristic parameters and the water invasion degrees at different depths of the well includes:
and adopting a preset correlation analysis algorithm to calculate the correlation degree between the characteristic parameters and the water invasion degrees at different well depths.
Optionally, the establishing the preset feature extraction model according to the target feature parameter includes:
taking the target characteristic parameters as input of the preset characteristic extraction model, taking the corresponding water invasion degrees of different depths of the well as output of the preset characteristic extraction model, and training the preset characteristic model;
and distributing weights to the target characteristic parameters so as to establish the preset characteristic extraction model.
Optionally, the assigning weights to the target feature parameters includes:
and adopting a self-adaptive enhancement integration algorithm to allocate weights to the target characteristic parameters.
Optionally, the establishing the preset feature extraction model according to the target feature parameter further includes:
obtaining a nonlinear relation data model between the standard transient electromagnetic response signal and the water invasion degrees at different well depths by adopting a regression random forest algorithm and a classification regression tree fitting mode;
and taking the nonlinear relation data model as the preset feature extraction model.
In the feature extraction method of the underground transient electromagnetic signal, a preset feature extraction model is firstly obtained, the preset feature extraction model can be an established standard feature extraction model, the standard feature extraction model is obtained through multiple data calculation, and corresponding detailed feature information can be effectively obtained through substituting parameters. Then, a transient electromagnetic logging instrument is put into the well to be measured, actual measurement transient electromagnetic response signals of different depths in the well to be measured are measured, the obtained actual measurement transient electromagnetic response signals are further optimized, characteristic parameters with correlation degrees larger than preset correlation degrees with water invasion degrees of different depths in the well to be measured are screened out, and the characteristic parameters are used as target characteristic parameters. And finally, substituting the target characteristic parameters into a preset characteristic extraction model to further obtain the topographic characteristics of the stratum with the water invasion degrees at different well depths corresponding to the actually measured transient electromagnetic signals. According to the feature extraction method, the preset feature extraction model is added, and the actual measurement transient electromagnetic response signals at different well depths are subjected to feature parameter screening, so that stratum feature information corresponding to the actual measurement electromagnetic response signals in the well to be detected is extracted more comprehensively, the accuracy is higher, and further, the evaluation of relevant information and residual oil gas of water invasion degrees of different strata is more accurate.
The above, as well as additional objectives, advantages, and features of the present invention will become apparent to those skilled in the art from the following detailed description of a specific embodiment of the present invention when read in conjunction with the accompanying drawings.
Drawings
Some specific embodiments of the invention will be described in detail hereinafter by way of example and not by way of limitation with reference to the accompanying drawings. The same reference numbers will be used throughout the drawings to refer to the same or like parts or portions. It will be appreciated by those skilled in the art that the drawings are not necessarily drawn to scale. In the accompanying drawings:
FIG. 1 is a schematic flow chart of a method of feature extraction of a downhole transient electromagnetic signal according to one embodiment of the invention;
FIG. 2 is a schematic flow chart of a feature extraction method according to another embodiment of the invention;
FIG. 3 is a schematic flow chart of a feature extraction method according to another embodiment of the invention;
FIG. 4 is a schematic flow chart of a feature extraction method according to another embodiment of the invention;
FIG. 5 is a signal profile of a downhole transient electromagnetic signal prior to data preprocessing in accordance with one embodiment of the present invention;
FIG. 6 is a signal profile of a downhole transient electromagnetic signal after data preprocessing in accordance with one embodiment of the present invention.
Detailed Description
A method for extracting characteristics of a downhole transient electromagnetic signal according to an embodiment of the present invention is described below with reference to fig. 1 to 6. In the description of the present embodiment, it should be understood that the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature, i.e. one or more such features. In the description of the present invention, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise. When a feature "comprises or includes" a feature or some of its coverage, this indicates that other features are not excluded and may further include other features, unless expressly stated otherwise.
In the description of the present embodiment, a description referring to the terms "one embodiment," "some embodiments," "illustrative embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Fig. 1 is a schematic flow chart of a feature extraction method of a downhole transient electromagnetic signal according to an embodiment of the invention, as shown in fig. 1, and referring to fig. 2 to 6, the embodiment of the invention provides a feature extraction method of a downhole transient electromagnetic signal, the feature extraction method comprising the steps of:
step S1: acquiring a preset feature extraction model;
step S2: obtaining actual measurement transient electromagnetic response signals of different depths in a well to be measured, and obtaining target characteristic parameters of the corresponding depths according to the actual measurement transient electromagnetic response signals, wherein the target characteristic parameters are characteristic parameters with correlation degree larger than preset correlation degree with water invasion degree of different depths;
step S3: substituting the target characteristic parameters into a preset characteristic extraction model to obtain the topographic characteristics of the well to be measured corresponding to the depth of the well.
In the step S1, a preset feature extraction model is obtained, and the preset feature extraction model may be an established standard feature extraction model, which is obtained by multiple data calculations. Corresponding detailed feature information can be obtained by substituting parameters into the feature extraction model.
In the step S2, a transient electromagnetic logging instrument is placed in the well to be measured, actual measurement transient electromagnetic response signals of different depths in the well to be measured are measured, then the obtained actual measurement transient electromagnetic response signals are further optimized, and characteristic parameters with correlation degree larger than preset correlation degree with water invasion degree of different depths in the well to be measured are screened out, and the characteristic parameters are used as target characteristic parameters.
And in the step S3, substituting the target characteristic parameters into a preset characteristic extraction model to further obtain the topographic characteristics of the stratum with the water invasion degrees at different well depths corresponding to the actually measured transient electromagnetic signals.
According to the feature extraction method provided by the embodiment, the feature extraction model is preset in advance, and the actual measurement transient electromagnetic response signals at different well depths are subjected to feature parameter screening, so that stratum feature information corresponding to the actual measurement electromagnetic response signals in the well to be detected is extracted more comprehensively, the accuracy is higher, stratum related information at different well depths can be effectively identified, and evaluation of residual oil gas in the well is more accurate.
In some embodiments of the present invention, as shown in fig. 2, the method for obtaining the preset feature extraction model in the step S1 includes:
step S101: obtaining standard transient electromagnetic response signals of different depths in a standard well;
step S103: calculating characteristic parameters of multiple dimensions of the standard transient electromagnetic response signal;
step S104: acquiring the correlation degree of each characteristic parameter and the water invasion degree at different well depths, and screening out the characteristic parameters with the correlation degree of the water invasion degree larger than the preset correlation degree as target characteristic parameters;
step S105: and establishing a preset feature extraction model according to the target feature parameters.
In this embodiment, a method for establishing a preset feature extraction model is provided, and first, standard transient electromagnetic response signals of different depths in a standard well are measured by a transient electromagnetic logging instrument which is placed in the standard well. Then, calculating characteristic parameters of multiple dimensions of each standard transient electromagnetic response signal, and screening out characteristic parameters with correlation degree with water invasion degree larger than preset correlation degree as target characteristic parameters. And finally, establishing a preset feature extraction model according to the target feature parameters. The preset correlation may be set as a correlation coefficient of a characteristic parameter having an absolute value of 0.5.
By means of the method, the preset feature extraction model is built, the data dimension of the collected actual measurement transient electromagnetic response signals can be reduced, the quality of data samples is improved, the efficiency of building the preset feature extraction model is higher, and meanwhile the accuracy of feature extraction through the preset feature extraction model is improved.
In some embodiments of the present invention, as shown in fig. 3, before calculating the characteristic parameters of the multiple dimensions of the standard transient electromagnetic response signal, the method for obtaining the preset characteristic extraction model further includes:
s102, data preprocessing is carried out on the standard transient electromagnetic response signals.
In this embodiment, as shown in fig. 5 and 6, the downhole transient electromagnetic response signal with the well depth of 2230m-2270m is measured as an example, and may be subjected to data preprocessing by using a data cleaning method, so as to obtain signal feature diagrams before and after preprocessing the transient electromagnetic response signal. The data samples can be optimized by preprocessing the data of the standard transient electromagnetic response signals, so that the processed samples of the electromagnetic response signals are purer, and the subsequent calculation of the multidimensional characteristic parameters of the electromagnetic response signals is facilitated. It should be noted that h in the present embodiment j Well depth values of 2230m-2270 m.
In some embodiments of the invention, data preprocessing of the standard transient electromagnetic response signal comprises: data deduplication.
In this embodiment, when the data preprocessing is performed on the electromagnetic response signal, the data deduplication can be performed on the standard transient electromagnetic response signal, mainly the data that is repeatedly measured is removed, so that the repeated computation on the standard transient electromagnetic response signal is effectively avoided, the occupation of the storage space is reduced, and the data processing efficiency is improved.
Furthermore, in order to optimize sample data, the missing value processing can be performed on the data samples of the standard transient electromagnetic response signals at the same time, and in the missing value processing process, a direct deleting method, a missing value filling method or a missing value fitting method can be adopted to perfect the data samples and improve the accuracy and reliability of the sample data.
Furthermore, the abnormal value processing can be carried out on the data samples of the standard transient electromagnetic response signals at the same time, and the methods of deleting the abnormal value, replacing the abnormal value, carrying out the box division processing, carrying out characteristic engineering by utilizing the abnormal value, establishing a robust model and the like can be adopted in the abnormal value processing process so as to further improve the accuracy and the reliability of sample data.
Furthermore, data noise reduction can be performed on the data samples of the standard transient electromagnetic response signals at the same time, and in the data noise reduction process, data noise reduction can be performed by adopting methods such as mean value filtering, median filtering, wavelet transformation, connected graph method or support vector machine, and the like, so that the accuracy and the reliability of sample data are further improved. Similarly, a zero-drift baseline normalization and recursive filtering method can also be used for signal noise reduction.
Of course, in order to achieve a better data preprocessing effect, data deduplication, missing value processing, outlier processing and data noise reduction can be performed on the data samples of the standard transient electromagnetic response signals at the same time.
In particular, during the actual data preprocessing operation, to measure the depth h of the well depth j The repetitive transient electromagnetic response signal in the continuous sampling mode is removed for the unique identifier. Sliding window with s=3 size, sliding window interpolation method is adopted for each depth h j Is subjected to missing value filling. Judging an abnormal value by drawing a standard transient electromagnetic response signal scatter diagram, and adopting h j-1 And h j+1 Filling the average value of the signals at the same time of the high-frequency electromagnetic signals, and removing the interference of the high-frequency electromagnetic signals by adopting median smoothing filtering. The sliding window with s=3 refers to a sliding window with the length of 3 transient electromagnetic response signals as a detection target in a single sliding, if the transient electromagnetic response signals are missing, the missing values can be identified more quickly, and an interpolation method is adopted to fill the missing values. The arrangement can improve the efficiency of identifying the missing value in the data preprocessing process, and meanwhile, the missing value can be conveniently filled in the later stage. And, the high-frequency electromagnetic signal is filtered by adopting median smoothing filtering. The median smoothing filtering mainly comprises the steps of identifying the frequencies of 3 transient electromagnetic response signals, identifying high-frequency abnormal values through median calculation, and directly filtering the identified high-frequency abnormal values.
In some embodiments of the invention, the plurality of dimensions includes a statistical dimension, a time-frequency domain dimension, a shape metric dimension, and/or a hysteresis dimension.
In the present embodiment, each depth h j The transient electromagnetic response signal after data preprocessing is defined as the value x= { X 1 ，x 2 ，…，x i ，…，x n }。
Wherein h is j Representative is the well depth value at height j. X is x i Representing the ith transient electromagnetic response signal.
The calculated characteristic parameter values of different dimensions are defined as f= { F 1 ，F 2 ，...，F m }。
Wherein = 1 Representative is x 1 Corresponding values of multidimensional characteristic parameters of the transient electromagnetic response signals. m represents the number of samples of the transient electromagnetic response signal.
By calculating characteristic parameters of multiple dimensions of the standard transient electromagnetic response signal and carrying out characteristic fusion on characteristics of different dimensions, space-time characteristics of the standard transient electromagnetic response signal with more resolution and more comprehensiveness can be obtained, and further target characteristic parameters for establishing a preset characteristic extraction model are more comprehensiveness, and the established model has more reliability and accuracy.
In some embodiments of the invention, the characteristic values of the statistical dimension include a maximum value, a minimum value, a mean value, a variance, a standard deviation, a peak value, a kurtosis, and/or a skewness. The characteristic values of the time-frequency domain dimension include center of gravity frequency, average frequency, frequency root mean square, and/or frequency variance. Characteristic values of the shape metric dimension include shannon entropy, a his Gu Qi fractal dimension, a kaz fractal dimension, and/or a generalized hurst index. The characteristic value of the hysteresis dimension includes calculating an autocorrelation characteristic value of the time series.
In this embodiment, the feature value of the statistical dimension includes a maximum value F 1 Minimum value F 2 Mean F 3 Variance F 4 Standard deviation F 5 Peak value F 6 Kurtosis F 7 And degree of deviation F 8 . Wherein, the calculation formula of each characteristic value is as follows:
F 1 ＝max(X(i))；
F 2 ＝min(X(i))；
F 6 ＝F 1 -F 2 ；
wherein x= { X 1 ，x 2 ，…，x i ，…，x n Each depth h is represented by } j Transient electromagnetic response signals after data preprocessing; f= { F 1 ,F 2 ,…,F m -representing multidimensional feature parameter values of a plurality of different dimensions; n represents the length of the transient electromagnetic response signal X; m represents the number of samples in the transient electromagnetic response signal.
The characteristic values of the time-frequency domain dimension include calculating the center of gravity frequency F 9 Average frequency F 10 Root mean square F of frequency 11 Frequency variance F 12 . Wherein, the calculation formula of each characteristic value is as follows:
k in the above formula represents the number of intervals after the transient electromagnetic response signal X is segmented; fk represents the frequency amplitude of the point corresponding to X; s (k) represents the power spectrum value corresponding to X; m represents the number of samples in the transient electromagnetic response signal.
The feature values of the shape metric dimension include calculating shannon entropy F 13 Fractal dimension F of Hi Gu Qi 14 Kaz fractal dimension F 15 Generalized Hersteter index F 16 . Wherein, the calculation formula of each characteristic value is as follows:
p (x) in the above formula i ) Representing the proportion of the number of variable classes to which a single sample variable belongs to, that is to say obtaining x i Probability of value;
l (k) represents the average value of Lm (k) for each k;
l represents the total length of the signal;
d represents the plane distance of the waveform;
τ represents the time lag over the time window;
q represents an order;
t=v, 2 v, …, kv, T. Wherein T represents the observation period, and v represents the time resolution.
The characteristic value of the hysteresis dimension includes calculating an autocorrelation characteristic value F of the time series 17 The calculation formula is as follows:
wherein μ represents an overall average value of the transient electromagnetic response signal X;
l represents a hysteresis coefficient;
σ represents the standard deviation of the transient electromagnetic response signal X.
Through the calculation of the multi-dimensional characteristic parameter values, characteristic parameter values of different dimensions corresponding to each transient electromagnetic response signal can be obtained, so that characteristic fusion can be conveniently carried out on the characteristic parameters of different dimensions, space-time characteristics of standard transient electromagnetic response signals with more resolution and more comprehensiveness can be obtained, further, target characteristic parameters for establishing a preset characteristic extraction model are more comprehensible, and the established model has more reliability and accuracy.
In some embodiments of the present invention, obtaining the correlation between each characteristic parameter and the water invasion levels at different depths of the well includes using a preset correlation analysis algorithm to calculate the correlation between the characteristic parameter and the water invasion levels at different depths of the well.
In this embodiment, first, the characteristic parameters with a larger influence on the water invasion degree are screened by adopting a preset correlation analysis algorithm in combination with the water invasion data of the standard well. The preset correlation analysis algorithm may employ a pearson correlation analysis algorithm. For each depth h j Transient electromagnetic response signal x= { X 1 ，x 2 ，…，x i ，…，x n Multi-dimensional feature parameter value f= { F 1 ，F 2 ，...，F m Sequence of water intrusion values y= { Y } 1 ,y 2 ,…,y j And forming a j-row (m+1) column two-dimensional matrix with depth points as rows and characteristic parameters and water invasion values as columns. Calculate the characteristic parameters F of each column i Pearson correlation coefficient with the n+1th column Y sequenceWherein->
Meanwhile, the pearson correlation coefficient mentioned above ranges from-1 to 1, and the closer the value is to 1 or the stronger the linear relationship between the two variables is represented by-1, the closer the value is to 0, which means that the linear relationship between the two variables is weaker or has no linear relationship, and the absolute value is greater than or equal to 0.5, which represents a strong linear relationship. In the calculated correlation coefficient set R= { R 1 ,r 2 ,…,r m And characteristic parameters with absolute values larger than or equal to 0.5 are screened out as target characteristic parameters, so that the efficiency of extracting transient electromagnetic response signals of different depths of a well is further improved, and further, characteristic parameters with high correlation with water invasion degrees of different depths of the well can be screened out more quickly. Meanwhile, the preset correlation degree may be set to a value having an absolute value equal to 0.5, and a correlation degree larger than the preset correlation degree may be used as a criterion for screening.
For example, the resulting pearson correlation coefficient for F1 is 0.37, the pearson correlation coefficient for F17 is 0.78, and so on. As shown in the table below,
the perason values in the table above represent pearson correlation coefficients. The method is easy to obtain through the table, all multi-dimensional characteristic parameter values corresponding to absolute values of pearson correlation coefficients which are larger than or equal to 0.5 are used as target characteristic parameters, and the obtained accuracy of the corresponding water invasion degrees at different well depths is highest.
In other embodiments of the present invention, other correlation analysis methods may be used, such as: gray correlation analysis. The basic idea of gray correlation analysis is to determine whether the relationship between different sequences is tight based on the degree of similarity of the sequence curve geometry. The basic idea is to convert the discrete behavior observation value of the system factors into a piecewise continuous broken line by a linear interpolation method, and then construct a model for measuring the association degree according to the geometric characteristics of the broken line. The closer the polyline geometry is, the greater the degree of association between the corresponding sequences and vice versa. In the practical application process, a data sequence reflecting the behavior characteristics of the system can be used as a reference sequence. The data sequence consisting of factors affecting the behavior of the system is used as a comparison series. And comparing the association degree value of the number sequence and the reference number sequence at each moment to serve as the association degree. Can be at each depth h j The transient electromagnetic response signal is used as a comparison sequence, and the water invasion degree value is used as a reference sequence, so that the degree of association of the two is obtained. And then, judging the influence of the characteristic parameters of different transient electromagnetic response signals on the water invasion degrees at different well depths by comparing the correlation degrees.
In some embodiments of the present invention, establishing the preset feature extraction model according to the target feature parameters includes:
taking the target characteristic parameters as input of a preset characteristic extraction model, taking the corresponding water invasion degrees of different depths of the well as output of the preset characteristic extraction model, and training the preset characteristic extraction model;
and (5) distributing weights for the target feature parameters to establish a preset feature extraction model.
In this embodiment, in order to establish a preset feature extraction model, the screened feature parameters are directly taken as target feature parameters, the target feature parameters are taken as input, and meanwhile, the corresponding water invasion degrees of different depths of the well are taken as output, so that sample training is performed on the preset feature extraction model. And then respectively distributing weights for the target characteristic parameters, such as: when the weights are distributed, large weights are distributed for the characteristic parameters with high correlation degree with the water invasion degree at different well depths, so that a more accurate preset characteristic extraction model is obtained.
In some embodiments of the invention, assigning weights to the target feature parameters includes employing an adaptive enhancement integration algorithm to assign weights to the target feature parameters.
In this embodiment, the adaptive boost integration algorithm is an adaptive boost algorithm. The core idea of the adaptive enhancement integration algorithm is to train a series of similar models that depend on each other in series, i.e. the latter model is used to correct the output result of the former model. Thus, after a series of serial models are fitted, an integrated model with more accurate results is finally obtained. The self-adaptive enhancement integration algorithm is used for distributing the weight to the target characteristic parameters, so that the weight distributed to each characteristic parameter is more reasonable, the achieved accuracy is higher, the purpose of automatically distributing the weight to each characteristic parameter is achieved, and the efficiency is higher.
In some embodiments of the present invention, the method further includes:
a regression random forest algorithm is adopted, and a nonlinear relation data model between a standard transient electromagnetic response signal and water invasion degrees at different well depths is obtained through a classification regression tree fitting mode;
and taking the nonlinear relation data model as a preset feature extraction model.
In the embodiment, a quantitative analysis data model between the multidimensional characteristic parameter F and the water invasion degree Y is mainly constructed by adopting a regression random forest based method. When sample training is carried out, the data in the training set is a combined characteristic quantity extracted from 2100 groups of transient electromagnetic response signals and a water invasion degree value under the corresponding well depth, and the training data set E= { (U) 1 ,Y 1 ),(U 2 ,Y 2 ),…,(U 2100 ,Y 2100 ) U, where i ＝{F i1 ,F i2 ,…F im The combined characteristic quantity of transient electromagnetic signals at a certain depth is represented by Y i Representing the water invasion level value corresponding to the well depth. Training through input 2100 sets of samples, a base is establishedAnd a nonlinear relation model between the combination characteristic F of the transient electromagnetic signals of the regression random forest algorithm and the water invasion degree Y. And then testing by using 900 groups of test samples, and correcting the established nonlinear relation model to obtain a final quantitative analysis model. The quantitative analysis model is a preset feature extraction model.
Further, in the embodiment, quantitative analysis and detection are mainly performed by adopting random forest regression, so that a data model between the transient electromagnetic response signal and the water invasion degree is obtained. Due to the nonlinear relation between the transient electromagnetic response signal and the water invasion degree, a random forest method is adopted, a nonlinear relation model between a relatively accurate transient electromagnetic response signal U and the water invasion degree Y can be obtained by training a classifier comprising a plurality of decision trees, the nonlinear relation between the transient electromagnetic response signal U and the water invasion degree Y is fitted more accurately, and the obtained preset feature extraction model is more accurate and has stronger reliability.
In other embodiments of the present invention, the random forest algorithm used in the above embodiments may also use other machine learning algorithms, such as: neural networks, etc.
In some embodiments of the present invention, as shown in fig. 4, the feature extraction method of the downhole transient electromagnetic signal further includes:
s4, verifying the validity of the preset feature extraction model.
In the practical application process, in the embodiment, actually measured transient electromagnetic response signals at different well depths are measured in a certain gas field well to obtain 3000 groups of field experimental data. And substituting the 3000 groups of field experimental data into a preset feature extraction model to obtain a calculation result of the water invasion degree of the corresponding well depth of the well. Through analysis, the average prediction accuracy of the 3000 sample data is 86.45%, and the good practical application standard is achieved. That is, the feature extraction method can calculate the water invasion degree of different well depths of different wells to be measured accurately and quantitatively, and further can effectively identify stratum related information of different well depths so as to well evaluate the information of residual oil gas, and has a good application prospect.
By now it should be appreciated by those skilled in the art that while a number of exemplary embodiments of the invention have been shown and described herein in detail, many other variations or modifications of the invention consistent with the principles of the invention may be directly ascertained or inferred from the present disclosure without departing from the spirit and scope of the invention. Accordingly, the scope of the present invention should be understood and deemed to cover all such other variations or modifications.
Claims (10)
1. A method for extracting characteristics of a downhole transient electromagnetic signal, comprising:
acquiring a preset feature extraction model;
obtaining actual measurement transient electromagnetic response signals of different depths in a well to be measured, and obtaining target characteristic parameters of the corresponding depths according to the actual measurement transient electromagnetic response signals, wherein the target characteristic parameters are characteristic parameters with correlation degree larger than preset correlation degree with water invasion degree at different depths;
substituting the target characteristic parameters into the preset characteristic extraction model to obtain the topographic characteristics of the well to be detected corresponding to the depth of the well.
2. The method for extracting features of a downhole transient electromagnetic signal according to claim 1, wherein the method for obtaining a predetermined feature extraction model comprises:
obtaining standard transient electromagnetic response signals of different depths in a standard well;
calculating characteristic parameters of multiple dimensions of the standard transient electromagnetic response signal;
acquiring the correlation degree of each characteristic parameter and the water invasion degree at different well depths, and screening out the characteristic parameter with the correlation degree of the water invasion degree larger than the preset correlation degree as a target characteristic parameter;
and establishing the preset feature extraction model according to the target feature parameters.
3. The method for feature extraction of a downhole transient electromagnetic signal of claim 2,
before calculating the characteristic parameters of the plurality of dimensions of the standard transient electromagnetic response signal, the method for obtaining the preset characteristic extraction model further comprises the following steps:
and carrying out data preprocessing on the standard transient electromagnetic response signal.
4. A method for feature extraction of a downhole transient electromagnetic signal according to claim 3,
the data preprocessing of the standard transient electromagnetic response signal comprises the following steps: data deduplication, missing value processing, outlier processing, and/or data denoising.
5. The method for feature extraction of a downhole transient electromagnetic signal of claim 2,
the plurality of dimensions includes a statistical dimension, a time-frequency domain dimension, a shape metric dimension, and/or a hysteresis dimension.
6. The method for feature extraction of a downhole transient electromagnetic signal of claim 5,
the characteristic values of the statistical dimension comprise a maximum value, a minimum value, a mean value, a variance, a standard deviation, a peak value, kurtosis and/or skewness;
the characteristic values of the time-frequency domain dimension comprise center-of-gravity frequency, average frequency, frequency root mean square and/or frequency variance;
the characteristic values of the shape measurement dimension comprise shannon entropy, a Hi Gu Qi fractal dimension, a Katz fractal dimension and/or a generalized Hersteter index;
the characteristic value of the hysteresis dimension includes calculating an autocorrelation characteristic value of the time series.
7. The method for feature extraction of a downhole transient electromagnetic signal of claim 2,
the obtaining the correlation between each characteristic parameter and the water invasion degree at different well depths comprises the following steps:
and adopting a preset correlation analysis algorithm to calculate the correlation degree between the characteristic parameters and the water invasion degrees at different well depths.
8. The method for feature extraction of a downhole transient electromagnetic signal of claim 2,
the establishing the preset feature extraction model according to the target feature parameters comprises the following steps:
taking the target characteristic parameters as input of the preset characteristic extraction model, taking the corresponding water invasion degrees of different depths of the well as output of the preset characteristic extraction model, and training the preset characteristic model;
and distributing weights to the target characteristic parameters so as to establish the preset characteristic extraction model.
9. The method for feature extraction of a downhole transient electromagnetic signal of claim 8,
the step of distributing weights for the target characteristic parameters comprises the following steps:
and adopting a self-adaptive enhancement integration algorithm to allocate weights to the target characteristic parameters.
10. The method for extracting features of a downhole transient electromagnetic signal according to claim 2, wherein said establishing the preset feature extraction model according to the target feature parameters further comprises:
obtaining a nonlinear relation data model between the standard transient electromagnetic response signal and the water invasion degrees at different well depths by adopting a regression random forest algorithm and a classification regression tree fitting mode;
and taking the nonlinear relation data model as the preset feature extraction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311400966.XA CN117538937A (en) | 2023-10-26 | 2023-10-26 | Feature extraction method of underground transient electromagnetic signals |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311400966.XA CN117538937A (en) | 2023-10-26 | 2023-10-26 | Feature extraction method of underground transient electromagnetic signals |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117538937A true CN117538937A (en) | 2024-02-09 |
Family
ID=89792774
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311400966.XA Pending CN117538937A (en) | 2023-10-26 | 2023-10-26 | Feature extraction method of underground transient electromagnetic signals |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117538937A (en) |
-
2023
- 2023-10-26 CN CN202311400966.XA patent/CN117538937A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111046341B (en) | Unconventional natural gas fracturing effect evaluation and productivity prediction method based on principal component analysis | |
| US20140114892A1 (en) | Apparatus and methods of analysis of pipe and annulus in a wellbore | |
| CN109613611B (en) | Method and system for determining input seismic waves for structural seismic time-course analysis | |
| CN110555247A (en) | structure damage early warning method based on multipoint sensor data and BilSTM | |
| CN110398647B (en) | Transformer state monitoring method | |
| CN103336305B (en) | A kind of method dividing Sandstone Gas Reservoir high water cut based on gray theory | |
| CN112836393B (en) | Method for analyzing reservoir heterogeneity based on multi-scale entropy | |
| CN114880940A (en) | Compaction quality detection method and device based on multi-domain analysis and artificial neural network | |
| CN111626377B (en) | Lithology recognition method, device, equipment and storage medium | |
| CN116699724B (en) | Time domain induced polarization data quality evaluation method, system and system | |
| CN117538937A (en) | Feature extraction method of underground transient electromagnetic signals | |
| CN116188266A (en) | Spatial interpolation method for geological investigation based on fuzzy theory and ensemble learning | |
| CN110552693A (en) | layer interface identification method of induction logging curve based on deep neural network | |
| CN112612935B (en) | Complete logging data acquisition method based on self-reasoning model | |
| CN112859192A (en) | Rock core saturation prediction model construction method and rock core saturation prediction method | |
| CN116432054A (en) | Partial discharge severity identification method based on cluster analysis | |
| CN113139681B (en) | Neural network rock burst prediction method based on time sequence data | |
| CN112069621B (en) | Method for predicting residual service life of rolling bearing based on linear reliability index | |
| CN117972365B (en) | Tunnel bottom pile pressure bearing monitoring system and monitoring method | |
| CN117970479B (en) | Surrounding rock grading method and grading system based on earthquake body waves and machine learning | |
| CN112668814B (en) | Cementing grade determining method and device | |
| CN115266914B (en) | Pile sinking quality monitoring system and method based on acoustic signal processing | |
| CN116738149B (en) | Intelligent recognition method for reinforcing stratum by vibroflotation gravel pile | |
| CN111929744B (en) | Kendall's coefficient-based multivariate information reservoir classification method | |
| CN110134918B (en) | Automatic core homing method and system based on sliding window method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |