A Rock Magnetic Measurement Method Based on an Ultra-High Sensitivity Triaxial Magnetic Field Measurement Device

Abstract

A method for measuring rock magnetism based on an ultra-high sensitivity triaxial magnetic field measuring device is disclosed. This method combines linearly polarized light rotation angle detection with circularly polarized light absorption. The device utilizes two linearly polarized beams and one circularly polarized beam to construct an ultra-high sensitivity triaxial magnetic field measuring device, enabling triaxial magnetic field measurement of the rock sample. A sample delivery system controls the movement of the sample within the measurement area, recording the magnetic field response at each measurement point of the triaxial magnetic field measuring device during sample movement. Based on the measurement data, a machine learning model is trained using a stochastic gradient descent algorithm. The magnetic moment is determined by the weight parameters obtained from the model training, thus achieving the measurement of the rock sample's magnetism. This invention further utilizes a high-sensitivity magnetic field measuring device to construct a multi-channel sensor. Compared to traditional methods, the magnetic field sensitivity of this device can be increased to the order of fT. By placing the sample in a low remanence environment, it can measure the magnetism of substances that are difficult to detect with current instruments, such as lunar soil magnetism.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of quantum sensing, and particularly relates to a rock magnetic measurement method based on an ultrahigh-sensitivity three-axis magnetic field measurement device. BACKGROUND

[0002] The measurement of rock magnetism is crucial for paleomagnetism and rock magnetism research. The rock age can be inferred, the plate tectonics theory can be verified, the dynamics of the earth's magnetic field can be studied, and the flow direction of ore veins and lava flows can be estimated through the rock magnetism. The strength of the rock magnetism is mainly represented by the size of the magnetic moment of the rock. The current instruments for measuring the magnetic moment use magnetic sensors including an induction coil, a fluxgate sensor, a magneto-impedance sensor, and a superconducting quantum interference device (SQUID). The induction coil is commonly used in high-frequency vibrating sample magnetometers or rotating magnetometers. Its structure limits it to measuring large-volume samples. The sensitivity of the fluxgate sensor and the magneto-impedance sensor is not outstanding, and they cannot measure rock samples with extremely weak magnetism. In comparison, the SQUID is mainly used in high-precision magnetic material analysis field due to its high sensitivity and high bandwidth, and it is irreplaceable. However, the SQUID can only measure small-volume samples. When measuring large-volume samples, the samples need to be crushed, which damages the integrity of the samples. On the other hand, the SQUID needs to be equipped with a low-temperature cooling device, which results in a large volume and high measurement cost. Compared with the SQUID, the magnetic field measurement device based on the spin-exchange-relaxation-free (SERF) effect has the advantages of ultrahigh sensitivity and no need for low-temperature cooling. It has surpassed the SQUID and become the instrument with the highest sensitivity for magnetic field measurement, reaching the sensitivity of sub-fT level. In recent years, it has been gradually applied to the research in the field of paleomagnetism.

[0003] Related document CN202411466136 describes a magnetic moment comparator device, including a background magnetic field generating component, an atomic magnetometer for measuring absolute magnetic fields, a first current-carrying standard coil component, a magnetic sample, and a measurement module. This invention provides a magnetic moment comparator device and measurement method. Using the condition that the background magnetic field remains constant, the current flowing through the standard coil is changed in real time. The magnetic moment of the magnetic sample is then equal in magnitude and opposite in direction to the magnetic moment of the current-carrying standard coil, thus the magnetic moment of the magnetic sample is measured using the magnetic moment of the current-carrying standard coil. This invention uses an atomic magnetometer as the magnetic field sensor, with a range of 100 nT to 100,000 nT. To ensure that the atomic magnetometer can effectively respond to the weak magnetic field generated by the magnetic sample, a background magnetic field of 200 nT to 20,000 nT needs to be generated within a magnetic shielding cylinder using a background magnetic field coil, confining the measured magnetic field within the sensor's range. However, for ultra-weak magnetic samples such as rocks, the magnetic field at 50 mm is only on the order of about 0.1 nT. At this time, the background magnetic field set by the device is significantly higher than the sample signal intensity. The strong background magnetic field will completely drown out the extremely weak rock sample signal, making it difficult to accurately measure its magnetic moment information. Summary of the Invention

[0004] This invention addresses the deficiencies and shortcomings of existing technologies by proposing a method for measuring the magnetic properties of rocks based on an ultra-high sensitivity triaxial magnetic field measuring device. It employs a combination of linearly polarized light rotation angle detection and circularly polarized light absorption. The ultra-high sensitivity triaxial magnetic field measuring device is constructed using two linearly polarized beams and one circularly polarized beam to achieve triaxial magnetic field measurement of the sample. A sample delivery system controls the movement of the sample within the measurement area, recording the magnetic field response at each measurement point of the triaxial magnetic field measuring device during sample movement. Based on the measurement data, a machine learning model is trained using a stochastic gradient descent algorithm. The magnetic moment is determined through the weight parameters obtained from the model training, thereby achieving the measurement of the magnetic properties of the rock sample.

[0005] The technical solution of the present invention is as follows:

[0006] A method for measuring rock magnetism based on an ultra-high sensitivity triaxial magnetic field measuring device, characterized by comprising the following steps:

[0007] Step 1: Configure the ultra-high sensitivity triaxial magnetic field measuring device and the sample delivery system so that the rock sample to be tested can be located at each preset measuring point in the measurement area;

[0008] Step 2: The ultra-high sensitivity triaxial magnetic field measuring device is used to perform triaxial magnetic field measurement by combining the two-beam linearly polarized light rotation angle detection method with the one-beam circularly polarized light absorption method. The first beam of linearly polarized light passes through the gas cell along the Y-axis, the second beam of linearly polarized light passes through the gas cell along the X-axis, and the third beam of circularly polarized light passes through the gas cell along the Z-axis.

[0009] Step 3: Use the sample delivery system to control the movement of the rock sample to be tested within the measurement area, and record the magnetic field response of each measurement point of the triaxial magnetic field measuring device during the movement of the rock sample to be tested;

[0010] Step 4: Based on the measurement data, a machine learning model is trained using the stochastic gradient descent algorithm. The magnetic moment is determined by the weight parameters obtained after the model training, thereby realizing the measurement of the magnetism of the rock sample.

[0011] Step 3 includes: moving the rock sample to be tested horizontally along the Y-axis using the sample delivery system, stopping the rock sample for 5 seconds at distances of 30mm, 32mm, 34mm, 36mm, 38mm, 40mm, 42mm, 44mm, 46mm, 48mm, 50mm, and 52mm from the gas chamber, and recording the response signal of the triaxial magnetic field measuring device after stabilization at each point. , , and These are the measurement points. The x-axis magnetic field, y-axis magnetic field and z-axis magnetic field will As the corresponding measurement point A set of measured data Each measurement yielded 12 sets of data.

[0012] Step 4 involves using the stochastic gradient descent algorithm to optimize the regression machine learning model based on the 12 sets of data obtained from each measurement, thus transforming the problem of solving the unknown magnetic moment into a machine learning problem.

[0013] Step 4 includes the following expression:

[0014] ,

[0015] ,

[0016] ,

[0017] ,

[0018] in In magnetic dipole coordinates The theoretical magnetic field generated along the x-axis In magnetic dipole coordinates The theoretical magnetic field generated along the y-axis In magnetic dipole coordinates The theoretical magnetic field generated along the z-axis The permeability of free space, , , respectively are the x-axis magnetic moment, y-axis magnetic moment, z-axis magnetic moment of the measured rock sample, R is the measurement distance, G is the gradient function, is the gradient of the mean square error is the gradient operator, is the three-axis theoretical magnetic field, is the three-axis measured magnetic field, n is the total number of measurement points, i is the sequence number of the measurement point, is the updated weight parameter, is the model weight parameter, and s is the learning rate.

[0019] Step 4 includes: 12 groups of data are optimized by the random gradient descent algorithm to complete the optimization of the model as a training cycle, and the training cycle is set to 500 rounds. Finally, the weight parameter of the model convergence state can obtain the magnetic moment of the measured rock sample, thereby realizing the measurement of the magnetism of the measured rock sample.

[0020] Step 2 includes obtaining the voltage value output by the Z-axis photoelectric acquisition module by using the third beam of circularly polarized light along the Z-axis passing through the gas chamber , obtaining the voltage value output by the X-axis photoelectric acquisition module by using the second beam of linearly polarized light along the X-axis passing through the gas chamber , and obtaining the voltage value output by the Y-axis photoelectric acquisition module by using the first beam of linearly polarized light along the Y-axis passing through the gas chamber .

[0021] In step 1, the ultra-high sensitivity three-axis magnetic field measuring device includes a beam splitter connected to the first laser, a first detection light barrel module, a first detection light barrel channel, a gas chamber, a first detection light barrel channel, a first detection light barrel channel, a first detection light barrel module and a first photoelectric acquisition module connected in sequence. The second split output end of the beam splitter is connected to the second detection light barrel module, the second detection light barrel channel, the gas chamber, the second detection light barrel channel, the second mirror, the second detection light barrel module, the second photoelectric acquisition module and the upper computer in sequence. The upper computer is connected to the sample feeding system through the sample feeding system driver, and the sample feeding system includes a dovetail slot sliding table and a rotating shaft connected to the motor, respectively. One end of the rotating shaft is connected to the dovetail slot sliding table, and the other end is connected to the sample groove and extends into the measurement area.

[0022] In step 1, the ultra-high sensitivity three-axis magnetic field measuring device includes a second laser, a pump light barrel module, a pump light barrel channel, a gas chamber, a pump light barrel channel, a third mirror and a third photoelectric acquisition module connected in sequence. The gas chamber is located in the oven, and the oven is located in the coil. The coil includes a Y-axis magnetic compensation coil, a Z-axis magnetic compensation coil and an X-axis magnetic compensation coil. The coil is located in an amorphous magnetic shielding barrel, which is located in a five-layer permalloy magnetic shielding barrel. The coil is connected to the signal generator. ​

[0023] The technical effects of this invention are as follows: This invention provides a method for measuring the magnetic properties of rocks based on an ultra-high sensitivity triaxial magnetic field measuring device. By employing a linear polarization optical rotation angle detection method combined with a circular polarization optical absorption method, an ultra-high sensitivity triaxial magnetic field measuring device is constructed based on two beams of linearly polarized light and one beam of circularly polarized light, enabling triaxial magnetic field measurement of the rock sample. A sample delivery system is used to control the movement of the sample within the measurement area, recording the magnetic field response at each measurement point of the triaxial magnetic field measuring device during sample movement. Based on the measurement data, a machine learning model is trained using a stochastic gradient descent algorithm. The magnetic moment is determined through the weight parameters obtained from the model training, thereby achieving the measurement of the magnetic properties of the rock sample.

[0024] Compared with the prior art, the features of the present invention are as follows:

[0025] 1. This invention proposes a method for measuring triaxial magnetic fields by combining linearly polarized light rotation angle detection with circularly polarized light absorption. This method has not been found in domestic patent publications.

[0026] 2. This invention proposes a method for measuring rock magnetism based on an ultra-high sensitivity triaxial magnetic field measuring device. Compared with the array magnetometer commonly used in existing multi-channel rock magnetism measurement schemes, this invention builds a multi-channel sensor using a high-sensitivity magnetic field measuring device. Compared with traditional methods, the magnetic field sensitivity of this device can be improved to the order of fT.

[0027] 3. This invention utilizes a high-sensitivity triaxial magnetic field measuring device to place the sample to be tested in a low remanent magnetic environment, which can measure the magnetic properties of substances that are difficult to detect with current instruments, such as lunar soil magnetism. Attached Figure Description

[0028] Fig. 1 This is a schematic diagram of the XY plane structure of the ultra-high sensitivity triaxial magnetic field measuring device with a sample delivery system involved in the rock magnetic field measuring method based on the ultra-high sensitivity triaxial magnetic field measuring device of the present invention.

[0029] Fig. 2 This is a schematic diagram of the YZ plane structure of the ultra-high sensitivity triaxial magnetic field measuring device involved in the rock magnetic field measuring method based on the ultra-high sensitivity triaxial magnetic field measuring device of the present invention.

[0030] The reference numerals in the attached diagram are explained as follows: 1-Five-layer permalloy magnetic shielding barrel; 2-First detection light entry channel; 3-First detection light entry module; 4-First laser; 5-Beam splitter; 6-Second detection light entry channel; 7-Second detection light entry module; 8-Signal generator; 9-Y-axis magnetic compensation coil; 10-Z-axis magnetic compensation coil; 11-X-axis magnetic compensation coil; 12-Amorphous magnetic shielding barrel; 13-Oven; 14-Gas chamber; 15-First detection light exit channel; 16-First reflector; 17-First detection light exit module; 18-First photoelectric acquisition module (i.e., Y-axis photoelectric acquisition module). ; 19-Second detection light exit channel; 20-Second reflector; 21-Second detection light exit module; 22-Second photoelectric acquisition module (i.e., x-axis photoelectric acquisition module); 23-Host computer; 24-Sample delivery system driver; 25-Motor; 26-Dovetail groove slide; 27-Rotation axis; 28-Sample slot; 29-Second laser; 30-Pump light into the barrel module; 31-Pump light into the barrel channel; 32-Pump light out of the barrel channel; 33-Third reflector; 34-Third photoelectric acquisition module (i.e., z-axis photoelectric acquisition module); XYZ-Cartesian coordinate system three axes (i.e., X-axis, Y-axis, and Z-axis). Detailed Implementation

[0031] The following is in conjunction with the attached diagram ( Figs. 1-2 The present invention will be described in conjunction with the embodiments.

[0032] Fig. 1 This is a schematic diagram of the XY plane structure of the ultra-high sensitivity triaxial magnetic field measuring device with a sample delivery system involved in the rock magnetic field measuring method based on the ultra-high sensitivity triaxial magnetic field measuring device of the present invention. Fig. 2 This is a schematic diagram of the YZ-plane structure of the ultra-high sensitivity triaxial magnetic field measuring device involved in the rock magnetic field measurement method based on the ultra-high sensitivity triaxial magnetic field measuring device of the present invention. (Reference) Figs. 1-2 As shown, a method for measuring the magnetism of rocks based on an ultra-high sensitivity triaxial magnetic field measuring device includes the following steps: Step 1, configuring the ultra-high sensitivity triaxial magnetic field measuring device and the sample delivery system so that the rock sample to be tested can be located at each preset measurement point in the measurement area; Step 2, using the ultra-high sensitivity triaxial magnetic field measuring device, a triaxial magnetic field measurement is performed by combining a two-beam linearly polarized light rotation angle detection method with a one-beam circularly polarized light absorption method. The first beam of linearly polarized light passes through the gas chamber along the Y-axis, the second beam of linearly polarized light passes through the gas chamber along the X-axis, and the third beam of circularly polarized light passes through the gas chamber along the Z-axis; Step 3, using the sample delivery system to control the movement of the rock sample to be tested within the measurement area, and recording the magnetic field response of each measurement point of the triaxial magnetic field measuring device during the movement of the rock sample; Step 4, using the measurement data, a machine learning model is trained using a stochastic gradient descent algorithm, and the magnetic moment is determined by the weight parameters obtained after model training, thereby realizing the measurement of the magnetism of the rock sample.

[0033] Step 3 includes: moving the rock sample to be tested horizontally along the Y-axis using the sample delivery system, stopping the rock sample for 5 seconds at distances of 30mm, 32mm, 34mm, 36mm, 38mm, 40mm, 42mm, 44mm, 46mm, 48mm, 50mm, and 52mm from the gas chamber, and recording the response signal of the triaxial magnetic field measuring device after stabilization at each point. , , and These are the measurement points. The x-axis magnetic field, y-axis magnetic field and z-axis magnetic field will As the corresponding measurement point A set of measured data Each measurement yields 12 sets of data. Step 4 involves using the stochastic gradient descent algorithm to optimize the regression machine learning model based on each of the 12 sets of data obtained from the measurements, transforming the problem of solving for the unknown magnetic moment into a machine learning problem.

[0034] Step 4 includes the following expression:

[0035] ,

[0036] ,

[0037] ,

[0038] ,

[0039] in In magnetic dipole coordinates The theoretical magnetic field generated along the x-axis In magnetic dipole coordinates The theoretical magnetic field generated along the y-axis In magnetic dipole coordinates The theoretical magnetic field generated along the z-axis The permeability of free space, , , These represent the x-axis, y-axis, and z-axis magnetic moments of the rock sample being tested, respectively; R is the measurement distance; and G is the gradient function. Mean square error gradient, It is a triaxial theoretical magnetic field. This is a triaxial measurement of the magnetic field, where n is the total number of measurement points and i is the index of each measurement point. It updates the weight parameters. is the model weight parameter, and s is the learning rate.

[0040] Step 4 includes: optimizing the model using the stochastic gradient descent algorithm with 12 sets of data as one training cycle, setting the training cycle to 500 rounds, and finally obtaining the magnetic moment of the rock sample by the weight parameters of the model's convergence state, thereby realizing the measurement of the magnetism of the rock sample.

[0041] Step 2 includes obtaining the voltage value output by the Z-axis photoelectric acquisition module using a third circularly polarized light beam passing through the gas chamber along the Z-axis. The voltage value output by the X-axis photoelectric acquisition module is obtained using the second beam of linearly polarized light passing through the gas chamber along the X-axis. The voltage value output by the Y-axis photoelectric acquisition module is obtained using the first beam of linearly polarized light passing through the gas chamber along the Y-axis. .

[0042] The ultra-high sensitivity triaxial magnetic field measuring device in step 1 includes a beam splitter 5 whose input end is connected to a first laser 4. The first beam splitting output end of the beam splitter 5 is sequentially connected to a first detection light inlet module 3, a first detection light inlet channel 2, a gas chamber 14, a first detection light outlet channel 15, a first reflector 16, a first detection light outlet module 17, and a first photoelectric acquisition module 18. The second beam splitting output end of the beam splitter 5 is sequentially connected to a second detection light inlet module 7, a second detection light inlet channel 6, a gas chamber 14, a second detection light outlet channel 19, a second reflector 20, a second detection light outlet module 21, a second photoelectric acquisition module 22, and a host computer 23. The host computer 23 is connected to a sample delivery system through a sample delivery system driver 24. The sample delivery system includes a dovetail slide 26 and a rotating shaft 27, which are respectively connected to a motor 25. One end of the rotating shaft 27 is connected to the dovetail slide 26, and the other end is connected to the sample slot 28 and extends into the measurement area.

[0043] The ultra-high sensitivity triaxial magnetic field measuring device in step 1 includes a second laser 29, a pump light entry module 30, a pump light entry channel 31, a gas chamber 14, a pump light exit channel 32, a third reflector 33, and a third photoelectric acquisition module 34 connected in sequence. The gas chamber 14 is located inside an oven 13, which is located inside a coil. The coil includes a Y-axis magnetic compensation coil 9, a Z-axis magnetic compensation coil 10, and an X-axis magnetic compensation coil 11. The coil is located inside an amorphous magnetic shielding barrel 12, which is located inside a five-layer permalloy magnetic shielding barrel 1. The coil is connected to a signal generator 8.

[0044] This invention relates to a method for measuring the magnetic properties of rocks based on an ultra-high sensitivity triaxial magnetic field measuring device. The method combines linearly polarized light rotation angle detection with circularly polarized light absorption. An ultra-high sensitivity triaxial magnetic field measuring device is constructed using two linearly polarized beams and one circularly polarized beam to achieve triaxial magnetic field measurement of the sample. A sample delivery system controls the movement of the sample within the measurement area, recording the magnetic field response at each measurement point of the triaxial magnetic field measuring device during sample movement. Based on the measurement data, a machine learning model is trained using a stochastic gradient descent algorithm. The magnetic moment is determined through the weight parameters obtained from the model training, thereby achieving the measurement of the magnetic properties of the rock sample.

[0045] Fig. 1The XY plane structure of this invention mainly includes a precision sample delivery system and a triaxial magnetic field measuring device. The precision sample delivery system consists of a dovetail slide 26, a motor 25, a rotating shaft 27, and a sample slot 28. The signal input terminal of the delivery system driver 24 is connected to the host computer 23, and the signal output terminal is connected to the motor 25 to drive the motor 25 to work. The motor 25, the rotating shaft 27, and the sample slot 28 are located above the dovetail slide 26. The sample to be measured is placed in the sample slot 28. The rotating stage and the sample slot 28 are controlled by the motor 25 to rotate and move. The triaxial magnetic field measuring device includes a five-layer permalloy magnetic shielding barrel 1, an amorphous magnetic shielding barrel 12, a signal generator 8, an X-axis magnetic compensation coil 11, a Y-axis magnetic compensation coil 9, a Z-axis magnetic compensation coil 10, an oven 13, an air chamber 14, a laser (first laser 4), a beam splitter 5, a first detection light inlet channel 2, a second detection light inlet channel 6, a first detection light outlet channel 15, a second detection light outlet channel 19, a first detection light inlet module 3, a second detection light inlet module 7, a first detection light outlet module 17, a second detection light outlet module 21, a first reflector 16, a second reflector 20, a first photoelectric acquisition module 18, and a second photoelectric acquisition module 21. Module 22 consists of a five-layer permalloy magnetic shielding barrel 1 on the outermost layer and an amorphous magnetic shielding barrel 12 inside the five-layer permalloy magnetic shielding barrel 1. Both the five-layer permalloy magnetic shielding barrel 1 and the amorphous magnetic shielding barrel 12 are used to shield against interference from external magnetic fields. After shielding and demagnetization, the residual magnetism at the center in the X, Y, and Z directions can all reach 0.3 nT. A triaxial magnetic compensation coil is located inside the amorphous magnetic shielding barrel 12. The coil constant for the X-axis is 16.2 nT / mA, the coil constant for the Y-axis is 16.16 nT / mA, and the coil constant for the Z-axis is 77.71 nT / mA. Driven by the signal generator 8, the coil can be used to further eliminate the residual magnetism inside the barrel, making the barrel reach a near-zero magnetic environment. An oven 13 is located inside the triaxial magnetic compensation coil and is used to provide high-temperature operating conditions for the air chamber 14, which is located in the center of the oven 13. The first beam of light from the laser, after passing through the beam splitter, enters the first detection light entry module 7. The light emitted from the first detection light entry module 7 enters the gas chamber 14 through the first detection light entry channel 2, and then enters the reflector (first reflector 16) through the first detection light exit channel 15. It is reflected and enters the first detection light exit module 17, where the first photoelectric acquisition module 18 collects the photoelectric information to achieve the detection of the optical rotation angle of the Y-axis polarized light. The other beam of light from the laser, after passing through the beam splitter 5, enters the second detection light entry module 7. The light emitted from the second detection light entry module 7 enters the gas chamber 14 through the second detection light entry channel 6, and then enters the reflector (second reflector 20) through the second detection light exit channel 19. It is reflected and enters the second detection light exit module 21, where the second photoelectric acquisition module 22 collects the photoelectric information to achieve the detection of the optical rotation angle of the X-axis polarized light. The host computer 23 is connected to the first photoelectric acquisition module 18 and the second photoelectric acquisition module 22 for real-time display and control.

[0046] Fig. 2 The present invention has an XY plane structure. The laser (second laser 29) is connected to the pump light inlet module 30. The light emitted from the pump light inlet module 30 enters the gas chamber 14 through the pump light inlet channel 31. After the alkali metal atoms inside the gas chamber 14 are polarized by the pump light to reach the SERF state, they can highly sensitively sense changes in the external magnetic field. The pump light emitted from the gas chamber 14 enters the reflector (third reflector 33) through the pump light outlet channel 32 and is reflected into the third photoelectric acquisition module 34 to collect photoelectric information, thereby realizing the detection of Z-axis circularly polarized light absorption.

[0047] The above steps completed the construction of the triaxial magnetic field measurement device. The sample to be measured was sequentially moved to preset measurement points using a precision sample delivery system. The magnetic field response signal of the triaxial magnetic field device at each point was recorded. Based on the collected magnetic field spatial distribution dataset, a machine learning model was trained using the Stochastic Gradient Descent (SGD) algorithm, transforming the magnetic moment calculation problem into a machine learning problem, ultimately determining the equivalent magnetic moment vector parameters of the sample. The specific measurement method is as follows:

[0048] The circularly polarized pump light along the Z-axis not only completes the polarization of atoms in the alkali metal gas cell, but its output intensity can also reflect the change in atomic polarizability projected along this direction. Therefore, the atomic spin precession signal is first detected by the circularly polarized light absorption method along the Z-axis, thereby achieving coarse compensation for the triaxial magnetic field. The magnetic field generated by the rock sample being tested is a static magnetic field. When the external magnetic field changes, the absorption of circularly polarized light by alkali metal atoms will change. This change is reflected in the detection signal of the photodetector at the end of the optical path. The output of the Z-axis magnetic field measurement device is as follows:

[0049] ,

[0050] ,

[0051] ,

[0052] ,

[0053] in, It is the voltage value output by the Z-axis photoelectric acquisition module. The optical depth for pumping light. The frequency of the pump light, The initial polarizability, , , These are dimensionless parameters concerning the X, Y, and Z axis magnetic fields. Electron gyromagnetic ratio, , , For the magnetic fields in the X, Y, and Z directions, R op R is the optical pump rate. rel denoted as the transverse relaxation rate of the atom. This indicates the initial voltage value of the photodetector; first, adjust the magnetic compensation coils in the X, Y, and Z directions to output the voltage along the Z axis. Adjust to the maximum to complete the coarse magnetic field compensation.

[0054] When an external magnetic field is input, the optical rotation angle of the linearly polarized light used as the detection light will change accordingly; when the detection light is along the X-direction, the output of the magnetic field measuring device in the X-direction is as follows:

[0055] ,

[0056] in, It is the voltage value output by the X-axis photoelectric acquisition module. denoted as the conversion coefficient of the photodetector. To detect the intensity of incident light in front of the gas cell, The modulation angle, To detect the optical depth of light. To detect the light frequency, This indicates the optical path length of the pump light interacting with the atoms. The atomic density of alkali metals, Let c represent the classical electron radius and c represent the speed of light. , These represent the oscillation intensities of the D1 and D2 lines of the alkali metal atoms, respectively. , These are the center frequencies of the D1 and D2 lines for alkali metal atoms, respectively. , These represent the pressure broadening of atomic spectral lines at lines D1 and D2, respectively. A triangular wave is used to scan the Y-axis magnetic field. Adjust to the minimum, then use a triangular wave to scan the Z-axis magnetic field, and continue adjusting Vout to near 0mV. At this point, the magnetic field compensation for the Y and Z axes is complete, and the compensated magnetic fields for the Y and Z axes are now: , .

[0057] When the detection light is along the Y direction, the output of the Y-direction magnetic field measuring device is as follows:

[0058] ,

[0059] in, This is the voltage value output by the Y-axis photoelectric acquisition module. A triangular wave is used to scan the X-axis magnetic field. Adjusted to near 0mV, the compensation magnetic field along the X-axis is now... .

[0060] The sample was moved horizontally along the Y-axis using a sample delivery system. The sample was stopped for 5 seconds at distances of 30mm, 32mm, 34mm, 36mm, 38mm, 40mm, 42mm, 44mm, 46mm, 48mm, 50mm, and 52mm from the gas chamber. The response signals of the triaxial magnetic field measuring device after stabilization at each point were recorded. ,Will As each measurement point corresponding Each measurement yields 12 sets of data. Based on the measured data, the stochastic gradient descent algorithm is used to optimize the regression machine learning model, transforming the problem of solving the unknown magnetic moment into a machine learning problem.

[0061] Assuming the dipole of the object being measured is located at the origin, the coordinates of the triaxial magnetic field measuring device relative to the magnetic dipole of the object being measured are: The theoretical formula for calculating the magnetic field generated by a magnetic dipole is:

[0062] ,

[0063] ,

[0064] in , , These are the theoretical magnetic fields generated by the three axes. The permeability of free space, , , R is the triaxial magnetic moment of the rock sample being tested, and R is the measurement distance.

[0065] Set the position coordinates of each measurement point As the input layer of the model, The target magnetic moment is measured as the output layer of the model. As weight parameters in the training process, the 12 sets of data are randomly divided into 3 batches, with each batch containing 4 sets. A batch of data is randomly selected to begin training the model, ensuring that the model does not get stuck in local optima and enhancing its generalization ability. First, a set of data is randomly generated within the specified magnetic moment range. As model weights, the gradient function G of the current weights is calculated. , That is, the theoretical magnetic field in three directions at this location. With actual measurement The gradient of the mean squared error (MSE) of the difference between the two values ​​is calculated as follows:

[0066] ,

[0067] Define a learning rate s, which is used in conjunction with the gradient function to update the weight parameters. The calculation is as follows:

[0068] ,

[0069] This process is repeated until the gradient function for the current set of data converges to its minimum value. Then, the same iterative process is repeated for the next set of data in the current batch. After the current batch of data is trained, the next batch of data is trained. When all four batches of data have been trained, that is, all 12 sets of data have been optimized by the stochastic gradient descent algorithm, this is called a training cycle. The training cycle is set to 500 rounds. Finally, the magnetic moment of the sample can be obtained by the weight parameters of the model's convergence state, thus realizing the measurement of the sample's magnetism.

[0070] Contents not described in detail in this specification are prior art known to those skilled in the art. It is hereby indicated that the above description is intended to help those skilled in the art understand this invention, but does not limit the scope of protection of this invention. Any equivalent substitutions, modifications, improvements, and / or simplifications of the above descriptions that do not depart from the essential content of this invention fall within the scope of protection of this invention.

Claims

1. A rock magnetic measurement method based on an ultra-high sensitivity three-axis magnetic field measurement device, characterized by, The method comprises the following steps: Step 1: configure the ultra-high sensitivity three-axis magnetic field measuring device with the sample feeding system, so that the rock sample to be measured can be positioned at the preset measurement points in the measurement area; Step 2: measure the three-axis magnetic field by using the ultra-high sensitivity three-axis magnetic field measuring device with two linearly polarized light optical rotation angle detection methods and one circularly polarized light absorption method, the first linearly polarized light passes through the gas chamber along the Y axis, the second linearly polarized light passes through the gas chamber along the X axis, and the third circularly polarized light passes through the gas chamber along the Z axis; Step 3: use the sample feeding system to control the movement of the rock sample to be measured in the measurement area, and record the magnetic field response of each measurement point of the three-axis magnetic field measuring device during the movement of the rock sample to be measured; Step 4: train the machine learning model based on the measurement data using the stochastic gradient descent algorithm, and determine the magnetic moment through the weight parameters of the model training to realize the measurement of the magnetism of the rock sample.

2. The rock magnetic measurement method based on the ultra-high sensitivity three-axis magnetic field measurement device according to claim 1, characterized in that, Step 3 includes: moving the rock sample to be tested horizontally along the Y-axis using the sample delivery system, stopping the rock sample for 5 seconds at distances of 30mm, 32mm, 34mm, 36mm, 38mm, 40mm, 42mm, 44mm, 46mm, 48mm, 50mm, and 52mm from the gas chamber, and recording the response signal of the triaxial magnetic field measuring device after stabilization at each point. , , and These are the measurement points. The x-axis magnetic field, y-axis magnetic field and z-axis magnetic field will As the corresponding measurement point A set of measured data Each measurement yielded 12 sets of data.

3. The rock magnetic measurement method based on the ultra-high sensitivity three-axis magnetic field measurement device according to claim 2, characterized in that, In step 4, 12 groups of data obtained by each measurement are used to optimize the regression type machine learning model by using the stochastic gradient descent algorithm, and the solution of the unknown magnetic moment is converted into a machine learning problem.

4. The rock magnetic measurement method based on the ultra-high sensitivity three-axis magnetic field measurement device according to claim 1, characterized in that, In step 4, the following expression is included: ， ， ， ， in In magnetic dipole coordinates The theoretical magnetic field generated along the x-axis In magnetic dipole coordinates The theoretical magnetic field generated along the y-axis In magnetic dipole coordinates The theoretical magnetic field generated along the z-axis The permeability of free space, , , These represent the x-axis, y-axis, and z-axis magnetic moments of the rock sample being tested, respectively; R is the measurement distance; and G is the gradient function. Mean square error gradient, It is the gradient operator. It is a triaxial theoretical magnetic field. This is a triaxial measurement of the magnetic field, where n is the total number of measurement points and i is the index of each measurement point. It updates the weight parameters. is the model weight parameter, and s is the learning rate.

5. The rock magnetic measurement method based on the ultra-high sensitivity three-axis magnetic field measurement device according to claim 3, characterized in that, In step 4, 12 groups of data are optimized by the stochastic gradient descent algorithm to complete the optimization of the model as a training cycle, and the training cycle is set to 500 rounds. Finally, the weight parameters of the model convergence state can obtain the magnetic moment of the rock sample to be measured, so as to realize the measurement of the magnetism of the rock sample to be measured.

6. The rock magnetic measurement method based on the ultra-high sensitivity three-axis magnetic field measurement device according to claim 1, characterized in that, The third beam of circularly polarized light along the Z axis through the gas cell is used to obtain the voltage value of the Z-axis photoelectric acquisition module output in step 2 The second beam of linearly polarized light along the X axis through the gas cell is used to obtain the voltage value of the X-axis photoelectric acquisition module output The first beam of linearly polarized light along the Y axis through the gas cell is used to obtain the voltage value of the Y-axis photoelectric acquisition module output .

7. The rock magnetic measurement method based on the ultra-high sensitivity three-axis magnetic field measurement device according to claim 1, characterized in that, In step 1, the ultra-high sensitivity three-axis magnetic field measuring device comprises a beam splitter connected with a first laser, a first detection light barrel module, a first detection light barrel channel, a gas chamber, a first detection light barrel channel, a first mirror, a first detection light barrel module and a first photoelectric collection module connected in sequence at the first split output end of the beam splitter, a second detection light barrel module, a second detection light barrel channel, a gas chamber, a second detection light barrel channel, a second mirror, a second detection light barrel module, a second photoelectric collection module and an upper computer connected in sequence at the second split output end of the beam splitter, and the upper computer is connected with a sample feeding system through a sample feeding system driver. The sample feeding system comprises a dovetail slot sliding table and a rotating shaft connected with a motor, respectively, one end of the rotating shaft is connected with the dovetail slot sliding table, and the other end is connected with a sample groove and extends into the measurement area.

8. The rock magnetic measurement method based on the ultra-high sensitivity three-axis magnetic field measurement device according to claim 1, characterized in that, In step 1, the ultra-high sensitivity three-axis magnetic field measuring device comprises a second laser, a pump light barrel module, a pump light barrel channel, a gas chamber, a pump light barrel channel, a third mirror and a third photoelectric collection module connected in sequence, the gas chamber is located in an oven, the oven is located in a coil, the coil comprises a Y-axis magnetic compensation coil, a Z-axis magnetic compensation coil and an X-axis magnetic compensation coil, the coil is located in an amorphous magnetic shielding barrel, the amorphous magnetic shielding barrel is located in a five-layer permalloy magnetic shielding barrel, and the coil is connected with a signal generator.

Images