CN110907997A - Rapid identification method and device for double-parameter time domain electromagnetic weak polarization effect - Google Patents
Rapid identification method and device for double-parameter time domain electromagnetic weak polarization effect Download PDFInfo
- Publication number
- CN110907997A CN110907997A CN201910998449.4A CN201910998449A CN110907997A CN 110907997 A CN110907997 A CN 110907997A CN 201910998449 A CN201910998449 A CN 201910998449A CN 110907997 A CN110907997 A CN 110907997A
- Authority
- CN
- China
- Prior art keywords
- polarization
- parameters
- zero
- electromagnetic
- unit
- 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.)
- Granted
Links
- 230000010287 polarization Effects 0.000 title claims abstract description 115
- 230000000694 effects Effects 0.000 title claims abstract description 59
- 238000000034 method Methods 0.000 title claims abstract description 50
- 230000004044 response Effects 0.000 claims abstract description 74
- 238000013145 classification model Methods 0.000 claims abstract description 29
- 238000012216 screening Methods 0.000 claims abstract description 20
- 238000012706 support-vector machine Methods 0.000 claims abstract description 15
- 238000000605 extraction Methods 0.000 claims description 26
- 238000005259 measurement Methods 0.000 claims description 21
- 239000013598 vector Substances 0.000 claims description 17
- 238000005070 sampling Methods 0.000 claims description 14
- 238000004458 analytical method Methods 0.000 claims description 7
- 230000001960 triggered effect Effects 0.000 claims description 6
- 239000006185 dispersion Substances 0.000 claims description 4
- 230000001360 synchronised effect Effects 0.000 claims description 4
- 239000000284 extract Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000004088 simulation Methods 0.000 claims description 3
- 230000003247 decreasing effect Effects 0.000 claims 1
- 238000001514 detection method Methods 0.000 abstract 2
- 238000012360 testing method Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 230000009977 dual effect Effects 0.000 description 4
- 206010027336 Menstruation delayed Diseases 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 239000002184 metal Substances 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 230000003321 amplification Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000005672 electromagnetic field Effects 0.000 description 1
- 230000005670 electromagnetic radiation Effects 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
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
- G01V3/08—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
Landscapes
- Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Electromagnetism (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention relates to a method and a device for rapidly identifying a double-parameter time domain electromagnetic weak polarization effect, which are used for screening two parameters which have the greatest influence on a polarization identification result aiming at the problems of difficult identification of weak polarization response characteristics, low interpretation precision and the like in time domain electromagnetic detection data; taking a parameter with the largest influence as input, and taking 'whether polarization occurs' as output, and establishing a polarization effect rapid classification model based on a support vector machine; and rapidly identifying the measured data through a polarization effect rapid classification model. According to the invention, the weak polarization response in electromagnetic detection can be rapidly and accurately identified only by extracting double parameters, so that the accuracy of electromagnetic data interpretation is improved.
Description
Technical Field
The invention relates to a method for identifying a polarization effect, in particular to a method and a device for quickly identifying a double-parameter time domain electromagnetic weak polarization effect.
Background
Induced Polarization Effects (IP) are an important electrochemical phenomenon, usually occurring in metal ores, impregnated mineral resources and water-bearing geology, with extremely high economic value, mainly represented by the phenomena of rapid attenuation and sign reversal of electromagnetic signals. The method for performing multi-parameter joint interpretation on geology by utilizing polarization data is also widely applied to wider fields of searching metal ores, water resources, geothermal resources, karst and the like, and becomes a research hotspot of scholars at home and abroad. However, for unknown geological structures, whether the geological structures have polarization characteristics or not is concluded by visually recognizing whether sign reversal (negative response) single characteristics of measured electromagnetic response data occur or not, and for strong polarization media, the electromagnetic response data can be mostly distinguished, while for weak polarization or more complex polarization terrains, the visual recognition mode has great subjective assumption and contingency, so that the method has limitations in practical application and can cause weak polarization effects to be ignored. Therefore, the rapid and accurate identification of the time domain weak polarization effect is a key problem to be solved urgently.
British patent CN101189533A discloses a method and apparatus for object recognition and classification of electromagnetic signals, in which electromagnetic radiation is emitted by a sensor, the analysis is performed by comparing the received signal with stored characteristic values, and the classification of the object is deduced from the analysis. The identification method mainly completes the whole identification classification by comparing the received signals with stored graphs and analysis after the received signals are transformed, mainly processes electromagnetic signals, and has more complicated steps and more complex processes.
Chinese patent CN108830325A discloses a learning-based vibration information terrain classification and identification method, which comprises the steps of collecting vibration information, extracting feature vectors, and performing off-line learning training on the feature vectors converted into a frequency domain by using a multilayer feedforward neural network to obtain a trained multilayer feedforward neural network; and carrying out online classification and identification to obtain the terrain type. However, the classification model inputs all the characteristic parameters into the classification model, the influence of redundant characteristic parameters is not considered, the processing of a large amount of data is not facilitated, and the identification and judgment of the polarization geology are not involved.
In 2007, the qining et al uses 6 typical polarized layered place models as recognition standards, maps a one-dimensional polarization response curve to a two-dimensional space image by a time-frequency conversion method, observes energy distribution characteristics of data in a space, calculates Euclidean distance, absolute value distance and similarity of the image, totals 18 parameters, and classifies the image by using a minimum value as a criterion. However, in the method, the selected standard curve is strongly polarized and has a negative response value, the polarization geological conditions of weak polarization and no negative response value are not considered, and the method is not beneficial to comprehensively grasping the underground polarization information, so that a method and a device capable of quickly and accurately identifying the weak polarization effect are required to be designed.
Disclosure of Invention
The invention aims to provide a double-parameter rapid identification method and a device for time domain electromagnetic polarization effect, including weak polarization effect, according to the characteristics of a polarization response curve of time domain electromagnetic data aiming at the defects of the existing weak polarization effect identification method.
The present invention is achieved in such a way that,
a fast identification method of double-parameter time domain electromagnetic weak polarization effect comprises the following steps:
screening two parameters which have the largest influence on the polarization recognition result;
taking a parameter with the largest influence as input, and taking 'whether polarization occurs' as output, and establishing a polarization effect rapid classification model based on a support vector machine;
and rapidly identifying the measured data through a polarization effect rapid classification model.
Further, the process of screening the parameters having the greatest influence on the polarization recognition result comprises the following steps:
s1, referring to the actual range of polarization parameters in geological data, including conductivity, polarizability, time constant and frequency dispersion coefficient, setting a plurality of groups of polarized and non-polarized three-dimensional earth models, and obtaining three-component electromagnetic response attenuation curves through numerical simulation;
s2, performing characteristic analysis on the electromagnetic response of S1, sampling an electromagnetic response curve around two large polarization characteristics of fast attenuation and negative response to obtain a plurality of characteristic parameters, and calculating the correlation between the characteristic parameters and the recognition result;
and S3, screening out two parameters with the maximum correlation with the output result according to the correlation result.
Further, the characteristic parameters in S2 include the following:
three-component magnetic field Bx(t)、By(t)、Bz(T) duration of negative response of the respective decay curves DeltaTx、ΔTy、ΔTz(ii) a Three-component magnetic field positive response maximum Bmax+And negative response maximum Bmax-Ratio P ofx、Py、Pz(ii) a Fitting late slope k of three-component attenuation curve under log-log coordinates by adopting least square methodx、ky、kz(ii) a N negative response amplitudes A with sampling interval delta t1、A2…ANN +9 characteristic parameters are totally processed by normalization. For a curve in which no negative response occurs in the non-polarized medium or the polarized medium, the negative response duration, the negative response amplitude, and the ratio of the positive to negative response amplitudes are all zero.
Further, the S2 further includes the following steps:
marking X as any characteristic parameter vector in the characteristic parameters, marking Y as 1 for polarized media and marking Y as-1 for non-polarized media, and calculating mutual information between each characteristic parameter X and an output result Y according to a formula (1):
in the formula, pijIs a joint probability density, p, between two variablesipjThe larger the I (X; Y) value is, the stronger the correlation between the two is.
Further, in S3, the akage pool information criterion is used as a screening end condition, the current akage pool information value is calculated for the screened feature vectors, and when the feature vectors are not reduced, the current feature vectors are considered to be used to optimize the fitting, so that the screening of the optimal set is ended.
Further, the parameters with the largest influence are calculated to obtain the sign inversion time and the late slope.
A method for rapidly identifying a double-parameter time domain electromagnetic weak polarization effect comprises the following steps:
zero scanning is carried out on the actually measured data to obtain a symbol inversion moment, and a late slope is extracted through nonlinear least square fitting;
taking the sign inversion time and the late slope as input, and taking 'whether polarization occurs' as output to establish a polarization effect rapid classification model based on a support vector machine;
the model is transplanted to a time domain electromagnetic receiving device, and the measured data is rapidly identified through a polarization effect rapid classification model.
Further, the zero-point scanning the measured data to obtain the symbol inversion time includes: the electromagnetic attenuation curve in the polarized medium is gradually reduced from a positive value to a negative value, the actually measured signal value is compared with a zero value, and when the curve is lower than zero for the first time, the current time value is recorded to obtain the sign reversal time;
extracting the late slope by nonlinear least squares fitting further comprises: and when the late slope is extracted, a polarization attenuation curve with the time period of 1ms is intercepted from the tail end of the curve, and a late slope value under a double-exponential coordinate is obtained by adopting nonlinear least square fitting.
A fast identification device for double-parameter time domain electromagnetic weak polarization effect comprises an actual measurement signal reading unit, a zero feature extraction unit, a synchronization unit and a main control unit, wherein the actual measurement signal reading unit, the zero feature extraction unit, the synchronization unit and the main control unit are arranged in the same plane, and the main control unit is used for controlling the actual measurement signal reading unit and the
The actual measurement signal reading unit comprises a receiving coil front end amplifying circuit and is used for acquiring the actual measurement electromagnetic response in the field;
the zero point characteristic extraction unit is used for extracting a symbol reversal moment from an actual measurement electromagnetic response curve according to actual measurement electromagnetic response data acquired by the actual measurement signal reading unit;
the master control unit comprises a late slope characteristic extraction unit, a model establishment module and a rapid classification unit, and extracts the late slope of the electromagnetic response curve by adopting a least square fitting method through the late slope characteristic extraction unit;
the model establishing module is used for establishing a polarization effect rapid classification model based on a support vector machine by taking the sign reversal time and the late slope as input and taking 'whether polarization occurs' as output;
the rapid classification unit is used for rapidly identifying the double parameters with the electrodeless effect by using a pre-constructed polarization effect classification model;
the synchronization unit adopts GPS synchronization, is directly connected with the main control unit and the zero point feature extraction unit, and is used for realizing the synchronization of the receiver and the transmitter and triggering the counter in the zero point feature extraction unit;
further, the zero point feature extraction unit comprises a zero-crossing comparator, a counter and an AD collector, and after receiving the synchronous signal, the AD collector starts sampling and converts the received voltage quantity into a digital quantity; and meanwhile, the counter starts counting, when the zero-crossing comparator detects the zero crossing point of the signal, the counter is triggered to stop counting, and the symbol inversion time is calculated according to the counting size of the counter.
Compared with the prior art, the invention has the beneficial effects that: the method and the device can not only extract double parameters to complete quick identification, but also accurately identify the weak polarization effect, thereby improving the accuracy of electromagnetic data interpretation. The method solves the technical problem of low accuracy caused by subjective assumption and contingency when the polarization effect is identified by naked eyes in the prior art, greatly simplifies the identification steps, realizes the rapid and accurate identification of the polarization effect, and reduces the exploration cost.
Drawings
FIG. 1 is a schematic flow chart of the method of the present invention.
FIG. 2 is a graph of the present invention for least squares fitting to a curve to extract slope values.
FIG. 3 is a diagram of the test results of the two-parameter fast classification model established based on the support vector machine according to the present invention.
FIG. 4 is a structural diagram of the fast dual-parameter classification device for weak polarization effect according to the present invention.
FIG. 5 is a graph of measured curves from an actual engineering survey.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
A fast identification method of double-parameter time domain electromagnetic weak polarization effect comprises the following steps:
screening two parameters which have the largest influence on the polarization recognition result;
taking a parameter with the largest influence as input, and taking 'whether polarization occurs' as output, and establishing a polarization effect rapid classification model based on a support vector machine; and rapidly identifying the measured data through a polarization effect rapid classification model.
The method specifically comprises the following steps:
s1, setting a plurality of groups of polarized and non-polarized three-dimensional layered earth models by referring to the actual range of polarization parameters in geological data, including conductivity, polarizability, time constant and frequency dispersion coefficient, and obtaining three-component electromagnetic response attenuation curves through numerical simulation;
s2, performing characteristic analysis on the electromagnetic response of S1, sampling the electromagnetic response curve at equal intervals around two large polarization characteristics of fast attenuation and negative response to obtain a plurality of characteristic parameters, and calculating mutual information among the characteristic parameters and between the characteristic parameters and an output result according to a mutual information formula;
s3, screening two parameters with the strongest correlation with the output result according to the correlation result: sign inversion time, late slope;
s4, carrying out zero point scanning on the measured data to obtain the sign inversion time, and extracting the late slope through nonlinear least square fitting; the zero point scanning of the measured data to obtain the symbol inversion time comprises the following steps: the electromagnetic attenuation curve in the polarized medium is gradually reduced from a positive value to a negative value, the actually measured signal value is compared with a zero value, and when the curve is lower than zero for the first time, the current time value is recorded to obtain the sign reversal time;
extracting the late slope by nonlinear least squares fitting further comprises: and when the late slope is extracted, a polarization attenuation curve with the time period of 1ms is intercepted from the tail end of the curve, and a late slope value under a double-exponential coordinate is obtained by adopting nonlinear least square fitting.
S5, taking the double parameters obtained by screening as an input sample set, taking 'whether polarization occurs' as output, and establishing a double-parameter polarization effect rapid classification model based on a support vector machine;
s6, directly transplanting the method to a field receiving device, and rapidly identifying the measured data on site.
In S2, performing feature analysis on the electromagnetic response curve of S1, and extracting feature parameters around two large polarization features of fast attenuation and negative response, including:
three-component magnetic field Bx(t)、By(t)、Bz(T) decay Curve negative response duration Δ Tx、ΔTy、ΔTz；
Three-component magnetic field positive response maximum Bmax+And negative response maximum Bmax-Ratio P ofx、Py、Pz；
Fitting late slope k of three-component attenuation curve under log-log coordinates by adopting least square methodx、ky、kz；
N negative response amplitudes A with sampling interval delta t1、A2…AN；
N +9 characteristic parameters in total, and all the characteristic parameters are subjected to normalization processing;
for a curve without negative response (weak polarization) in a non-polarized medium or a polarized medium, the negative response duration, the negative response amplitude and the ratio of the positive response amplitude to the negative response amplitude are all zero; for polarized media, YiMarked 1, Y for non-polarized mediaiLabeled-1; (ii) a Calculating mutual information among all characteristic parameters and between all characteristic parameters and output results according to a formula (1):
in the formula, pijIs a joint probability density, p, between two variablesipjThe larger the I (X; Y) value is, the stronger the correlation is;
when the optimal characteristic parameters are screened in the step S3, the determination condition that the AIC value of the akage pool information criterion is not reduced any more is taken as the end of screening, which includes the following steps:
a. inputting the correlation value obtained in step S2;
b. selecting the characteristic parameter X with the maximum correlation valueqOutputting to an optimal set S;
c. calculate AIC value:
wherein n is the number of observed data, uiAnd p is the number of the selected variables, which is the residual error. Judging whether the AIC is reduced, if not, turning to the step h, otherwise, turning to the step d;
d. removing XqWith other parameters X1、X2...XnA coupling between them;
e. respectively calculating and eliminating coupled X1、X2...XnThe relevance value of (a);
f. selecting the characteristic parameter X with the maximum relevance value againPOutputting to an optimal set S;
g. turning to step c;
h. and outputting the optimal sample set S.
In the screening process, after obtaining the double parameters of the sign inversion time and the late slope, the AIC value starts to be reduced, the screening is finished, and the double parameters are output.
In step S5, the fast classification model is trained, which includes the following steps:
1) taking the optimal polarization sample set S obtained in the step S3 as an input vector, taking whether polarization Y occurs as an output vector, outputting the polarization as 1 if the polarization occurs, and otherwise outputting the polarization as-1;
2) the sample sequence is disturbed, 2/3 of the total samples are randomly selected as a training set, and the rest are test sets;
3) selecting a Gaussian mapping function RBF to map the low-dimensional feature vector to a high-dimensional feature vector space, so that linear irreversibility is converted into linear divisible;
4) establishing a hyperplane equation, setting constraint conditions, introducing a Lagrange operator to convert an original problem into a dual problem, and substituting the dual problem into a training set;
5) and (4) substituting the fast recognition model of the polarization effect of the trained support vector machine into the test set to test the recognition precision.
In the step S6, in the field discrimination of the actually measured electromagnetic field data, the fast classification model of the two-parameter weak polarization effect established in S5 is packaged in the main control chip of the time domain electromagnetic receiving device, and the field fast judgment is performed on the electromagnetic response collected by the field receiver.
A fast recognition device for double-parameter time domain electromagnetic weak polarization effect comprises an actual measurement signal reading unit, a zero feature extraction unit, a synchronization unit and a main control unit:
the actual measurement signal reading unit comprises a receiving coil and a front-end amplifying circuit, is connected with the input end of the zero point characteristic extraction unit, and obtains the field actual measurement electromagnetic response through the amplification of the front-end amplifying circuit after receiving the signal through the receiving coil;
the zero point characteristic extraction unit comprises a zero crossing comparator, a counter and an AD collector, is connected with the input end of the main control unit and is used for extracting the sign inversion time from the response curve; after receiving the synchronous signal, the AD collector starts sampling, and converts the received voltage quantity into a digital quantity readable by the singlechip; meanwhile, the counter starts counting, when the zero crossing comparator detects the zero crossing point of the signal, the counter is triggered to stop counting, and the symbol inversion time is calculated according to the counting size of the counter;
the main control unit comprises a late slope characteristic extraction unit, a model establishment module and a rapid classification unit, wherein the late slope characteristic extraction unit adopts least square fitting to extract the slope of a curve in the late period, the model establishment module takes the symbol inversion moment and the late slope as input, and takes 'whether polarization occurs' as output to establish a polarization effect rapid classification model based on a support vector machine;
and the rapid classification unit uses a pre-constructed polarization effect classification model to rapidly identify the double parameters with the polarization effect. And the synchronization unit adopts GPS synchronization, is directly connected with the main control unit and the counter and is used for realizing the synchronization of the receiver and the transmitter and triggering the timer.
Also includes: the display unit is directly connected with the main control unit and is used for displaying a judgment result and the like; the storage unit is directly connected with the main control unit and adopts a micro SD card for storing a judgment result and the like; the power supply unit is directly connected with each unit in a mode of combining a low-voltage difference linear voltage-stabilized power supply module AMS1117-3.3 and an intelligent power supply module TPS767D301 and is used for supplying power to the whole time domain polarization effect quick identification device; the AMS1117-3.3 operates in standby mode and once triggered, the entire identification appliance will be powered by the TPS767D 301.
Example (b):
the invention provides a method for rapidly identifying a double-parameter time domain electromagnetic weak polarization effect, which comprises the following steps as shown in figure 1:
s1, setting 250 groups of three-dimensional models of uniform earth nonpolarization, uniform earth polarization, layered earth polarization and uniform earth nonpolarization according to the actual range of polarization parameters in geological data, including conductivity, polarizability, time constant and frequency dispersion coefficient, and obtaining an electromagnetic response curve;
s2, analyzing the electromagnetic response of S1 to obtain curve characteristics of fast attenuation and negative response, and extracting characteristic parameters around two large polarization characteristics:
three-component magnetic field Bx(t)、By(t)、Bz(T) decay Curve negative response duration Δ Tx、ΔTy、ΔTz；
Three-component magnetic field positive response maximum Bmax+And negative response maximum Bmax-Ratio P ofx、Py、Pz；
Fitting slope k of attenuation curve by least square methodx、ky、kz；
7 negative response amplitudes A with a sampling interval of 1500 microseconds1、A2…AN；
There are 16 polarization characteristic parameters, and all are normalized.
For curves where no negative response (weak polarization) occurs in the non-polarized medium or the polarized medium, the negative response duration, the negative response amplitude, and the ratio of the positive and negative response maxima are all zero. For polarized media, YiMarked 1, Y for non-polarized mediaiThe label is-1.
As shown in fig. 2, when different sampling times are selected for equal-interval sampling, slope value characteristic parameters of different segments can be obtained through fitting, and by analogy, different characteristic parameters can be obtained by changing sampling intervals.
Calculating the correlation among the characteristic parameters and the correlation between the characteristic parameters and the recognition result according to a formula (1):
in the formula, pijIs a joint probability density, p, between two variablesipjProbability density of independent distribution. The larger the I (X; Y) value, the stronger the correlation.
S3, when the optimal characteristic parameters are screened, taking the fact that the value of the Chichi pool information criterion AIC is not reduced any more as a judgment condition for the screening end, the method comprises the following steps:
a. inputting the correlation value obtained at S2;
b. selecting the characteristic parameter X with the maximum correlation valueqOutputting to an optimal set S;
c. calculate AIC value:
wherein n is the number of observed data, uiAnd p is the number of the selected variables, which is the residual error. Judging whether the AIC is reduced, if not, turning to the step h, otherwise, turning to the step d;
d. removing XqWith other parameters X1、X2...XnA coupling between them;
e. respectively calculating and eliminating coupled X1、X2...XnThe relevance value of (a);
f. selecting the characteristic parameter X with the maximum relevance value againPOutputting to an optimal set S;
g. turning to step c;
h. and outputting the optimal sample set S.
In the screening process, after obtaining the double parameters of the sign inversion time and the late slope, the AIC value starts to be reduced, the screening is finished, and the double parameters are output.
S4, carrying out zero point scanning on the measured data to obtain a symbol inversion moment, and extracting a late slope through nonlinear least square fitting;
s5, establishing a double-parameter polarization effect fast classification model based on a support vector machine by taking double parameters as input and 'whether polarization occurs' as output, and training the double-parameter polarization effect fast classification model by adopting a sample set S, wherein the method comprises the following steps:
1) taking the optimal polarization sample set S obtained in S4 as an input vector, taking whether polarization Y occurs as an output vector, outputting the polarization as 1 if the polarization occurs, and otherwise outputting the polarization as-1;
2) the sample sequence is disturbed, 2/3 of the total samples are randomly selected as a training set, and the rest are test sets;
3) selecting a Gaussian mapping function RBF to map the low-dimensional feature vector to a high-dimensional feature vector space, so that linear irreversibility is converted into linear divisible;
4) establishing a hyperplane equation, setting constraint conditions, introducing a Lagrange operator to convert an original problem into a dual problem, and substituting the dual problem into a training set;
5) and substituting the trained SVM polarization effect fast recognition model into the test set to test the recognition accuracy.
Fig. 3 is a test result diagram of an embodiment of the present invention, and it can be seen from the diagram that the recognition accuracy is more than 93%, the recognition time is 1 second, and the validity of the dual-parameter fast recognition method based on the support vector machine is fully verified.
S6, packaging the polarization effect fast classification model established in S5 in a main control chip of the time domain electromagnetic receiving device, and rapidly judging the electromagnetic response collected by the field receiver on site, wherein as shown in FIG. 4, the structure diagram of the double-parameter fast classification device for the time domain electromagnetic weak polarization effect comprises the following units:
the actual measurement signal reading unit consists of a receiving coil and a front end amplifying circuit, is connected with the input end of the zero point characteristic extraction unit and is used for acquiring the field actual measurement electromagnetic response;
the zero point characteristic extraction unit consists of a zero crossing comparator, a counter and an AD collector, is connected with the input end of the main control unit and is used for extracting the sign inversion time from the response curve; after receiving the synchronous signal, the AD collector starts sampling, and converts the received voltage quantity into a digital quantity readable by the singlechip; meanwhile, the counter starts counting, when the zero crossing comparator detects the zero crossing point of the signal, the counter is triggered to stop counting, and the symbol inversion time is calculated according to the counting size of the counter;
the main control unit consists of a late slope characteristic extraction unit, a model establishment module and a rapid classification unit, wherein the late slope characteristic extraction unit adopts least square fitting to extract the slope of a curve in a late period, the model establishment module takes the symbol inversion moment and the late slope as input, takes 'whether polarization occurs' as output, and establishes a polarization effect rapid classification model based on a support vector machine; the rapid classification unit uses a pre-constructed polarization effect classification model to rapidly identify the double parameters with the electrodeless effect;
the display unit is directly connected with the main control unit and is used for displaying a judgment result and the like;
the synchronization unit adopts GPS synchronization, is directly connected with the main control unit and the counter and is used for realizing the synchronization of the receiver and the transmitter and triggering the timer;
the storage unit is directly connected with the main control unit and adopts a micro SD card for storing a judgment result and the like;
the power supply unit is directly connected with each unit in a mode of combining a low-voltage difference linear voltage-stabilized power supply module AMS1117-3.3 and an intelligent power supply module TPS767D301 and is used for supplying power to the whole time domain polarization effect quick identification device; the AMS1117-3.3 operates in standby mode and once triggered, the entire identification appliance will be powered by the TPS767D 301.
Fig. 5 is a measured curve diagram in actual engineering exploration, as shown in the figure, negative values do not appear in the total response observed in the polarized geology, however, the polarization effect can be successfully identified through the method, and therefore the method provides a new idea and a new method for accurately identifying weak polarization in time domain electromagnetism.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. A fast identification method for a double-parameter time domain electromagnetic weak polarization effect is characterized by comprising the following steps:
screening two parameters which have the largest influence on the polarization recognition result;
taking a parameter with the largest influence as input, and taking 'whether polarization occurs' as output, and establishing a polarization effect rapid classification model based on a support vector machine;
and rapidly identifying the measured data through a polarization effect rapid classification model.
2. The method of claim 1, wherein the step of screening the parameters that most affect the polarization recognition result comprises:
s1, referring to the actual range of polarization parameters in geological data, including conductivity, polarizability, time constant and frequency dispersion coefficient, setting a plurality of groups of polarized and non-polarized three-dimensional earth models, and obtaining three-component electromagnetic response attenuation curves through numerical simulation;
s2, performing characteristic analysis on the electromagnetic response of S1, sampling an electromagnetic response curve around two large polarization characteristics of fast attenuation and negative response to obtain a plurality of characteristic parameters, and calculating the correlation between the characteristic parameters and the recognition result;
and S3, screening out two parameters with the maximum correlation with the output result according to the correlation result.
3. The method of claim 2, wherein the characteristic parameters in S2 include the following:
three-component magnetic field Bx(t)、By(t)、Bz(T) duration of negative response of the respective decay curves DeltaTx、ΔTy、ΔTz(ii) a Three-component magnetic field positive response maximum Bmax+And negative response maximum Bmax-Ratio P ofx、Py、Pz(ii) a Fitting late slope k of three-component attenuation curve under log-log coordinates by adopting least square methodx、ky、kz(ii) a N negative response amplitudes A with sampling interval delta t1、A2…ANN +9 characteristic parameters are totally processed by normalization. For a curve in which no negative response occurs in the non-polarized medium or the polarized medium, the negative response duration, the negative response amplitude, and the ratio of the positive to negative response amplitudes are all zero.
4. The method of claim 2, wherein said S2 further comprises the following:
marking X as any characteristic parameter vector in the characteristic parameters, marking Y as 1 for polarized media and marking Y as-1 for non-polarized media, and calculating mutual information between each characteristic parameter X and an output result Y according to a formula (1):
in the formula, pijIs a joint probability density, p, between two variablesipjThe larger the I (X; Y) value is, the stronger the correlation between the two is.
5. The method as claimed in claim 2, wherein the step S3 is performed by using the akage information criterion as a screening end condition, calculating a current akage information value for the screened feature vectors, and when the current akage information value is not decreased any more, considering that the current feature vectors are used to optimize the fitting, thereby obtaining the end of screening of the optimal set.
6. A method according to claim 1, characterized in that the most influential parameters are calculated to obtain the sign inversion instant and the late slope.
7. A fast identification method for a double-parameter time domain electromagnetic weak polarization effect is characterized by comprising the following steps:
zero scanning is carried out on the actually measured data to obtain a symbol inversion moment, and a late slope is extracted through nonlinear least square fitting;
taking the sign inversion time and the late slope as input, and taking 'whether polarization occurs' as output to establish a polarization effect rapid classification model based on a support vector machine;
the model is transplanted to a time domain electromagnetic receiving device, and the measured data is rapidly identified through a polarization effect rapid classification model.
8. The quick identification method according to claim 7,
the zero point scanning of the measured data to obtain the symbol inversion time comprises the following steps: the electromagnetic attenuation curve in the polarized medium is gradually reduced from a positive value to a negative value, the actually measured signal value is compared with a zero value, and when the curve is lower than zero for the first time, the current time value is recorded to obtain the sign reversal time;
extracting the late slope by nonlinear least squares fitting further comprises: and when the late slope is extracted, a polarization attenuation curve with the time period of 1ms is intercepted from the tail end of the curve, and a late slope value under a double-exponential coordinate is obtained by adopting nonlinear least square fitting.
9. The device is characterized by comprising an actual measurement signal reading unit, a zero point feature extraction unit, a synchronization unit and a main control unit, wherein the actual measurement signal reading unit, the zero point feature extraction unit, the synchronization unit and the main control unit are arranged in the device, and the main control unit is used for controlling the actual measurement signal reading unit and the zero point feature extraction unit to perform the zero point feature extraction on the
The actual measurement signal reading unit comprises a receiving coil front end amplifying circuit and is used for acquiring the actual measurement electromagnetic response in the field;
the zero point characteristic extraction unit is used for extracting a symbol reversal moment from an actual measurement electromagnetic response curve according to actual measurement electromagnetic response data acquired by the actual measurement signal reading unit;
the master control unit comprises a late slope characteristic extraction unit, a model establishment module and a rapid classification unit, and extracts the late slope of the electromagnetic response curve by adopting a least square fitting method through the late slope characteristic extraction unit;
the model establishing module is used for establishing a polarization effect rapid classification model based on a support vector machine by taking the sign reversal time and the late slope as input and taking 'whether polarization occurs' as output;
the rapid classification unit is used for rapidly identifying the double parameters with the electrodeless effect by using a pre-constructed polarization effect classification model;
and the synchronization unit adopts GPS synchronization, is directly connected with the main control unit and the zero feature extraction unit, and is used for realizing synchronization of the receiver and the transmitter and triggering a counter in the zero feature extraction unit.
10. The quick identifying device according to claim 9, wherein the zero point feature extracting unit includes a zero-crossing comparator, a counter, and an AD collector,
after receiving the synchronous signal, the AD collector starts sampling and converts the received voltage quantity into a digital quantity; and meanwhile, the counter starts counting, when the zero-crossing comparator detects the zero crossing point of the signal, the counter is triggered to stop counting, and the symbol inversion time is calculated according to the counting size of the counter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910998449.4A CN110907997B (en) | 2019-10-21 | 2019-10-21 | Rapid identification method and device for double-parameter time domain electromagnetic weak polarization effect |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910998449.4A CN110907997B (en) | 2019-10-21 | 2019-10-21 | Rapid identification method and device for double-parameter time domain electromagnetic weak polarization effect |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110907997A true CN110907997A (en) | 2020-03-24 |
| CN110907997B CN110907997B (en) | 2024-04-05 |
Family
ID=69815893
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910998449.4A Active CN110907997B (en) | 2019-10-21 | 2019-10-21 | Rapid identification method and device for double-parameter time domain electromagnetic weak polarization effect |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110907997B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5936734A (en) * | 1997-12-23 | 1999-08-10 | J.A. Woollam Co. Inc. | Analysis of partially polarized electromagnetic radiation in ellipsometer and polarimeter systems |
| EP1876473A1 (en) * | 2005-03-24 | 2008-01-09 | Obshchestvo S Ogranichennoy Otvetstvennostyu "SIBI | Electromagnetic sounding method using a transient field spatial derivation on several separations |
| RU2015148374A (en) * | 2015-11-10 | 2017-05-16 | Закрытое акционерное общество "Аэрогеофизическая разведка" | Method of geoelectrical exploration (options) |
| CN110133733A (en) * | 2019-04-28 | 2019-08-16 | 吉林大学 | A kind of conductance based on particle swarm optimization algorithm-polarizability multi-parameter imaging method |
| CN210639288U (en) * | 2019-10-21 | 2020-05-29 | 吉林大学 | Quick recognition device of time domain electromagnetism weak polarization effect |
-
2019
- 2019-10-21 CN CN201910998449.4A patent/CN110907997B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5936734A (en) * | 1997-12-23 | 1999-08-10 | J.A. Woollam Co. Inc. | Analysis of partially polarized electromagnetic radiation in ellipsometer and polarimeter systems |
| EP1876473A1 (en) * | 2005-03-24 | 2008-01-09 | Obshchestvo S Ogranichennoy Otvetstvennostyu "SIBI | Electromagnetic sounding method using a transient field spatial derivation on several separations |
| RU2015148374A (en) * | 2015-11-10 | 2017-05-16 | Закрытое акционерное общество "Аэрогеофизическая разведка" | Method of geoelectrical exploration (options) |
| CN110133733A (en) * | 2019-04-28 | 2019-08-16 | 吉林大学 | A kind of conductance based on particle swarm optimization algorithm-polarizability multi-parameter imaging method |
| CN210639288U (en) * | 2019-10-21 | 2020-05-29 | 吉林大学 | Quick recognition device of time domain electromagnetism weak polarization effect |
Non-Patent Citations (1)
| Title |
|---|
| 李永祯;程旭;李棉全;王雪松;: "极化信息在雷达目标检测中的得益分析", 现代雷达, no. 02, 15 February 2013 (2013-02-15) * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110907997B (en) | 2024-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108564565A (en) | A kind of power equipment infrared image multi-target orientation method based on deep learning | |
| CN107169469B (en) | Material identification method of MIMO radar based on machine learning | |
| CN101893704A (en) | Rough set-based radar radiation source signal identification method | |
| CN103778409A (en) | Human face identification method based on human face characteristic data mining and device | |
| CN107274679B (en) | Vehicle identification method, device, equipment and computer readable storage medium | |
| CN103824093A (en) | SAR (Synthetic Aperture Radar) image target characteristic extraction and identification method based on KFDA (Kernel Fisher Discriminant Analysis) and SVM (Support Vector Machine) | |
| CN110348494A (en) | A kind of human motion recognition method based on binary channels residual error neural network | |
| CN102158486A (en) | Method for rapidly detecting network invasion | |
| CN109858428A (en) | ANA flourescent sheet automatic identifying method based on machine learning and deep learning | |
| CN104125050B (en) | Ultrahigh frequency RFID card reader protocol conformance test method | |
| CN106338662A (en) | Transformer winding deformation diagnosis method based on mathematical morphology | |
| CN105300267A (en) | Metal plate thickness measuring method based on electromagnetic eddy current and data fusion algorithm | |
| CN102982345B (en) | Semi-automatic classification method for timing sequence remote sensing images based on continuous wavelet transforms | |
| CN114776304A (en) | Method, device, equipment and medium for identifying abnormal deep sea mineral areas | |
| CN108983296A (en) | A kind of transient electromagnetic and effect of induced polarization decoupling method | |
| CN110907997B (en) | Rapid identification method and device for double-parameter time domain electromagnetic weak polarization effect | |
| CN116486177A (en) | Underwater target identification and classification method based on deep learning | |
| CN116894223A (en) | Airborne underwater anomaly detection method based on self-adaptive cancellation and ResNet neural network | |
| CN104751630B (en) | Road traffic state acquisition method based on Kernel-KNN matching | |
| CN102750547A (en) | Fruit size grading method based on compressed sensing | |
| CN105842742A (en) | Specific metal detecting device and specific metal intelligent identification learning method | |
| CN104933858B (en) | A kind of road traffic state acquisition methods based on space communication characteristic Kernel KNN matchings | |
| CN108763742A (en) | A kind of strong electromagnetic pulse effect Vulnerability assessment method and device | |
| CN115016008A (en) | Electrical source induction-polarization symbiotic effect multi-parameter imaging method based on neural network | |
| CN109360289B (en) | Power meter detection method fusing inspection robot positioning information |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |