CN117647846A - Electromagnetic signal noise reduction method - Google Patents
Electromagnetic signal noise reduction method Download PDFInfo
- Publication number
- CN117647846A CN117647846A CN202210988015.8A CN202210988015A CN117647846A CN 117647846 A CN117647846 A CN 117647846A CN 202210988015 A CN202210988015 A CN 202210988015A CN 117647846 A CN117647846 A CN 117647846A
- Authority
- CN
- China
- Prior art keywords
- noise
- signal
- electromagnetic signal
- noise reduction
- component
- 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
- 230000009467 reduction Effects 0.000 title claims abstract description 87
- 238000000034 method Methods 0.000 title claims abstract description 58
- 238000012545 processing Methods 0.000 claims abstract description 12
- 238000000354 decomposition reaction Methods 0.000 claims description 14
- 230000008569 process Effects 0.000 claims description 13
- 238000012549 training Methods 0.000 claims description 9
- 238000009826 distribution Methods 0.000 claims description 4
- 238000011478 gradient descent method Methods 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 22
- 230000001052 transient effect Effects 0.000 abstract description 4
- 238000005259 measurement Methods 0.000 description 18
- 230000015654 memory Effects 0.000 description 14
- 238000012360 testing method Methods 0.000 description 10
- 230000005291 magnetic effect Effects 0.000 description 9
- 238000004088 simulation Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 7
- 238000004458 analytical method Methods 0.000 description 6
- 230000002950 deficient Effects 0.000 description 6
- 230000000694 effects Effects 0.000 description 6
- 239000010410 layer Substances 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 239000002131 composite material Substances 0.000 description 3
- 230000007547 defect Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000004044 response Effects 0.000 description 3
- 239000000523 sample Substances 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 239000002356 single layer Substances 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 108010076504 Protein Sorting Signals Proteins 0.000 description 1
- 229910000831 Steel Inorganic materials 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000007797 corrosion Effects 0.000 description 1
- 238000005260 corrosion Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000005674 electromagnetic induction Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000005284 excitation Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000005294 ferromagnetic effect Effects 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 239000010959 steel Substances 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
Classifications
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Testing Electric Properties And Detecting Electric Faults (AREA)
Abstract
The invention belongs to the technical field of transient electromagnetic signal processing, and particularly relates to an electromagnetic signal noise reduction method. Firstly, acquiring an original electromagnetic signal which changes along with time, and inputting the original electromagnetic signal into a trained noise model to obtain a noise signal comprising oil pipe eccentricity, mechanical high-frequency jitter and pipe column temperature drift; then, subtracting the noise signal from the original electromagnetic signal to obtain an electromagnetic signal after noise reduction. The method improves the signal-to-noise ratio of the electromagnetic signal, ensures the precision of the detection signal under the complex working condition, can improve the interpretation accuracy by utilizing the signal to interpret the field data, and is suitable for noise reduction of the oil casing detection signal.
Description
Technical Field
The invention belongs to the technical field of transient electromagnetic signal processing, and particularly relates to an electromagnetic signal noise reduction method.
Background
With the continuous deep development of oil and gas fields, the number of casing damage wells is increased year by year, the damage degree is increased year by year, the trend of regional development is pointed, the monitoring means are utilized for identifying, preventing and treating casing damage, and the method has important significance for the safe production of oil (gas) wells.
The electromagnetic flaw detection well-logging technology can detect damage conditions of the oil pipe and the sleeve in the oil pipe, judge deformation, corrosion, damage and perforation, accurately indicate the structure of a downhole tubular column and the position of a tool, and detect ferromagnetic substances outside the sleeve. Because the instrument external diameter is small (43 mm), the oil pipe can be used for measuring in the production well, the operation cost and time for lifting and lowering the oil pipe are saved, and the method has unique advantages in the damage detection of the oil (gas) well.
The physical basis of the electromagnetic flaw detection well-logging technology is Faraday electromagnetic induction law, current is supplied to a transmitting coil, an induced electromotive force which changes along with time is generated by a receiving coil, when the thickness of an oil casing is changed or a defect exists, the induced electromotive force is changed, and under the single-layer and double-layer pipe column structure, cracks and holes of the pipe column can be judged through analysis and calculation, so that the average residual wall thickness of the pipe column is obtained.
The pulsed eddy current detection scene of the oil (gas) well tubular column structure is complex, the multilayer tubular column structure is more complex than the single-layer tubular column structure, magnetic force lines penetrate through the oil pipe and enter the sleeve, circular currents are respectively generated in the oil and the sleeve, a secondary magnetic field is generated for overcoming the attenuation of the original magnetic field intensity, and induced electromotive force is generated in the receiving coil. When the oil pipe receives the pulse eddy excitation signal, secondary field mutual inductance is generated, so that the space magnetic field intensity near the sleeve is increased, the voltage amplitude of the detection coil is integrally improved, the attenuation speed is reduced, and the response noise interference of non-tubular column information is enhanced. By analyzing interference of field test data, the noise sources of the detection signals are summarized to be three types: firstly, oil pipe eccentricity can lead to certain deviation between an oil casing detection signal and a single casing; secondly, mechanical high-frequency shaking, steel wire or cable operation pulls the instrument during logging, and the instrument is influenced by the up-and-down mechanical shaking of the instrument, and the up-and-down drift of the position of the magnetic field probe at the receiving and transmitting time causes the fluctuation of a noise curve; thirdly, the temperature of the pipe column drifts, the temperature gradually rises along with the increase of the depth of the pipe column, and the detection voltage signal shows a sectional drift phenomenon under the influence of the temperature. In order to improve the accuracy of the interpretation of the field test data, data noise reduction processing is required to improve the signal to noise ratio and enhance the response of useful signals.
The traditional electromagnetic signal noise reduction method at present mainly comprises moving average noise reduction, median noise reduction, wavelet noise reduction and EMD noise reduction. The moving average noise reduction is mainly used for denoising the Gaussian white noise which is distributed randomly, and distortion is easy to occur to high-frequency noise; the median noise reduction can eliminate the interference of isolated points and line segments, but is not suitable for signals with larger fluctuation amplitude; wavelet noise reduction has good time-frequency localization property, but has slightly poorer signal effect on low signal-to-noise ratio; EMD noise reduction improves the processing effect on signals with low signal-to-noise ratio, but it is difficult to remove co-band noise.
For example, the chinese patent application with publication number CN110221349a discloses a method for noise reduction of transient electromagnetic signals based on wavelet transform and sine wave estimation, which processes white noise and power frequency interference mainly through wavelet basis function, interpolation extension and FFT transform, and improves the signal-to-noise ratio of the transient electromagnetic signals. However, both wavelet analysis and FFT analysis in this method can only eliminate single noise. Under the actual measurement condition, electromagnetic noise has randomness, diversity and complexity, has no obvious rule in depth, and the noise elimination and noise reduction treatment is carried out by using a single method, so that the effect is not ideal, and the noise reduction treatment under the complex condition of the multi-layer tubular column of the oil casing can not be satisfied.
Disclosure of Invention
The invention aims to provide an electromagnetic signal noise reduction method which is used for solving the problem that the noise reduction effect of the prior art method is poor.
In order to solve the technical problems, the invention provides an electromagnetic signal noise reduction method, which comprises the following steps:
1) Acquiring an original electromagnetic signal which changes along with time, and inputting the original electromagnetic signal into a trained noise model to obtain a noise signal comprising oil pipe eccentricity, mechanical high-frequency jitter and pipe column temperature drift; wherein, the noise model is:
wherein y represents a noise signal; softmax () represents the softmax function; x represents an input signal; x represents time; w (W) 1 And W is 2 Are weight matrixes; b 1 、b 2 And b 3 All represent deviation values; a represents the vibration scale; s is a variance parameter of white noise obeying normal distribution N; and W is 1 、W 2 、b 1 、b 2 、b 3 Training to obtain the A;
2) Subtracting the noise signal from the original electromagnetic signal to obtain an electromagnetic signal after primary noise reduction.
The beneficial effects are as follows: the invention provides a noise model comprising oil pipe eccentricity, mechanical high-frequency jitter and pipe column temperature drift through electromagnetic detection mechanism and noise source analysis in the detection process, and inputs an original electromagnetic signal which changes with time into the noise model, so that a main noise signal of the oil pipe comprising the oil pipe eccentricity, the mechanical high-frequency jitter and the pipe column temperature drift can be obtained, and further the original electromagnetic signal is subtracted to obtain an electromagnetic signal after noise reduction once, thereby improving the signal-to-noise ratio of the electromagnetic signal, ensuring the accuracy of detection signals under complex working conditions, and the accuracy of interpretation can be improved by utilizing the signal to perform on-site data interpretation, thereby being applicable to noise reduction of oil casing detection signals.
Further, in the training process, the difference value of the measured electromagnetic signal of the real well logging minus the measured electromagnetic signal of the non-defective experimental well is used as a tag noise signal, so that the measured electromagnetic signal of the real well logging and the corresponding tag noise signal are used as training data.
Further, the root mean square error of the predicted noise signal and the label noise signal obtained by inputting the electromagnetic signal of the real well logging into the noise model is used as a loss function, and the noise model is optimized by using a gradient descent method so as to obtain a trained noise model.
Further, steps 3) to 5) are included: 3) Performing modal decomposition on the electromagnetic signal subjected to primary noise reduction to obtain a plurality of components; 4) Determining whether each of the decomposed modal components is an effective component or a noise component: if the noise component is the noise component, noise reduction processing is carried out; 5) And reconstructing the effective component and the noise-reduced component to obtain an electromagnetic signal after secondary noise reduction.
The beneficial effects are as follows: the electromagnetic signal after primary noise reduction is subjected to secondary noise reduction, so that other noise signals except for oil pipe eccentricity, mechanical high-frequency jitter and pipe column temperature drift can be further eliminated, and the method is suitable for complex practical conditions. And only noise components are subjected to noise reduction during noise reduction, loss of effective components is avoided, and the effect of the noise reduction is ensured.
Further, in step 4), the noise component is determined by the following method: calculating the correlation coefficient of each component:R(c i x') represents the ith component c i X' represents the electromagnetic signal after primary noise reduction, and C () represents the covariance; a component whose correlation coefficient is smaller than the set correlation threshold is regarded as a noise component.
The beneficial effects are as follows: the noise component is identified by utilizing the correlation between the component and the electromagnetic signal after primary noise reduction, and the judgment of low correlation is the noise component.
Further, the correlation threshold is set to 0.1.
Further, the noise reduction process in step 4) is a wavelet threshold denoising process, and the wavelet base selected when the wavelet threshold denoising process is performed is Sym6, and the unbiased likelihood estimation method is used to select the threshold.
Further, the modal decomposition is an empirical modal decomposition.
Drawings
FIG. 1 is a flow chart of a method of electromagnetic signal noise reduction of the present invention;
FIG. 2 (a) is a raw data comparison plot;
FIG. 2 (b) is a graph of data alignment after gain removal;
FIG. 2 (c) is a graph comparing logarithmic data after gain removal;
FIG. 3 (a) is a summary graph of the error between the experimental and measured wells after de-gain;
FIG. 3 (b) is a graph of the signal-to-noise ratio of experimental and measured well errors after de-boosting;
FIG. 4 (a) is a graph of noise test set error variation;
FIG. 4 (b) is a graph of signal-to-noise ratio after denoising using the method of the present invention;
FIG. 5 (a) is a graph of response signals after secondary denoising;
fig. 5 (b) is a partial enlarged view of fig. 5 (a);
FIG. 6 (a) is a graph of noise test set error variation;
fig. 6 (b) is a signal-to-noise ratio diagram after denoising.
Detailed Description
The noise reduction method is a composite noise reduction method, and the composite noise reduction method adopts a composite noise reduction method of twice noise reduction. The primary noise reduction is mainly based on the established noise model to realize primary noise reduction treatment of electromagnetic signals of the oil pipe and the sleeve, and comprises eccentric noise reduction, shaking noise reduction and temperature noise reduction; the secondary noise reduction is mainly to extract effective components and noise components in the signals after the primary noise reduction is processed through empirical mode decomposition, and to perform the noise reduction on the noise components so as to complete the secondary noise reduction and effectively improve the signal to noise ratio. The present invention will be described in detail with reference to the accompanying drawings and examples.
Electromagnetic signal noise reduction method embodiment:
the overall flow of the electromagnetic signal noise reduction method embodiment of the invention is shown in figure 1, and the steps are as follows:
step one, constructing and training a noise model to obtain a trained noise model.
1. And constructing a forward simulation signal sample, and establishing a corresponding relation with the measurement data of the flawless experimental well.
1) And constructing a non-defective double-layer pipe column forward simulation model of the oil pipe and the sleeve, and determining a forward simulation signal which is a noiseless electromagnetic signal and is taken as a standard perfect sample.
2) The test data of the experimental well X1 is selected to be a defect-free section, and is compared and analyzed with the forward simulation data, the effect diagram is shown in fig. 2 (a) to 2 (c), the simulation signal length is 300 sampling points (0-0.2 ms), the test well measurement signal is 11 sampling points (0.015-0.09 ms), the forward simulation signal without gain is larger than the test well signal in phase difference from fig. 2 (a), the signal without gain and the logarithmic signal show the same trend from fig. 2 (b) and 2 (c), and the signal without gain and the logarithmic signal show the same trend and are relatively similar in value.
3) And calculating the Pearson linear correlation coefficient of the forward simulation signal and the measurement signal of the defect-free experimental well at each depth, and establishing a linear fitting function. Defining input data X as a measurement signal of a defect-free experimental well, and defining a target signal Y as a forward simulation signal, wherein the model is defined as follows:
Y＝W T X+b (1)
where W is a weight matrix and b is a deviation amount.
4) Randomly selecting 1500 groups of measurement signals of flawless experimental wells as samples, setting root mean square error as a loss function, optimizing parameters by a gradient descent method, iterating for 100 periods to obtain an optimal model, and fitting the difference between the 1500 groups of measurement signals of flawless experimental wells and forward simulation signals by using the model shown in the formula (1).
2. Noise analysis is performed on the measurement signals of the non-defective experimental well X1 and the measurement signals of the non-defective real well D1, and due to noise interference, certain errors exist in the measurement signals of the non-defective experimental well and the measurement signals of the non-defective real well, as shown in fig. 3 (a) and fig. 3 (b), the abscissa of the two graphs is depth data, and the average of error signal to noise ratios after gain removal is 7.3825. The noise sources of signals, mainly oil pipe eccentricity, mechanical high-frequency jitter and pipe column temperature drift, are analyzed aiming at noise interference existing in field test data, and a targeted model is built respectively.
1) For noise generated in electromagnetic detection of oil pipe eccentricity, white noise item is added into linear modelObtaining a linear-noise model, defining input data x as time and target data y as noise signals, wherein the model is defined as follows:
where W is a weight matrix, b is a deviation amount, and s is a variance parameter of white noise subjected to normal distribution.
2) For noise generated by jitter in electromagnetic detection, it is proposed to approximate mechanical vibration noise using the sine curve of asinx+b, resulting in a vibration-linear-noise model:
wherein W is a weight matrix, b 1 And b 2 Are deviation amounts, s is a variance parameter of white noise obeying normal distribution N, and A is a vibration scale.
3) For noise generated by temperature in electromagnetic detection, a linear offset weighting coefficient caused by temperature is calculated by adopting a softmax function, an input data x is defined as time by combining a previous vibration noise formula, a target data y is a noise signal, and a new depth-vibration-linearity-noise model (namely a noise model) is defined as follows:
in which W is i And b i Is a weight matrix and a bias value.
4. Taking the measurement signal without defects of the experimental well X1 as a clean signal, subtracting the clean signal from the measurement signal of the actual measurement well D1 to obtain a label noise signal, selecting a noise model shown in a depth and time data input formula (4) of 200 groups of training data (total 2000 groups) to obtain a prediction noise signal, taking the root mean square error of the generated prediction noise signal and the label noise signal as a loss function, using a gradient descent optimization noise model to obtain unknown parameters in the model, wherein the unknown parameters comprise W 1 、W 2 、b 1 、b 2 、b 3 And A. And the rest data are selected as test data to be input into an optimized noise model, the relative errors of the measurement signals of the experimental well and the actual well are calculated and compared randomly, and the relative errors of the signals after denoising the measurement signals of the actual well and the measurement signals of the experimental well are calculated, so that the trained noise model meets the requirements under the condition that the relative errors meet certain requirements, and the whole training and the test are completed, so that the method can be used for predicting the actual noise signals.
Step two, obtaining an original electromagnetic signal (comprising time) of the actual measurement well along with the time change, and inputting the original electromagnetic signal into a trained noise model to obtain a noise signal comprising oil pipe eccentricity, mechanical high-frequency jitter and pipe column temperature drift. The noise signal is subtracted from the original electromagnetic signal, so as to obtain an electromagnetic signal after noise reduction, the error and the signal-to-noise ratio of which are shown in fig. 4 (a) and fig. 4 (b), the abscissa of the two figures is the depth, and the average signal-to-noise ratio after noise removal is 12.1582. The electromagnetic signal after primary noise reduction can eliminate most of interference existing in field data, including three kinds of noise including casing eccentricity, mechanical high-frequency jitter and pipe column temperature drift. However, the actual situation is complex, and other types of noise exist besides the three types of noise, and the step three is executed to perform subsequent processing to realize secondary noise reduction.
Thirdly, performing secondary noise reduction on the electromagnetic signal subjected to primary noise reduction by using a mode decomposition method and wavelet threshold noise reduction combined method. The specific treatment process is as follows:
1. and (3) performing modal decomposition of the signal on the electromagnetic signal obtained in the step two after primary noise reduction by adopting an Empirical Mode Decomposition (EMD) method, so as to obtain a plurality of IMF components. Let the electromagnetic signal after one noise reduction be x '(t), the electromagnetic signal after one noise reduction x' (t) can be expressed as:
wherein, c i (t) represents the decomposed ith IMF component, r n (t) represents the decomposed difference signal sequence.
2. The correlation coefficient for each IMF component is calculated according to the following formula:
wherein R (c) i X') denotes the ith IMF component c i X' represents the electromagnetic signal after one noise reduction, and C () represents the covariance.
3. Judging whether each IMF component is an effective component or a noise component by using a set correlation threshold value: if the correlation coefficient is smaller than the set correlation threshold value of 0.1, the component is a noise component, otherwise, the component is an effective component.
4. And for the noise component, adopting a wavelet threshold denoising processing method to perform denoising processing. The wavelet threshold denoising is in the prior art, and the whole processing flow is as follows: 1) Wavelet decomposition, namely selecting a wavelet to carry out wavelet decomposition on an input signal, wherein Sym6 is selected as a wavelet basis in the process; 2) Thresholding, namely thresholding the decomposed coefficient of each layer to obtain an estimated wavelet coefficient, wherein an unbiased likelihood estimation threshold method in a soft threshold is used for selecting a threshold in the process; 3) And carrying out wavelet reconstruction according to the denoised estimated wavelet coefficient to obtain a denoised signal.
5. And reconstructing the effective component and the noise-reduced component to obtain an electromagnetic signal after secondary noise reduction. The electromagnetic corresponding signal after secondary noise cancellation is shown in fig. 5 (a), and a partially enlarged schematic view of the square area in fig. 5 (a) is shown in fig. 5 (b).
Thus, the electromagnetic signal noise reduction method is completed. The method is applied to double-pipe signals of field real well logging, and processing analysis is carried out. The error and snr curves after denoising the measured well defect segment are shown in fig. 6 (a) and fig. 6 (b), respectively, the abscissa of the two graphs is depth data, and the average snr after denoising is 20.1496. After the two-time noise reduction processing method is applied, the average signal to noise ratio is improved to 20.14, and the average signal to noise ratio is greatly improved.
In summary, firstly, through analyzing the electromagnetic detection mechanism and the noise sources in the detection process, an intelligent electromagnetic detection noise reduction method is innovated, a noise reduction model combining the oil pipe mechanism characteristics based on a deep learning theory is provided for one-time noise reduction, the noise reduction model can eliminate noise, the signal to noise ratio is improved, and the interpretation accuracy of on-site data is improved; and secondly, combining the proposed noise model with modal decomposition, namely performing secondary noise reduction by using empirical mode decomposition after primary noise reduction, and enabling the result to be more suitable for noise reduction of the oil casing detection signal. Experiments prove that the noise reduction method has universality and solves the problem that electromagnetic detection signal noise is complex and difficult to eliminate.
Furthermore, the electromagnetic signal noise reduction method described above can be implemented by computer program codes, and in particular, an electromagnetic signal noise reduction device can be designed. Specifically, the device comprises a memory, a processor and an internal bus, wherein the processor and the memory are communicated with each other and data are interacted through the internal bus. The memory includes at least one software functional module stored in the memory, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory to implement an electromagnetic signal noise reduction method of the present invention. The processor may be a microprocessor MCU, a programmable logic device FPGA, or other processing device. The memory may be various memories for storing information by using electric energy, such as RAM, ROM, etc.; the magnetic storage device can also be various memories for storing information by utilizing a magnetic energy mode, such as a hard disk, a floppy disk, a magnetic tape, a magnetic core memory, a bubble memory, a U disk and the like; various memories for optically storing information, such as CDs, DVDs, etc.; of course, other types of memory are also possible, such as quantum memory, graphene memory, etc.
Claims (8)
1. A method for reducing noise of an electromagnetic signal, comprising the steps of:
1) Acquiring an original electromagnetic signal which changes along with time, and inputting the original electromagnetic signal into a trained noise model to obtain a noise signal comprising oil pipe eccentricity, mechanical high-frequency jitter and pipe column temperature drift; wherein, the noise model is:
wherein y represents a noise signal; softmax () represents the softmax function; x represents an input signal; x represents time; w (W) 1 And W is 2 Are weight matrixes; b 1 、b 2 And b 3 All represent deviation values; a represents the vibration scale; s is a variance parameter of white noise obeying normal distribution N; and W is 1 、W 2 、b 1 、b 2 、b 3 Training to obtain the A;
2) Subtracting the noise signal from the original electromagnetic signal to obtain an electromagnetic signal after primary noise reduction.
2. The electromagnetic signal noise reduction method according to claim 1, wherein in the training process, a difference obtained by subtracting the measured electromagnetic signal of the fault-free experimental well from the measured electromagnetic signal of the real well logging is used as a tag noise signal, so that the measured electromagnetic signal of the real well logging and the corresponding tag noise signal are used as training data.
3. The electromagnetic signal noise reduction method according to claim 2, wherein a root mean square error of a predicted noise signal and a tag noise signal obtained by inputting an electromagnetic signal of a real well logging into the noise model is used as a loss function, and the noise model is optimized using a gradient descent method to obtain a trained noise model.
4. The electromagnetic signal noise reduction method according to claim 1, further comprising steps 3) to 5):
3) Performing modal decomposition on the electromagnetic signal subjected to primary noise reduction to obtain a plurality of components;
4) Determining whether each of the decomposed modal components is an effective component or a noise component: if the noise component is the noise component, noise reduction processing is carried out;
5) And reconstructing the effective component and the noise-reduced component to obtain an electromagnetic signal after secondary noise reduction.
5. The method of noise reduction of electromagnetic signals according to claim 4, wherein in step 4), the noise component is determined by:
calculating the correlation coefficient of each component:R(c i x') represents the ith component c i X' represents the electromagnetic signal after primary noise reduction, and C () represents the covariance;
a component whose correlation coefficient is smaller than the set correlation threshold is regarded as a noise component.
6. The method of noise reduction of electromagnetic signals according to claim 5, wherein the correlation threshold is set to 0.1.
7. The electromagnetic signal denoising method according to claim 4, wherein the denoising process in step 4) is a wavelet threshold denoising process, and the wavelet basis selected when the wavelet threshold denoising process is performed is Sym6, and the threshold is selected using an unbiased likelihood estimation method.
8. The method of electromagnetic signal noise reduction according to claim 4, wherein the modal decomposition is an empirical mode decomposition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210988015.8A CN117647846A (en) | 2022-08-17 | 2022-08-17 | Electromagnetic signal noise reduction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210988015.8A CN117647846A (en) | 2022-08-17 | 2022-08-17 | Electromagnetic signal noise reduction method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117647846A true CN117647846A (en) | 2024-03-05 |
Family
ID=90046510
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210988015.8A Pending CN117647846A (en) | 2022-08-17 | 2022-08-17 | Electromagnetic signal noise reduction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117647846A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117970500A (en) * | 2024-04-02 | 2024-05-03 | 中国科学技术大学 | Method for extracting weak transient electromagnetic signals under strong noise background |
-
2022
- 2022-08-17 CN CN202210988015.8A patent/CN117647846A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117970500A (en) * | 2024-04-02 | 2024-05-03 | 中国科学技术大学 | Method for extracting weak transient electromagnetic signals under strong noise background |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Model‐based method with nonlinear ultrasonic system identification for mechanical structural health assessment | |
| CN108106830B (en) | A kind of Variable Speed Rotating Machinery method for diagnosing faults based on time-frequency spectrum segmentation | |
| CN103336049B (en) | A kind of pulse eddy current detection method and device eliminating Lift-off effect | |
| Wang et al. | Crack characterization in ferromagnetic steels by pulsed eddy current technique based on GA-BP neural network model | |
| CN110286155B (en) | Damage detection method and system for multilayer composite material | |
| CN112084982B (en) | Double-layer tubular column pulse vortex data denoising method based on noise model | |
| CN108108672B (en) | Random resonance current weak information identification method based on linear search strategy | |
| CN117647846A (en) | Electromagnetic signal noise reduction method | |
| Li et al. | An improved AIC onset-time picking method based on regression convolutional neural network | |
| CN114487129B (en) | Flexible material damage identification method based on acoustic emission technology | |
| CN114970646B (en) | Artificial source electromagnetic pseudorandom signal detrending and noise identification method | |
| US20110166842A1 (en) | Layer stripping method | |
| Han et al. | A novel rolling bearing fault diagnosis method based on generalized nonlinear spectral sparsity | |
| CN116432071A (en) | Rolling bearing residual life prediction method | |
| CN117332371A (en) | Material service behavior multidimensional detection method based on laser photoacoustic information fusion | |
| CN113624834B (en) | Defect depth identification method and system based on marginal spectrum centroid detection | |
| Song et al. | An improved structural health monitoring method utilizing sparse representation for acoustic emission signals in rails | |
| CN111505709B (en) | Attenuation qualitative analysis method based on sparse spectral decomposition | |
| CN111191502B (en) | Stick-slip and jump drill abnormal working condition identification method based on drill string vibration signal | |
| Liu et al. | Accurate wire rope defect MFL detection using improved Hilbert transform and LSTM neural network | |
| Li et al. | A new method for weak fault feature extraction based on improved MED | |
| CN117214398A (en) | Deep underground water body pollutant detection method and system | |
| CN112378895A (en) | Method and system for measuring water content of crude oil, storage medium and equipment | |
| CN116956119A (en) | Sediment noise simulation-based shale layer sequence stratum division method and device | |
| Liu et al. | Application of Hilbert–Huang transform for defect recognition in pulsed eddy current testing |
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 |