Automatic magnetic compensation method for a weak magnetic sensor

Abstract

An automatic magnetic compensation method of a weak magnetic sensor, according to the light intensity signal value control compensation magnetic field intensity, search light intensity signal extreme point, namely the magnetic field zero point, so as to realize automatic magnetic compensation, including the design of the magnetic field compensation algorithm along the pumping beam direction according to the Fibonacci sequence method and the design of the plane magnetic field compensation algorithm perpendicular to the pumping beam direction according to the Nelder-Mead simplex method. The steps of the automatic magnetic compensation method are as follows: first, according to the Fibonacci sequence method, control the compensation magnetic field intensity along the pumping beam direction, and compensate the magnetic field in this direction to the error range; then, according to the Nelder-Mead simplex method, control the compensation magnetic field intensity perpendicular to the pumping beam direction, and compensate the magnetic field in this plane to the error range, and complete the three-axis automatic magnetic compensation. The algorithm is simple and efficient, which can not only improve the magnetic compensation speed of the weak magnetic sensor, but also improve the stability of the magnetic compensation speed, and can be used for array weak magnetic sensor application.

Description

Technical Field

[0001] This invention belongs to the field of magnetic compensation technology for weak magnetic sensors, specifically relating to an automatic magnetic compensation method for weak magnetic sensors based on the Fibonacci sequence method and the Nelder-Mead simplex method. Background Technology

[0002] SERF-based weak magnetic sensors utilize SERF atomic states (SERF, Spin-Exchange-Relaxation-Free), which significantly improves the spin relaxation time of atoms and enhances atomic spin coherence compared to traditional weak magnetic sensors. This greatly improves the measurement sensitivity of weak magnetic sensors, and they have been widely used in fields such as human heart and brain magnetic measurements and extremely weak magnetic analysis of materials.

[0003] The realization of the SERF state requires three important conditions: pump light, high temperature, and weak magnetic field. Methods to achieve a weak magnetic environment include passive magnetic shielding and active magnetic compensation. Passive magnetic shielding is achieved through a magnetically shielded room or barrel, but the residual magnetic field after shielding is around 10 nT and is affected by fluctuations in the spatial magnetic field environment, which is not conducive to precise measurement of extremely weak magnetic fields. There are many active magnetic compensation methods, with the most common being cross-modulation, parametric modulation, and sequential compensation. The principle of cross-modulation is as follows: the pump beam is along the Z-axis, and the detection beam is along the X-axis. First, a modulated magnetic field is applied along the Z-axis, and the corresponding frequency component is extracted by a lock-in amplifier to obtain the X-axis magnetic field information and compensate for the X-axis magnetic field. Then, a modulated magnetic field is applied along the X-axis to obtain the Z-axis magnetic field information and compensate for the Z-axis magnetic field. After the X-axis and Z-axis magnetic fields are compensated, the output of the photodetector can be approximated as the Y-axis magnetic field information, and the Y-axis magnetic field is compensated accordingly. Parametric modulation utilizes the parametric modulation of the Z-magnetic field to suppress noise associated with airflow through the oven. Sequential compensation includes two main steps: no-pump compensation and sequential compensation with a specific sequence. Pump-free compensation primarily compensates for the magnetic field perpendicular to the pump and probe beam directions. Sequential compensation first applies a modulated magnetic field along the pump beam direction, extracts the corresponding frequency signal using a lock-in amplifier, and achieves magnetic field compensation along the probe beam direction; then, it applies the same modulated magnetic field along the probe beam direction and extracts the corresponding frequency signal using another lock-in amplifier, achieving magnetic field compensation along the pump beam direction. Both methods are complex to operate and difficult to implement automatically.

[0004] In the medical context of measuring the magnetic signals of the human heart and brain, the traditional active magnetic compensation method is complex to operate and slow to compensate. Therefore, the search for a new and efficient automatic magnetic compensation method using weak magnetic sensors has become a research hotspot.

[0005] To improve the magnetic compensation speed, an automatic magnetic compensation method based on the Fibonacci sequence method and the Nelder-Mead simplex method is used to compensate the ambient magnetic field. This method is simple to operate; first, the magnetic field along the pump beam direction is compensated to within the error range using the Fibonacci sequence method, and then the planar magnetic field perpendicular to the pump beam direction is compensated to within the error range using the Nelder-Mead simplex method. The Fibonacci sequence method is an iterative solution method for minimizing optimization problems where the objective function is a single-valued function. The light intensity signal collected by the photodetector in the weak magnetic field sensor system is a discrete signal, and the Fibonacci sequence method only uses the objective function value f during the iterative solution process; therefore, the Fibonacci sequence method is more suitable than other one-dimensional extremum search methods. The Nelder-Mead simplex method is used to solve problems where the function f(x) has a minimum value. The problem involves searching for local minima within a given region. Compared to other two-dimensional extremum search methods, the Nelder-Mead simplex method offers a simpler search process while maintaining the same accuracy. Therefore, employing an automatic magnetic compensation method based on the Fibonacci sequence and the Nelder-Mead simplex method can both improve the speed and stability of magnetic compensation. Summary of the Invention

[0006] The technical problem solved by this invention is to overcome the shortcomings of existing technologies and address the issues of complex operation and slow magnetic compensation speed in current active magnetic compensation methods. This invention provides an automatic magnetic compensation method for weak magnetic sensors, a novel method based on the Fibonacci sequence method and the Nelder-Mead simplex method. This method can improve the magnetic compensation speed and stability, providing technical support for applications requiring a high-quality zero-magnetic environment, such as weak magnetic sensors.

[0007] The technical solution of the present invention is as follows:

[0008] An automatic magnetic compensation method for a weak magnetic sensor is characterized by comprising controlling the strength of the compensation magnetic field based on the light intensity signal value, and searching for the extreme point of the light intensity signal, i.e., the zero point of the magnetic field, to achieve automatic magnetic compensation. The automatic magnetic compensation is implemented as follows: first, a magnetic field compensation algorithm along the pump beam direction is executed according to the Fibonacci sequence method to compensate the magnetic field along the pump beam direction to the error range; then, a planar magnetic field compensation algorithm perpendicular to the pump beam direction is executed according to the Nelder-Mead simplex method to compensate the planar magnetic field perpendicular to the pump beam direction to the error range, thus completing the triaxial automatic magnetic compensation.

[0009] The magnetic field compensation algorithm along the pump beam direction includes the following steps:

[0010] Step A1: Given the initial interval [a, b] and MIL. MIL is a constant representing the minimum interval length, and both a and b are coordinate values.

[0011] Step A2: Calculate B ycl and B ycr , B ycl and B ycr are the coordinate points in the interval [B ycl , B ycr generated by compressing the interval [a, b] according to the symmetric compression principle.

[0012] Step A3: Determine whether B ycr - B ycl < MIL. If so, generate the final compensation magnetic field B yc =(B ycr + B ycl ) / 2, then end the magnetic field compensation in the Y - axis direction and enter the magnetic field compensation in the X - axis and Z - axis directions. If not, enter Step A4.

[0013] Step A4: Generate the compensation magnetic field B ycl , and collect the output PD(B ycl ) of the photodetector.

[0014] Step A5: Generate the compensation magnetic field B ycr , and collect the output PD(B ycr ) of the photodetector.

[0015] Step A6: Determine whether PD(B ycl )> PD(B ycr ). If so, re - assign values to a, B ycl , B ycr in sequence and then return to Step A3. If not, re - assign values to b, B ycr , B ycl in sequence and then return to Step A3.

[0016] In Step A2, it includes:

[0017]

[0018]

[0019] where ρ1 represents the first element of the compression coefficient ρ k in, k is the element serial number, F k represents the k - th element in the Fibonacci sequence, N is the integer part of the ratio of the initial interval length to the minimum interval length, and ε is an extremely small positive number.

[0020] The algorithm flow of the plane magnetic field compensation perpendicular to the pumping beam direction includes the following steps:

[0021] Step 1: Select B0, B1, B2, Minx, and Minz, where Minx and Minz are given constants, and B0, B1, and B2 are the initial three points;

[0022] Step 2: Generate compensation magnetic fields B0, B1, and B2 sequentially, and collect the corresponding photodetector outputs PD(B0), PD(B1), and PD(B2).

[0023] Step 3: Sort PD(B0), PD(B1), and PD(B2) from largest to smallest, and define the points as B in order. l B nl B s ;

[0024] Step 4, according to B l B nl B s Calculate Maxx, Maxz, where Maxx is three points B. l B nl B s The maximum distance between the x-axis coordinates, Maxz, is the distance between the three points B. l B nl B s The maximum distance between the Z-axis coordinates;

[0025] Step 5: Determine if Maxx > Minx or Maxz > Min2. If not, generate the final compensation magnetic field B. l Then end the X-axis and Z-axis magnetic field compensation. If so, proceed to step 6.

[0026] Step 6, calculate center point B m ;

[0027] Step 7, calculate reflection point B r ;

[0028] Step 8, generate compensation magnetic field B r And collect the output PD(B) of the photodetector r );

[0029] Step 9, determine if it is PD(B) nl )≤PD(B r )≤PD(B l If so, then B r Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4; otherwise, proceed to step 10.

[0030] Step 10, determine if PD(B) is present. r )>PD(B l If yes, then return to step 4 after steps 10a to 10c; otherwise, proceed to step 11. Step 10a: Calculate the extension point B. e Step 10b: Generate the compensation magnetic field B e and collect PD(B) e Step 10c, determine if PD(B) is present. e )>PD(B r If so, then B e Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4; if not, then B. r Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4;

[0031] Step 11, determine if PD(B) is present. s ) <PD(B r ) <PD(B nl If so, then calculate the internal contraction point B. ic Then proceed to step 12. If not, calculate the external contraction point B. oc Then proceed to step 12;

[0032] Step 12, generate compensation magnetic field B ic Or B oc And collect the output PD(B) of the photodetector c );

[0033] Step 13, determine if it is PD(B) c )≥PD(B s If so, then B c Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4; otherwise, proceed to step 14.

[0034] Step 14: Generate three compensation magnetic fields sequentially, collect three PD values, compare the three PD values, and redefine three points B. l B nl B s Then return to step 4.

[0035] Includes the following steps:

[0036] (1) Based on the single-beam weak magnetic sensor model, in the XYZ coordinate system, the direction along the pump beam is defined as the Y direction, and a mathematical model of the light intensity signal and the triaxial magnetic field is established, which is expressed as:

[0037]

[0038] Where PD is the absorbed light intensity signal of the photodetector; k is the proportionality coefficient; s is the photon polarization of the pump light, which is equal to 1 for circularly polarized light; R op R is the pumping rate; rel B is the relaxation rate; x The magnetic field is along the X-axis; B y The magnetic field is along the Y-axis; B z γ is the magnetic field along the Z-axis; e The electron gyromagnetic ratio is denoted by C; C0 is a constant.

[0039] The light intensity signal carries a triaxial magnetic field B x B y and B z Information. Assume B x and B z Unchanged, for B y Find the partial derivative, expressed as:

[0040]

[0041] Where, P0 = sR op / (R op +R rel );

[0042] In B x and B z When B is not zero y When <0, When B y When = 0, When B y When >0,

[0043] When B y When PD = 0, PD has a minimum value. That is, when PD reaches the minimum value, the magnetic field in the Y-axis direction becomes zero. Therefore, the magnetic field compensation problem along the pump beam direction can be transformed into a one-dimensional minimum point search problem.

[0044] Assume B y =0, PD and B x and B z The relationship can be represented as:

[0045]

[0046] When B x =B z When PD = 0, PD has a maximum value. That is, when PD reaches a maximum value, the magnetic field in the X-axis and Z-axis directions becomes zero. Therefore, the planar magnetic field compensation problem perpendicular to the pump beam direction can be transformed into a two-dimensional maximum point search problem.

[0047] (2) Using the Fibonacci sequence method, the magnetic field along the pump beam direction is compensated to within the error range; the remaining magnetic field interval [a,b] along the pump beam direction, i.e., the Y-axis direction, is selected, and the interval [a,b] is used to generate the interval [B] according to the principle of symmetric compression. ycl B ycr ], compressibility coefficient ρ k From the Fibonacci sequence F k Generate; Y-axis coil outputs B sequentially ycl B ycr Record the corresponding photodetector output PD(B) ycl ), PD(B ycr By comparing PD(B) ycl ) and PD(B ycr The magnitude of the light intensity signal determines the interval where the minimum point of the light intensity signal, i.e., the zero point of the magnetic field, is located. For B ycl and B ycr Reassign; after multiple loops, when [B ycl B ycr When the interval length is less than the given parameter MIL, the Y-axis coil outputs the final compensation magnetic field B. yc That is, B ycl and B ycr The mean value of the magnetic field along the pump beam direction is compensated to the error range;

[0048] (3) The planar magnetic field perpendicular to the pump beam direction is simultaneously compensated using the Nelder-Mead simplex method, and is brought within the error range; the planar magnetic field is decomposed into the X-axis magnetic field B. x and Z-axis magnetic field B z (B) x B z To construct the planar magnetic field coordinate points, select three initial coordinate points B0, B1, and B2 to construct an initial simplex. The X-axis and Z-axis coils output B0, B1, and B2 sequentially, and record PD(B0), PD(B1), and PD(B2) sequentially. Sort these values ​​by size, and denote the point corresponding to the maximum value as B. l The point corresponding to the second largest value is denoted as B. nl The point corresponding to the minimum value is denoted as B. s B l B nl B sThe maximum distance between the X-axis coordinates is denoted as Maxx, and the maximum distance between the Z-axis coordinates is denoted as Maxz; B l and B nl The center point is denoted as B. m B s and B m The lines form the search direction. Reflection, extension, outward contraction, and inward contraction are performed along this direction to obtain the corresponding coordinate points. The light intensity signal value corresponding to the magnetic field value is recorded and compared with PD(B). l ), PD(B nl ), PD(B s The comparison is performed, replacing the point with the smaller light intensity signal value to construct a new simplex; after multiple iterations, when Maxx is less than the given parameter Minx or Maxz is less than the given parameter Minz, the X-axis and Z-axis coils output the final compensation magnetic field B. l The planar magnetic field perpendicular to the pump beam direction is compensated to the error range.

[0049] The technical effects of this invention are as follows: The automatic magnetic compensation method for a weak magnetic sensor is a non-modulation compensation method, which is more convenient to operate than the modulation compensation method and can realize automatic magnetic compensation; in addition, compared with the existing non-modulation method, the speed of searching for the zero point of the triaxial magnetic field, i.e. the extreme point of the light intensity signal, is faster, which effectively improves the magnetic compensation speed and improves the stability of the magnetic compensation speed.

[0050] The principle of this invention is as follows: This invention provides an automatic magnetic compensation method for weak magnetic sensors, which is based on the Fibonacci sequence method and the Nelder-Mead simplex method and is used for automatic magnetic compensation of weak magnetic sensors requiring high-quality zero-magnetic environments. This method utilizes the light intensity signal PD received by the photodetector and the triaxial magnetic field B... x B y and B z Relationship: B y When PD = 0, it has a minimum value. Adjusting the current in the Y-axis compensation coil, when PD reaches its minimum, the magnetic field in the Y-axis direction becomes zero. Therefore, the magnetic field compensation problem along the pump beam direction can be transformed into a one-dimensional minimum point search problem, and the magnetic field compensation can be completed using the Fibonacci sequence method; when B... x =B z When PD = 0, PD has a maximum value, that is, the current magnitude of the X-axis and Z-axis compensation coils is adjusted simultaneously. When PD reaches its maximum value, the magnetic field in the X-axis and Z-axis directions becomes zero. Therefore, the problem of compensating for the planar magnetic field perpendicular to the pump beam direction can be transformed into a two-dimensional maximum point search problem. The Nelder-Mead simplex method is used to complete the magnetic field compensation, realizing automatic magnetic compensation of the weak magnetic sensor. Attached Figure Description

[0051] Figure 1 A block diagram of an automatic magnetic compensation control system model related to the implementation of an automatic magnetic compensation method for a weak magnetic sensor according to the present invention. Figure 1 The system includes a mathematical model 1 for the light intensity signal and triaxial magnetic field of a single-beam magnetic weakening sensor, a magnetic field compensation algorithm 2 along the pump beam direction, and a planar magnetic field compensation algorithm 3 perpendicular to the pump beam direction. The mathematical model 1 for the light intensity signal and triaxial magnetic field of the single-beam magnetic weakening sensor includes a PD(B) model. x B y B z ), PD is the absorbed light intensity signal of the photodetector, B x For the magnetic field in the X-axis direction, B y For the magnetic field along the Y-axis, B z The magnetic field is along the Z-axis. The magnetic field compensation algorithm part 2 along the pump beam direction is based on the Fibonacci sequence method, controlling the compensation magnetic field strength along the pump beam direction according to the light intensity signal output by the photodetector, and searching for the minimum point of the light intensity signal, i.e., the zero point of the magnetic field. The planar magnetic field compensation algorithm part 3 perpendicular to the pump beam direction is based on the Nelder-Mead simplex method, controlling the planar compensation magnetic field strength perpendicular to the pump beam direction according to the light intensity signal output by the photodetector, and searching for the maximum point of the light intensity signal, i.e., the zero point of the planar magnetic field. B xc B zc B yc These are the X-axis coil output compensation magnetic field, Z-axis coil output compensation magnetic field, and Y-axis coil output compensation magnetic field, respectively. B xc B zc Enter part 1 from part 3, B yc Enter part 1 from part 2. PD enters parts 2 and 3 from part 1 respectively.

[0052] Figure 2 The graph shows the relationship between the partial derivative of the absorbed light intensity signal of the photodetector with respect to the magnetic field along the Y-axis and the response of the magnetic field along the Y-axis. Figure 2 The horizontal axis represents the magnetic field By(nT) in the Y-axis direction, with the horizontal axis coordinates being -15, -10, ..., 15. Figure 2 The vertical axis represents the partial derivative of the photodetector's absorption (PD) with respect to By (arb.units, relative units, with coordinates of -3, -2, ..., 3*10). -5 ). Figure 2 In B x and B z When B is not zero y When <0, When B y When = 0, When B y When >0,

[0053] When B y = 0, PD has a minimum value. That is, when PD reaches the minimum value, the magnetic field in the Y-axis direction becomes zero. Therefore, the magnetic field compensation problem along the pumping beam direction can be transformed into a one-dimensional minimum point search problem.

[0054] Figure 3 It is a response relationship diagram of the light intensity signal absorbed by the photodetector and the magnetic fields in the X-axis and Z-axis directions. Figure 3 It includes the magnetic field Bx in the X-axis direction (nT, coordinate points are -15, -10, ···, 15), the magnetic field Bz in the Z-axis direction (nT, coordinate points are -10, 0, ···, 10), and the light intensity absorbed by the photodetector PD (arb.units, relative unit, coordinate points are 0.008, 0.1, ···, 0.2). Figure 3 A maximum point appears on the curved surface. When B x = B z = 0, PD has a maximum value. That is, when PD reaches the maximum value, the magnetic fields in the X-axis and Z-axis directions become zero. Therefore, the magnetic field compensation problem for the plane perpendicular to the pumping beam direction can be transformed into a two-dimensional maximum point search problem.

[0055] Figure 4 It is a flowchart of the magnetic field compensation algorithm along the pumping beam direction involved in implementing an automatic magnetic compensation method for a weak magnetic sensor according to the present invention. Figure 4 It includes step 1, given an initial interval [a, b] and MIL, where MIL is a constant representing the minimum interval length, and both a and b are coordinate values; step 2, calculate B ycl and B ycr , B ycl and B ycr are the coordinate points in the interval [B ycl , B ycr generated by compressing the interval [a, b] according to the symmetric compression principle; step 3, judge whether B ycr - B ycl < MIL. If so, generate the final compensation magnetic field B yc = (B ycr + B ycl ) / 2, then end the magnetic field compensation in the Y-axis direction and enter the magnetic field compensation in the X-axis and Z-axis directions. If not, enter step 4; step 4, generate the compensation magnetic field B ycl , and collect the output PD (Bycl) of the photodetector; step 5, generate the compensation magnetic field B ycr , and collect the output PD (B ycr ) of the photodetector; step 6, judge whether PD (B ycl )> PD (B ycr ). If so, for a, B ycl , Bycr After reassigning values ​​sequentially, return to step 3. If not, then for b, B ycr B ycl After reassigning values ​​sequentially, return to step 3.

[0056] Figure 5 The flowchart illustrates the planar magnetic field compensation algorithm perpendicular to the pump beam direction, which is involved in implementing the automatic magnetic compensation method for a weak magnetic sensor according to the present invention. Figure 5 The process includes: Step 1, selecting B0, B1, B2, Minx, and Minz, where Minx and Minz are given constants, and B0, B1, and B2 are the initial three points; Step 2, sequentially generating compensation magnetic fields B0, B1, and B2, and acquiring the corresponding photodetector outputs PD(B0), PD(B1), and PD(B2); Step 3, sorting PD(B0), PD(B1), and PD(B2) from largest to smallest, and defining the points as follows: B... l B nl B s Step 4, according to B l B nl B s Calculate Maxx, Maxz, where Maxx is three points B. l B nl B s The maximum distance between the x-axis coordinates, Maxz, is the distance between the three points B. l B nl B s The maximum distance between the Z-axis coordinates; Step 5, determine whether Maxx > Minx or Maxz > Minz. If not, generate the final compensation magnetic field B. l After completing the X-axis and Z-axis magnetic field compensation, if so, proceed to step 6; Step 6: Calculate the center point B. m Step 7, calculate reflection point B r Step 8: Generate the compensation magnetic field B r And collect the output PD(B) of the photodetector r Step 9, determine if PD(B) is present. nl )≤PD(B r )≤PD(B l If so, then B r Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4; if not, proceed to step 10; in step 10, determine if PD(B) is correct. r )>PD(B lIf yes, then return to step 4 after steps 10a to 10c; otherwise, proceed to step 11. Step 10a: Calculate the extension point B. e Step 10b: Generate the compensation magnetic field B e and collect PD(B) e Step 10c, determine if PD(B) is present. e )>PD(B r If so, then B e Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4; if not, then B. r Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4; in step 11, determine whether PD(B) is active. s ) <PD(B r ) <PD(B nl If so, then calculate the internal contraction point B. ic Then proceed to step 12. If not, calculate the external contraction point B. oc Then proceed to step 12; in step 12, a compensating magnetic field B is generated. ic Or B oc And collect the output PD(B) of the photodetector c Step 13, determine if PD(B) is present. c )≥PD(B s If so, then B c Replace B s Compare the three PD values ​​and redefine the three points B. l B nl B s Then return to step 4; if not, proceed to step 14. In step 14, generate three compensation magnetic fields sequentially, collect three PD values, compare the three PD values, and redefine three points B. l B nl B s Then return to step 4. Detailed Implementation

[0057] The following is in conjunction with the attached diagram ( Figures 1-5 The invention will be described in the following sections and examples.

[0058] Figure 1 A block diagram of an automatic magnetic compensation control system model related to the implementation of an automatic magnetic compensation method for a weak magnetic sensor according to the present invention. Figure 2The graph shows the relationship between the partial derivative of the absorbed light intensity signal of the photodetector with respect to the magnetic field along the Y-axis and the response of the magnetic field along the Y-axis. Figure 3 The graph shows the relationship between the absorbed light intensity signal of the photodetector and the response of the magnetic fields along the X-axis and Z-axis. Figure 4 The flowchart illustrates the magnetic field compensation algorithm along the pump beam direction involved in implementing the automatic magnetic compensation method for a weak magnetic sensor according to the present invention. Figure 5 A flowchart illustrating the planar magnetic field compensation algorithm perpendicular to the pump beam direction, which is part of the automatic magnetic compensation method for a weak magnetic sensor according to the present invention. (Reference) Figures 1 to 5 As shown, an automatic magnetic compensation method for a weak magnetic sensor, based on the Fibonacci sequence method and the Nelder-Mead simplex method, includes the following steps:

[0059] (1) Based on the single-beam weak magnetic sensor model, in the XYZ coordinate system, the direction along the pump beam is defined as the Y direction, and a mathematical model of the light intensity signal and the triaxial magnetic field is established, which is expressed as:

[0060]

[0061] Where PD is the absorbed light intensity signal of the photodetector; k is the proportionality coefficient; s is the photon polarization of the pump light, which is equal to 1 for circularly polarized light; R op R is the pumping rate; rel B is the relaxation rate; x The magnetic field is along the X-axis; B y The magnetic field is along the Y-axis; B z γ is the magnetic field along the Z-axis; e The electron gyromagnetic ratio is denoted by C; C0 is a constant.

[0062] The light intensity signal carries a triaxial magnetic field B x B y and B z Information. Assume B x and B z Unchanged, for B y Find the partial derivative, expressed as:

[0063]

[0064] Where, P0 = sR op / (R op +R rel );

[0065] In B x and B z When B is not zero y When <0, When B y When = 0, When B y When >0,

[0066] When B y When PD = 0, PD has a minimum value. That is, when PD reaches the minimum value, the magnetic field in the Y-axis direction becomes zero. Therefore, the magnetic field compensation problem along the pump beam direction can be transformed into a one-dimensional minimum point search problem.

[0067] Assume B y =0, PD and B x and B z The relationship can be represented as:

[0068]

[0069] When B x =B z When PD = 0, PD has a maximum value. That is, when PD reaches its maximum value, the magnetic field in the X-axis and Z-axis directions becomes zero. Therefore, the planar magnetic field compensation problem perpendicular to the pump beam direction can be transformed into a two-dimensional maximum point search problem.

[0070] (2) Using the Fibonacci sequence method, the magnetic field along the pump beam direction is compensated to within the error range; the remaining magnetic field interval [a,b] along the pump beam direction, i.e., the Y-axis direction, is selected, and the interval [a,b] is used to generate the interval [B] according to the principle of symmetric compression. ycl B ycr ], compressibility coefficient ρ k From the Fibonacci sequence F k Generate; Y-axis coil outputs B sequentially ycl B ycr Record the corresponding photodetector output PD(B) ycl ), PD(B ycr By comparing PD(B) ycl ) and PD(B ycr The magnitude of the light intensity signal determines the interval where the minimum point of the light intensity signal, i.e., the zero point of the magnetic field, is located. For B ycl and B ycr Reassign; after multiple loops, when [B ycl B ycr When the interval length is less than the given parameter MIL, the Y-axis coil outputs the final compensation magnetic field B. yc That is, B ycl and B ycr The mean value of the magnetic field along the pump beam direction is compensated to the error range;

[0071] (3) The planar magnetic field perpendicular to the pump beam direction is simultaneously compensated using the Nelder-Mead simplex method, and is brought within the error range; the planar magnetic field is decomposed into the X-axis magnetic field B. x and Z-axis magnetic field B z (B)x B z To construct the planar magnetic field coordinate points, select three initial coordinate points B0, B1, and B2 to construct an initial simplex. The X-axis and Z-axis coils output B0, B1, and B2 sequentially, and record PD(B0), PD(B1), and PD(B2) sequentially. Sort these values ​​by size, and denote the point corresponding to the maximum value as B. l The point corresponding to the second largest value is denoted as B. nl The point corresponding to the minimum value is denoted as B. s B l B nl B s The maximum distance between the X-axis coordinates is denoted as Maxx, and the maximum distance between the Z-axis coordinates is denoted as Maxz; B l and B nl The center point is denoted as B. m B s and B m The lines form the search direction. Reflection, extension, outward contraction, and inward contraction are performed along this direction to obtain the corresponding coordinate points. The light intensity signal value corresponding to the magnetic field value is recorded and compared with PD(B). l ), PD(B nl ), PD(B s The comparison is performed, replacing the point with the smaller light intensity signal value to construct a new simplex; after multiple iterations, when Maxx is less than the given parameter Minx or Maxz is less than the given parameter Minz, the X-axis and Z-axis coils output the final compensation magnetic field B. l The planar magnetic field perpendicular to the pump beam direction is compensated to the error range.

[0072] The magnetic field compensation algorithm along the pump beam direction is designed based on the Fibonacci sequence method. The specific steps of this method are as follows:

[0073] (1) Given an initial interval [a, b] and MIL, where MIL is a constant representing the minimum interval length, calculate B. ycl and B ycr ;

[0074] B ycl and B ycr The formula for calculation is:

[0075]

[0076] Where ρ1 represents the compressibility coefficient ρ k The first element in;

[0077] Compression coefficient ρ k The formula for calculation is:

[0078]

[0079] Among them, F k represents the k-th element in the Fibonacci sequence, N is the integer ratio of the initial interval length to the minimum interval length, and ε is an extremely small positive number;

[0080] (2) Determine whether B ycr -B ycl <MIL holds. If it holds, drive the Y-axis coil through the current source to output the approximate solution B yc =(B ycr +B ycl ) / 2, end the magnetic field compensation in the Y-axis direction, and enter the magnetic field compensation in the X-axis and Z-axis directions. If it does not hold, enter the next step;

[0081] (3) Drive the Y-axis coil through the current source to output B ycl and B ycr in sequence, collect the output of the photodetector, and record it as PD(B ycl ) and PD(B ycr );

[0082] (4) Determine whether PD(B ycl )>PD(B ycr ) holds. If it holds, reassign values to a, B ycl , B ycr in sequence, and they are respectively expressed as:

[0083]

[0084] Among them, ρ k represents the k-th element in the compression coefficient;

[0085] If it does not hold, reassign values to b, B ycr , B ycl in sequence, and they are respectively expressed as:

[0086]

[0087] After the assignment, jump to step (2);

[0088] After the algorithm loop ends, the magnetic field along the pumping beam direction is compensated to within the error range.

[0089] The plane magnetic field compensation algorithm perpendicular to the pumping beam direction is designed based on the Nelder-Mead simplex method. The specific steps of this method are as follows:

[0090] (1) Given constants Minx and Minz, and select an initial 3 points to construct an initial simplex:

[0091] B i =B0 + λi e i = (B xci B zci ), i = 1, 2(8)

[0092] Where, λ i It is a positive number, and its value is determined by the specific size of the object being optimized; e i Represents a unit vector, which is a space The standard basis is given by e1 = (1,0) and e2 = (0,1).

[0093] (2) Drive the X-axis and Z-axis coils with a current source to output B0, B1, and B2 sequentially, and collect the outputs of the photodetector sequentially, recording them as PD(B0), PD(B1), and PD(B2). Sort PD(B0), PD(B1), and PD(B2) in ascending order. If multiple points have the same objective function value, assign a higher index to the newly generated point, and record the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s ;

[0094] (3) Calculate the three points B l B nl B s The maximum distance between the three points on the X-axis is denoted as Maxx. The maximum distance between the three points on the Z-axis is denoted as Maxz. It is then determined whether Maxx > Minx or Maxz > Minz. If not, the X-axis and Z-axis coils are driven by a current source to output B. l End the X-axis and Z-axis magnetic field compensation. If successful, proceed to the next step (4).

[0095] (4) Calculate the best point B l and the second best point B nl Center point B m Using the reflection coefficient ρ in B m worst point B in the direction s Perform reflection and calculate the reflection point B. r ;

[0096] B m and B r The formula for calculation is:

[0097]

[0098] Where ρ represents the reflection coefficient, ρ>0, and is generally taken as ρ=1;

[0099] The X-axis and Z-axis coils are driven by a current source to output B. rAnd collect the output of the photodetector and record it as PD(B) r ), determine PD(B) nl )≤PD(B r )≤PD(B l If true, use B. r Replace B s , PD(B r ), PD(B nl ), PD(B l Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s If the condition is not met, proceed to step (3). If the condition is not met, proceed to the next step (5).

[0100] (5) Determine PD(B) r )>PD(B l If the condition is true, proceed to the next step (6); if the condition is false, jump to step (7).

[0101] (6) Calculate the extension point B e , represented as:

[0102] B e =B g +χ(B r -B g (10)

[0103] Where χ represents the extension coefficient, χ>1, and is generally taken as χ=2;

[0104] The X-axis and Z-axis coils are driven by a current source to output B. e And collect the output of the photodetector and record it as PD(B) e ), determine PD(B) e )>PD(B r If true, use B. e Replace B s , PD(B e ), PD(B nl ), PD(B l Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s Jump to step (3). If it is not true, use B. r Replace B s , PD(B r ), PD(B nl ), PD(Bl Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s Proceed to step (3);

[0105] (7) Determine PD(B) s ) <PD(B r ) <PD(B nl If the condition is true, proceed to step (9); if not, proceed to the next step (8).

[0106] (8) For (B) r -B m Perform outward contraction and calculate the outward contraction point B. oc , represented as:

[0107] B oc =B m +γ(B r -B m (11)

[0108] Where γ represents the shrinkage coefficient, 0 < γ < 1, and is generally taken as γ = 0.5;

[0109] (9) Use B s Replace B r Perform inward contraction and calculate the inward contraction point B. ic , represented as:

[0110] B ic =B m +γ(B s -B m (12)

[0111] (10) Output B by driving the X-axis and Z-axis coils through a current source oc Or B ic And collect the output of the photodetector and record it as PD(B) c );

[0112] Determine PD(B) c )≥PD(B s If true, use B. c Replace B s , PD(B c ), PD(B nl ), PD(B l Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. sProceed to step (3);

[0113] If this is not true, only retain point B. l Replace all other points, with the following replacement relationship:

[0114] B i =B l +σ(B i -B l (13)

[0115] Where σ represents the compression coefficient, it is generally taken as σ = 0.5;

[0116] The X-axis and Z-axis coils are driven by a current source to output B. l B i The output of the photodetector is collected sequentially and recorded as PD(B) l ),PD(B i ). PD(B l ), PD(B i Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s Proceed to step (3);

[0117] After the algorithm loop ends, the magnetic field perpendicular to the pump beam direction is compensated to the error range, completing the triaxial automatic magnetic compensation of the weak magnetic sensor.

[0118] Specific implementation method one: See Figure 1 This embodiment describes an automatic magnetic compensation method for a weak magnetic sensor based on the Fibonacci sequence method and the Nelder-Mead simplex method. It includes a mathematical model part 1 of the light intensity signal of a single-beam weak magnetic sensor and a three-axis magnetic field, a magnetic field compensation algorithm part 2 along the pump beam direction, and a planar magnetic field compensation algorithm part 3 perpendicular to the pump beam direction.

[0119] The mathematical model 1 of the single-beam weak magnetic sensor light intensity signal and the triaxial magnetic field includes PD(B) xx B y B z ).

[0120] The magnetic field compensation algorithm part 2 along the pump beam direction is based on the Fibonacci sequence method. It controls the compensation magnetic field strength along the pump beam direction based on the light intensity signal output by the photodetector, and searches for the minimum point of the light intensity signal, i.e., the zero point of the magnetic field. The remaining magnetic field interval [a,b] along the pump beam direction, i.e., the Y-axis direction, is selected, and the interval [a,b] is compressed according to the principle of symmetry to generate the interval [B]. ycl B ycr], compressibility coefficient ρ k From the Fibonacci sequence F k Generate; Y-axis coil outputs B sequentially ycl B ycr Record the corresponding photodetector output PD(B) ycl ), PD(B ycr By comparing PD(B) ycl ) and PD(B ycr The magnitude of the light intensity signal determines the interval where the minimum point of the light intensity signal, i.e., the zero point of the magnetic field, is located. For B ycl and B ycr Reassign; after multiple loops, when [B ycl B ycr When the interval length is less than the given parameter MIL, the Y-axis coil outputs the final compensation magnetic field B. yc That is, B ycl and B ycr The mean value of the magnetic field along the pump beam direction is compensated to be within the error range.

[0121] The planar magnetic field compensation algorithm part 3, perpendicular to the pump beam direction, is based on the Nelder-Mead simplex method. It controls the strength of the planar compensation magnetic field perpendicular to the pump beam direction based on the light intensity signal output by the photodetector, and searches for the maximum point of the light intensity signal, i.e., the zero point of the planar magnetic field. This planar magnetic field is decomposed into the X-axis magnetic field B. x and Z-axis magnetic field B z (B) x B z To construct the planar magnetic field coordinate points, select three initial coordinate points B0, B1, and B2 to construct an initial simplex. The X-axis and Z-axis coils output B0, B1, and B2 sequentially, and record PD(B0), PD(B1), and PD(B2) sequentially. Sort these values ​​by size, and denote the point corresponding to the maximum value as B. l The point corresponding to the second largest value is denoted as B. nl The point corresponding to the minimum value is denoted as B. s B l B nl B s The maximum distance between the X-axis coordinates is denoted as Maxx, and the maximum distance between the Z-axis coordinates is denoted as Maxz; B l and B nl The center point is denoted as B. m B s and B m The lines form the search direction. Reflection, extension, outward contraction, and inward contraction are performed along this direction to obtain the corresponding coordinate points. The light intensity signal value corresponding to the magnetic field value is recorded and compared with PD(B). l ), PD(B nl ), PD(Bs The comparison is performed, replacing the point with the smaller light intensity signal value to construct a new simplex; after multiple iterations, when Maxx is less than the given parameter Minx or Maxz is less than the given parameter Minz, the X-axis and Z-axis coils output the final compensation magnetic field B. l The planar magnetic field perpendicular to the pump beam direction is compensated to the error range.

[0122] Specific Implementation Method Two: See Figure 1 , Figure 2 This embodiment further defines the automatic magnetic compensation method for a weak magnetic sensor based on the Fibonacci sequence method and the Nelder-Mead simplex method described in Specific Embodiment 1. The single-beam weak magnetic sensor's light intensity signal and the triaxial magnetic field mathematical model part 1 include PD(B x B y B z ).

[0123] In the XYZ coordinate system, PD(B x B y B z ) is represented as:

[0124]

[0125] Where PD is the absorbed light intensity signal of the photodetector; k is the proportionality coefficient; s is the photon polarization of the pump light, which is equal to 1 for circularly polarized light; R op R is the pumping rate; rel B is the relaxation rate; x The magnetic field is along the X-axis; B y The magnetic field is along the Y-axis; B z γ is the magnetic field along the Z-axis; e The electron gyromagnetic ratio is denoted by C; C0 is a constant.

[0126] The light intensity signal carries a triaxial magnetic field B x B y and B z Information. Assume B x and B z Unchanged, for B y Find the partial derivative, expressed as:

[0127]

[0128] Where, P0 = sR op / (R op +R rel );

[0129] like Figure 2 As shown, in B x and Bz When B is not zero y When <0, When B y When = 0, When B y When >0, When B y When PD = 0, PD has a minimum value. That is, when PD reaches the minimum value, the magnetic field in the Y-axis direction becomes zero. Therefore, the magnetic field compensation problem along the pump beam direction can be transformed into a one-dimensional minimum point search problem.

[0130] Assume B y =0, PD and B xx and B z The relationship can be represented as:

[0131]

[0132] like Figure 3 As shown, when B x =B z When PD = 0, PD has a maximum value. That is, when PD reaches its maximum value, the magnetic field in the X-axis and Z-axis directions becomes zero. Therefore, the planar magnetic field compensation problem perpendicular to the pump beam direction can be transformed into a two-dimensional maximum point search problem.

[0133] Specific implementation method three: See Figure 1 , Figure 4 This embodiment further defines the automatic magnetic compensation method for a weak magnetic sensor based on the Fibonacci sequence method and the Nelder-Mead simplex method described in Specific Embodiment 1. The magnetic field compensation algorithm part 2 along the pump beam direction includes the following steps:

[0134] Step 1: Given an initial interval [a, b] and MIL (where MIL is a constant representing the minimum interval length), calculate B. ycl and B ycr ;

[0135] B ycl and B ycr The formula for calculation is:

[0136]

[0137] Where ρ1 represents the compressibility coefficient ρ k The first element in;

[0138] Compression coefficient ρ k The formula for calculation is:

[0139]

[0140] Among them, Fk represents the k-th element in the Fibonacci sequence, N is the integer ratio of the initial interval length to the minimum interval length, and ε is an extremely small positive number;

[0141] Step 2, determine B ycr -B ycl Whether <MIL holds. If it holds, drive the Y-axis coil through the current source to output the approximate solution B yc =(B ycr +B ycl ) / 2, end the magnetic field compensation in the Y-axis direction, and enter the magnetic field compensation in the X-axis and Z-axis directions. If it does not hold, enter the next step;

[0142] Step 3, drive the Y-axis coil through the current source to sequentially output B ycl and B ycr , collect the output of the photodetector, and record it as PD(B ycl ) and PD(B ycr );

[0143] Step 4, determine whether PD(B ycl )>PD(B ycr ) holds. If it holds, reassign a, B ycl , B ycr sequentially, and they are respectively represented as:

[0144]

[0145] where ρ k represents the k-th element in the compression coefficient;

[0146] If it does not hold, reassign b, B ycr , B ycl sequentially, and they are respectively represented as:

[0147]

[0148] After the assignment, jump to Step 2;

[0149] After the algorithm loop ends, the magnetic field along the pumping beam direction is compensated within the error range.

[0150] Specific Embodiment 4: Refer to Figure 1 、 Figure 5 to illustrate this embodiment. This embodiment further limits the automatic magnetic compensation method of the weak magnetic sensor based on the Fibonacci sequence method and the Nelder-Mead simplex method described in Specific Embodiment 1. The plane magnetic field compensation algorithm part 3 perpendicular to the pumping beam direction includes the following steps:

[0151] Step 1: Given constants Minx and Minz, and selecting three initial points, construct the initial simplex:

[0152] B i =B0+λ i e i = (B xci B zci ), i = 1, 2 (34)

[0153] Where, λ i It is a positive number, and its value is determined by the specific size of the object being optimized; e i Represents a unit vector, which is a space The standard basis is given by e1 = (1,0) and e2 = (0,1).

[0154] Step two: Drive the X-axis and Z-axis coils with a current source to output B0, B1, and B2 sequentially, and then sequentially acquire the outputs of the photodetector, recording them as PD(B0), PD(B1), and PD(B2). Sort PD(B0), PD(B1), and PD(B2) in ascending order. If multiple points have the same objective function value, assign a higher index to the newly generated point, and record the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s ;

[0155] Step 3, calculate the three points B l B nl B s The maximum distance between the three points on the X-axis is denoted as Maxx. The maximum distance between the three points on the Z-axis is denoted as Maxz. It is then determined whether Maxx > Minx or Maxz > Minz. If not, the X-axis and Z-axis coils are driven by a current source to output B. l End the X-axis and Z-axis magnetic field compensation. If successful, proceed to the next step, step four.

[0156] Step 4: Calculate the best point B l and the second best point B nl Center point B m Using the reflection coefficient ρ in B m worst point B in the direction s Perform reflection and calculate the reflection point B. r ;

[0157] B m and B r The formula for calculation is:

[0158]

[0159] Where ρ represents the reflection coefficient, ρ>0, and is generally taken as ρ=1;

[0160] The X-axis and Z-axis coils are driven by a current source to output B. r And collect the output of the photodetector and record it as PD(B) r ), determine PD(B) nl )≤PD(B r )≤PD(B l If true, use B. r Replace B s , PD(B r ), PD(B nl ), PD(B l Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s If the condition is not met, proceed to step three. If not, proceed to step five.

[0161] Step 5, determine PD(B) r )>PD(B l If the condition is true, proceed to step six; otherwise, proceed to step seven.

[0162] Step 6, calculate extension point B e , represented as:

[0163] B e =B g +χ(B r -B g (36)

[0164] Where χ represents the extension coefficient, χ>1, and is generally taken as χ=2;

[0165] The X-axis and Z-axis coils are driven by a current source to output B. e And collect the output of the photodetector and record it as PD(B) e ), determine PD(B) e )>PD(B r If true, use B. e Replace B s , PD(B e ), PD(B nl ), PD(B l Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s Proceed to step three. If this is not successful, use B. rReplace B s , PD(B r ), PD(B nl ), PD(B l Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s Proceed to step three;

[0166] Step 7, determine PD(B) s ) <PD(B r ) <PD(B nl If the condition is true, proceed to step nine; otherwise, proceed to step eight.

[0167] Step 8, for (B) r -B m Perform outward contraction and calculate the outward contraction point B. oc , represented as:

[0168] B oc =B m +γ(B r -B m (37)

[0169] Where γ represents the shrinkage coefficient, 0 < γ < 1, and is generally taken as γ = 0.5;

[0170] Step nine, use B s Replace B r Perform inward contraction and calculate the inward contraction point B. ic , represented as:

[0171] B ic =B m +γ(B s -B m (38)

[0172] Step 10: Drive the X-axis and Z-axis coils to output B using a current source. oc Or B ic And collect the output of the photodetector and record it as PD(B) c );

[0173] Determine PD(B) c )≥PD(B s If true, use B. c Replace B s , PD(B c ), PD(B nl ), PD(B l Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B.l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s Proceed to step three;

[0174] If this is not true, only retain point B. l Replace all other points, with the following replacement relationship:

[0175] B i =B l +σ(B i -B l (39)

[0176] Where σ represents the compression coefficient, it is generally taken as σ = 0.5;

[0177] The X-axis and Z-axis coils are driven by a current source to output B. l B i The output of the photodetector is collected sequentially and recorded as PD(B) l ),PD(B i ). PD(B l ), PD(B i Sort the data from smallest to largest, and denote the point corresponding to the maximum value as B. l The second largest value is denoted as point B. nl The point corresponding to the minimum value is denoted as point B. s Proceed to step three;

[0178] After the algorithm loop ends, the magnetic field perpendicular to the pump beam direction is compensated to the error range, completing the triaxial automatic magnetic compensation of the weak magnetic sensor.

[0179] This invention can serve as a novel automatic magnetic compensation method for weak magnetic sensors. The method is simple, greatly improves the magnetic compensation speed and stability, and provides a high-quality zero-magnetic environment for the rapid and stable realization of the SERF state of weak magnetic sensors.

[0180] 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 method for automatic magnetic compensation of a field weakening sensor, characterized in that The automatic magnetic compensation is realized by controlling the compensation magnetic field intensity according to the light intensity signal value and searching for the light intensity signal extreme point, i.e. the magnetic field zero point, and the automatic magnetic compensation is realized in the following manner: firstly, a magnetic field compensation algorithm flow along the direction of the pumping light beam is executed according to the Fibonacci sequence method, and the magnetic field along the direction of the pumping light beam is compensated to within an error range; and secondly, a planar magnetic field compensation algorithm flow perpendicular to the direction of the pumping light beam is executed according to the Nelder-Mead simplex method, and the planar magnetic field perpendicular to the direction of the pumping light beam is compensated to within an error range, thereby completing three-axis automatic magnetic compensation. The magnetic field compensation algorithm flow along the direction of the pumping light beam comprises the following steps: Step A1, given initial interval and MIL, MIL is a constant, representing the minimum interval length, a and b are coordinate values; Step A2, Calculate and , and It is an interval Generate intervals according to the principle of symmetrical compression. The coordinates of the points in the middle; Step A3, judging whether , if yes, generating final compensation magnetic field end Y-axis direction magnetic field compensation, enter X-axis and Z-axis direction magnetic field compensation, if no, enter Step A4; Step A4, generating a compensating magnetic field , collecting photodetector output ; Step A5, generating a compensating magnetic field , collecting photodetector output ; Step A6, determine if If yes, reassign to , , in turn and return to step A3. If no, reassign to , , in turn and return to step A3.

2. The method of automatic magnetic compensation of a field weakening sensor according to claim 1, characterized in that The step A2 comprises the following steps: wherein, represents the compression coefficient represents the first element in the Fibonacci sequence, k is the element number, represents the kth element in the Fibonacci sequence, N is the integer of the ratio of the initial interval length to the minimum interval length, is a very small positive number.

3. The method of automatic magnetic compensation of a field weakening sensor according to claim 1, characterized in that The planar magnetic field compensation algorithm flow perpendicular to the direction of the pumping light beam comprises the following steps: Step 1, Selection , , Minx, Minz, where Minx and Minz are both given constants, , , are the initial 3 points; Step 2, the compensation magnetic field is generated in sequence , , , and the corresponding photodetector output is collected , , ; Step 3, , , In descending order, the definition points are in turn , , ; Step 4, according to , , , calculate , , is the maximum distance between the X-axis coordinates of the three points , , , is the maximum distance between the Z-axis coordinates of the three points , , ; Step 5, determine if or if not, generate final compensation magnetic field end X and Z axis magnetic field compensation, if yes, go to Step 6; Step 6, calculating the center point ; Step 7, calculating the reflection point ; Step 8, generating a compensating magnetic field and collecting the photodetector output ; Step 9, determine if , if yes, then instead of , compare 3 PD values, redefine 3 points , , go back to step 4, if no, go to step 10; Step 10, determine if If yes, then return to step 4 after steps 10a to 10c; otherwise, proceed to step 11. Step 10a: Calculate the extension point. Step 10b: Generate a compensation magnetic field. and collect Step 10c, determine whether If so, then replace Compare the three PD values ​​and redefine the three points. , , Then return to step 4; otherwise, replace Compare the three PD values ​​and redefine the three points. , , Then return to step 4; Step 11, determine if , if yes, calculate inner inflection point and go to Step 12, if no, calculate outer inflection point and go to Step 12; Step 12, generating a compensating magnetic field or and collecting the photodetector output ; Step 13, judge whether , if yes, then instead of , compare 3 PD values, redefine 3 points , , go back to step 4, if not, go to step 14; Step 14, three compensation magnetic fields are generated in turn, three PD values are collected, the three PD values are compared, and three points are redefined , , Return to step 4.

4. The method of automatic magnetic compensation of a field weakening sensor according to claim 1, characterized in that The planar magnetic field compensation algorithm flow perpendicular to the direction of the pumping light beam comprises the following steps: (1) Based on a single-beam weak magnetic sensor model, in an XYZ coordinate system, the direction along the pumping light beam is defined as the Y direction, a mathematical model of the light intensity signal and the three-axis magnetic field is established, and is expressed as: (1) wherein, is the absorption light intensity signal of the photodetector; is a proportionality coefficient; is the photon polarization of the pump light, equal to 1 for circularly polarized light; is the pump rate; is the relaxation rate; is the magnetic field in the X-axis direction; is the magnetic field in the Y-axis direction; is the magnetic field in the Z-axis direction; is the electron gyromagnetic ratio; is a constant; The light intensity signal carries information of the three-axis magnetic field 、 and Assuming and are constant, the partial derivative of is (2) wherein ; In and is not zero, when , ; when , ; when , ; when , has a minimum value, that is, when reaches the minimum value, the magnetic field in the Y-axis direction becomes zero, so the magnetic field compensation problem along the pumping light beam direction can be converted into a one-dimensional minimum point search problem; Assume , With And The relationship can be expressed as: (3) When time, has a maximum value, i.e. when reaches the maximum value, the magnetic field in the X-axis and Z-axis directions becomes zero, so the planar magnetic field compensation problem perpendicular to the pumping light beam direction can be converted into a two-dimensional maximum point search problem; (2) The magnetic field along the pumping beam direction is compensated to the error range by the Fibonacci sequence method; the residual magnetic field interval along the pumping beam direction, i.e. the Y-axis direction, is selected The interval is generated according to the symmetric compression principle The compression coefficient is generated by the Fibonacci sequence The Y-axis coil outputs , in turn, and the corresponding photodetector outputs , are recorded; the interval in which the minimum point of the light intensity signal, i.e. the magnetic field zero point, is determined by comparing the sizes of and , and and are revalued; after multiple cycles, when the interval length is less than the given parameter MIL, the Y-axis coil outputs the final compensation magnetic field , i.e. the average of and , and the magnetic field along the pumping beam direction is compensated to the error range. (3) By Nelder-Mead simplex method, the plane magnetic field perpendicular to the pumping beam direction is compensated at the same time, and is compensated to the error range; the plane magnetic field is decomposed into X-axis magnetic field and Z-axis magnetic field , constitute the plane magnetic field coordinate point, select three initial coordinate points , , , construct the initial simplex; the X-axis and Z-axis coils output in turn , , , and record in turn , , , sort them by size, the point corresponding to the maximum value is recorded as , the point corresponding to the second maximum value is recorded as , and the point corresponding to the minimum value is recorded as ; , , The maximum value of the distance between the X-axis coordinates of , and the maximum value of the distance between the Z-axis coordinates is recorded as ; and The center point is recorded as , and The connecting line constitutes the search direction, and the corresponding coordinate points are obtained by reflection, extension, outer contraction and inner contraction operations in this direction, and the light intensity signal values corresponding to the magnetic field values are recorded. Compare it with , , Replace the point with smaller light intensity signal value to construct a new simplex; after many cycles, when is less than the given parameter or is less than the given parameter , the X-axis and Z-axis coils output the final compensation magnetic field , and the plane magnetic field perpendicular to the pumping beam direction is compensated to the error range.

Images