阅读说明：本技术 一种磁场总场的补偿方法、装置、系统及存储介质 (Compensation method, device and system for total magnetic field and storage medium ) 是由 荣亮亮 代海宾 伍俊 邱隆清 宋正威 董慧 陶泉 裴易峰 谢晓明 于 2020-03-30 设计创作，主要内容包括：本申请公开了一种磁场总场的补偿方法、装置、系统及存储介质,通过超导三轴磁强计获取磁场在坐标系下的分量值,通过光泵传感器获取磁场的光泵总场值,对分量值进行校正,得到校正后的分量值。利用磁干扰补偿模型、校正后的分量值和光泵总场值得到方向余弦特征矩阵,通过惯导传感器获取姿态信息,基于姿态信息确定姿态特征矩阵,对方向余弦特征矩阵,姿态特征矩阵进行滤波处理,得到滤波后的方向余弦特征矩阵和滤波后的姿态特征矩阵。将滤波后的方向余弦特征矩阵和滤波后的姿态特征矩阵输入已训练好的磁场补偿模型,得到补偿的磁干扰值,根据光泵总场值和补偿的磁干扰值得到目标磁场总场值。如此,可以得到精度更高的磁场总场值。(The application discloses a compensation method, a device and a system of a magnetic field total field and a storage medium, wherein component values of the magnetic field under a coordinate system are obtained through a superconducting triaxial magnetometer, an optical pump total field value of the magnetic field is obtained through an optical pump sensor, and the component values are corrected to obtain corrected component values. The method comprises the steps of obtaining a direction cosine feature matrix by using a magnetic interference compensation model, corrected component values and an optical pump total field value, obtaining attitude information through an inertial navigation sensor, determining the attitude feature matrix based on the attitude information, and filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix. And inputting the filtered direction cosine characteristic matrix and the filtered attitude characteristic matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value, and obtaining a target magnetic field total field value according to the optical pump total field value and the compensated magnetic interference value. Thus, the total field value of the magnetic field with higher precision can be obtained.)

1. A method of compensating for a total field of a magnetic field, the method comprising:

acquiring component values of a magnetic field in a coordinate system through a superconducting triaxial magnetometer;

acquiring an optical pump total field value of the magnetic field through an optical pump sensor;

correcting the component value to obtain a corrected component value;

obtaining a direction cosine feature matrix by using a magnetic interference compensation model, the corrected component values and the optical pump total field value;

acquiring attitude information through an inertial navigation sensor, and determining an attitude characteristic matrix based on the attitude information;

filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix;

inputting the filtered direction cosine feature matrix and the filtered attitude feature matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value;

and obtaining a target magnetic field total field value according to the optical pump total field value and the compensated magnetic interference value.

2. The method of claim 1, further comprising the step of training the magnetic field compensation model;

the training obtains the magnetic field compensation model, including:

acquiring a sample data set, wherein the sample data set comprises a filtered direction cosine feature matrix, a filtered attitude feature matrix and reference magnetic interference corresponding to each sample magnetic field in a plurality of sample magnetic fields; the reference magnetic interference is obtained by filtering the total field value of the reference optical pump corresponding to the reference magnetic interference;

constructing a preset machine learning model, and determining the preset machine learning model as a current machine learning model;

based on the current machine learning model, performing magnetic interference prediction operation on the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field, and determining predicted magnetic interference corresponding to the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field;

determining a loss value based on the filtered direction cosine feature matrix corresponding to each sample magnetic field and the predicted magnetic interference and the reference magnetic interference corresponding to the filtered attitude feature matrix;

when the loss value is larger than a preset threshold value, performing back propagation based on the loss value, updating the current machine learning model to obtain an updated machine learning model, and re-determining the updated machine learning model as the current machine learning model; repeating the steps: based on the current machine learning model, performing magnetic interference prediction operation on the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field, and determining predicted magnetic interference corresponding to the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field;

when the loss value is less than or equal to the preset threshold value, determining the current machine learning model as the magnetic field compensation model.

3. The method according to claim 1, wherein said correcting said component value to obtain a corrected component value comprises:

and correcting the component values by utilizing a particle swarm optimization algorithm and the optical pump total field value to obtain corrected component values.

4. The method of claim 1, wherein said deriving a direction cosine feature matrix using a magnetic disturbance compensation model, said corrected component values and said total optical pump field value comprises:

obtaining a direction cosine angle under a carrier coordinate system according to the quotient of the corrected component value and the optical pump total field value;

determining permanent magnetic interference, inductive interference and eddy current interference based on the magnetic interference compensation model, the direction cosine angle and the total field value of the optical pump;

determining a total magnetic interference formula according to the magnetic field interference, the induction interference and the eddy current interference;

and extracting the direction cosine feature matrix from the total magnetic interference formula.

5. The method of claim 1, wherein the attitude information includes a roll angle, a pitch angle, and a yaw angle;

determining a pose feature matrix based on the pose information, comprising:

converting the roll angle, the pitch angle and the yaw angle into a first radian corresponding to the roll angle, a second radian corresponding to the pitch angle and a third radian corresponding to the yaw angle respectively;

solving the first radian, the second radian and the third radian to obtain a sine value and a cosine value corresponding to the first radian, a sine value and a cosine value corresponding to the second radian and a sine value and a cosine value corresponding to the third radian;

and generating the attitude characteristic matrix according to the sine value and the cosine value corresponding to the first radian, the sine value and the cosine value corresponding to the second radian and the sine value and the cosine value corresponding to the third radian.

6. The method according to claim 1, wherein the filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix, includes:

and filtering the direction cosine feature matrix and the attitude feature matrix by using a high-pass filter to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix.

7. The method of claim 1 or 2, wherein the magnetic field compensation model comprises a first submodel, a second submodel, and an output layer; the first sub-model corresponds to the filtered direction cosine feature matrix, and the second sub-model corresponds to the filtered attitude feature matrix;

the first sub-model and the second sub-model are arranged in parallel and are respectively connected with the output layer;

the first submodel and the second submodel have the same structure and respectively comprise a first hidden layer and a second hidden layer.

8. An arrangement for compensation of the total field of a magnetic field, the arrangement comprising:

the component value acquisition module is used for acquiring component values of the magnetic field in a coordinate system through the superconducting triaxial magnetometer;

the optical pump total field value acquisition module is used for acquiring the optical pump total field value of the magnetic field through an optical pump sensor;

the component value correction module is used for correcting the component value to obtain a corrected component value;

the characteristic matrix acquisition module is used for acquiring a direction cosine characteristic matrix by utilizing a magnetic interference compensation model, the corrected component values and the optical pump total field value; acquiring attitude information through an inertial navigation sensor, and determining an attitude characteristic matrix based on the attitude information;

the filtering module is used for filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix;

the magnetic field compensation determining module is used for inputting the filtered direction cosine feature matrix and the filtered attitude feature matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value;

and the total field value determining module is used for obtaining a target magnetic field total field value according to the optical pump total field value and the compensated magnetic interference value.

9. An apparatus comprising a processor and a memory, wherein at least one instruction or at least one program is stored in the memory, and wherein the at least one instruction or the at least one program is loaded by the processor and executes the method for compensating for a total field of a magnetic field according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that at least one instruction or at least one program is stored in the computer storage medium, which is loaded and executed by a processor to implement the method for compensating for the total field of a magnetic field according to any one of claims 1 to 7.

Technical Field

The present disclosure relates to the field of detection by an aviation magnetic field total field method, and in particular, to a method, an apparatus, a system, and a storage medium for compensating a magnetic field total field.

Background

The total field method detection of the aviation magnetic field has the advantages of high detection efficiency, wide detection area, relative cheapness and the like, and is one of the optimal choices of the magnetic method detection. However, in the actual flight detection process, since the sensor needs to be mounted on an aircraft, the magnetic substance existing in the aircraft carrier generates non-negligible magnetic interference on the magnetic sensor probe, and the magnetic interference around the sensor needs to be reasonably compensated in order to realize high-precision measurement.

At present, a magnetic interference compensation method in total field detection is based on a three-axis fluxgate modeling, wherein an optical pump realizes total field information acquisition, the three-axis fluxgate realizes carrier attitude information measurement, and algorithm compensation is carried out according to a Tolles-lawson linear compensation model proposed by Tolles in 1950 and a maneuvering scheme given by L eikak in 1961 so as to eliminate motion-related magnetic interference.

The problems of the total field compensation of the current aviation magnetic field are as follows: the existing magnetic compensation model mainly uses a fluxgate sensor to measure three components of a background magnetic field to solve direction cosine so as to perform magnetic compensation, and the fluxgate has high noise and a certain steering difference, so that the noise related to a motion attitude still exists in the total field output of the compensated optical pump.

Disclosure of Invention

The embodiment of the application provides a compensation method, a device and a system for a magnetic field total field and a storage medium, so that noise related to a motion attitude in the compensated magnetic field total field value is reduced and even removed, the compensation precision can be improved, and the magnetic field total field value with higher precision is obtained.

In one aspect, an embodiment of the present application provides a method for compensating a total field of a magnetic field, where the method includes:

acquiring component values of a magnetic field in a coordinate system through a superconducting triaxial magnetometer;

acquiring an optical pump total field value of a magnetic field through an optical pump sensor;

correcting the component value to obtain a corrected component value;

obtaining a direction cosine feature matrix by using the magnetic interference compensation model, the corrected component values and the total field value of the optical pump;

acquiring attitude information through an inertial navigation sensor, and determining an attitude characteristic matrix based on the attitude information;

filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix;

inputting the filtered direction cosine feature matrix and the filtered attitude feature matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value;

and obtaining the total field value of the target magnetic field according to the total field value of the optical pump and the compensated magnetic interference value.

In another aspect, an embodiment of the present application provides an apparatus for compensating a total field of a magnetic field, including:

the component value acquisition module is used for acquiring a component value of the magnetic field in a coordinate system through the superconducting triaxial magnetometer;

the optical pump total field value acquisition module is used for acquiring an optical pump total field value of a magnetic field through an optical pump sensor;

the component value correction module is used for correcting the component values to obtain corrected component values;

the characteristic matrix acquisition module is used for acquiring a direction cosine characteristic matrix by utilizing the magnetic interference compensation model, the corrected component values and the optical pump total field value; acquiring attitude information through an inertial navigation sensor, and determining an attitude characteristic matrix based on the attitude information;

the filtering module is used for filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix;

the magnetic field compensation determining module is used for inputting the filtered direction cosine feature matrix and the filtered attitude feature matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value;

and the total field value determining module is used for obtaining a target magnetic field total field value according to the optical pump total field value and the compensated magnetic interference value.

Another aspect provides a device, the terminal comprising a processor and a memory, the memory storing at least one instruction or at least one program, the at least one instruction or the at least one program being loaded by the processor and executed to implement the method for compensating for a total field of a magnetic field as described above.

Another aspect provides a computer-readable storage medium having at least one instruction or at least one program stored therein, the at least one instruction or the at least one program being loaded and executed by a processor to implement the method for compensating for a total field of a magnetic field as described above.

The compensation method, device, system and storage medium for the total magnetic field provided by the embodiment of the application have the following technical effects: acquiring component values of a magnetic field in a coordinate system through a superconducting triaxial magnetometer; acquiring an optical pump total field value of a magnetic field through an optical pump sensor; correcting the component value to obtain a corrected component value; obtaining a direction cosine characteristic matrix by using the magnetic interference compensation model, the corrected component values and the total field value of the optical pump; acquiring attitude information through an inertial navigation sensor, and determining an attitude characteristic matrix based on the attitude information; filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix; and inputting the filtered direction cosine characteristic matrix and the filtered attitude characteristic matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value, and obtaining a target magnetic field total field value according to the optical pump total field value and the compensated magnetic interference value. Therefore, the noise related to the motion attitude in the compensated magnetic field total field value can be reduced and even removed, so that the compensation precision can be improved, and the magnetic field total field value with higher precision can be obtained.

Drawings

To more clearly illustrate the technical solutions and advantages of the embodiments or the prior art of the present application, the drawings used in the description of the embodiments or the prior art are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without inventive efforts.

Fig. 1 is a schematic flowchart of a method for compensating a total field of a magnetic field according to an embodiment of the present disclosure;

fig. 2 is a schematic flowchart of a component value calibration method according to an embodiment of the present application;

fig. 3 is a flow chart illustrating a method for determining a direction cosine feature matrix according to an embodiment of the present application;

fig. 4 is a schematic flowchart of a method for determining an attitude feature matrix according to an embodiment of the present application;

FIG. 5 is a schematic structural diagram of a magnetic field compensation model provided in an embodiment of the present application;

FIG. 6 is a schematic flow chart of a training method of a magnetic field compensation model according to an embodiment of the present disclosure;

FIG. 7 is a schematic diagram of remanent magnetic interference provided by an embodiment of the present application;

FIG. 8 is a schematic structural diagram of a device for compensating for the total field of a magnetic field according to an embodiment of the present disclosure;

fig. 9 is a block diagram of a hardware structure of a server according to a method for compensating a total field of a magnetic field provided in an embodiment of the present application.

Detailed Description

The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

It should be noted that the terms "first," "second," and the like in the description and claims of this application and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the application described herein are capable of operation in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or server that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

While the following describes an embodiment of a method for compensating for a total magnetic field according to the present application, fig. 1 is a flow chart of a method for compensating for a total magnetic field according to an embodiment of the present application, and the present specification provides the method operation steps according to the embodiment or the flow chart, but more or less operation steps may be included based on conventional or non-inventive labor. The order of steps recited in the embodiments is merely one manner of performing the steps in a multitude of orders and does not represent the only order of execution. In practice, the system or server product may be implemented sequentially or in parallel (e.g., in the context of parallel processors or multi-threaded processing) according to the embodiments or methods shown in the figures. Specifically, as shown in fig. 1, the method may include:

in this embodiment of the present application, all the execution subjects of the implementation steps may be the same device, or may be executed by different devices, where the devices may be terminals, servers, or other computer devices.

S101: and acquiring component values of the magnetic field in a coordinate system through the superconducting triaxial magnetometer.

The component values in the coordinate system are measured by the superconducting triaxial magnetometer and have errors. The number of the component values is 3, and the component values can be regarded as three axial components of an X axis, a Y axis and a Z axis.

In this embodiment, the superconducting triaxial magnetometer in step S101 may be referred to as a superconducting Quantum Interference Device (SQUID), which is a known relatively sensitive magnetic sensor and is widely used for detecting weak magnetic signals. The superconducting quantum interference device is a superconducting loop that senses the magnetic field by sensing the magnetic flux induced in the superconducting loop. Because the superconducting loop of the superconducting quantum interference device has small area and the magnetic field in the loop is uniform, the magnetic flux measured by the superconducting quantum interference device is in direct proportion to the magnetic field intensity coupled with the superconducting loop of the superconducting quantum interference device. Therefore, the SQUID magnetometer is formed by using the superconducting quantum interference device superconducting loop plane coupling magnetic flux and the superconducting quantum interference device to measure the magnetic field. The magnetometer is a vector sensor and only measures the magnetic field in the normal direction of the SQUID superconducting ring plane.

In practical applications, to measure the total field of the magnetic field, three components of the magnetic field vector, i.e. three axial components of the X-axis, the Y-axis and the Z-axis, are measured, i.e. the component values in the coordinate system in step S101. The total field of the magnetic field is obtained by three axial components. To measure the total field of the magnetic field using the conventional SQUID magnetometers, three SQUID magnetometers are used, which are perpendicular to each other and are respectively defined as three perpendicular components of an X axis, a Y axis and a Z axis. The three-axis magnetometer measures the vector components of the vector magnetic field in three orthogonal directions XYZ, thereby obtaining a measure of the total field.

S103: and acquiring the total optical pump field value of the magnetic field through the optical pump sensor.

Alternatively, the optical pump total field value of the magnetic field obtained by the optical pump sensor is a reference value in this application. Compared with the component values under the coordinate system of the superconducting triaxial magnetometer, the total field value of the optical pump is closer to the true total field of the magnetic field, however, the total field value of the optical pump is not accurate enough due to the fact that the optical pump sensor still contains partial magnetic interference, and the requirement of the application cannot be met.

S105: and correcting the component values to obtain corrected component values.

Because the component values output by the superconducting triaxial magnetometer contain offsets in the coordinate system due to the characteristics of the components of the superconducting triaxial magnetometer, the component values need to be corrected to obtain corrected component values for use in subsequent steps.

In the embodiment of the application, the device may perform offset correction on the component value by using an optimization algorithm to obtain a corrected component value. The optimization algorithm may include a particle swarm optimization algorithm, a gradient descent algorithm, a newton method, a genetic algorithm, a simulated annealing algorithm, and the like. Next, the technical means of step S105 will be explained by taking a particle swarm optimization algorithm as an example.

In an optional implementation manner, the device may correct the component values by using a particle swarm optimization algorithm and an optical pump total field value to obtain corrected component values. Fig. 2 is a schematic flowchart of a component value correction method according to an embodiment of the present application, and as shown in fig. 2, the method includes:

s201: and initializing offset values corresponding to the three component values.

The number of component values is indicated above to be 3, therefore three component values may be defined as H_x， H_yAnd H_z. The initialization of the component values is to preset 3 offset values, O, corresponding to the 3 component values_x， O_yAnd O_z。

S202: and initializing the current speed and the current position of the particles in the particle swarm consisting of the offset value.

Wherein, the value range of the particles in the particle group can be-50 to 50, the number of the particles in the particle group can be set to be 20, and the speed range of the particles is-0.5 to 0.5.

In this step, the device may preset the current speed and the current position of each particle, where the current position and the current speed are within a value range of the particle and a speed range of the particle.

S203: and calculating the fitness of each particle under the current speed and the current position according to the total field value of the optical pump, and updating the current speed and the current position of each particle according to a preset rule.

Alternatively, the device may synthesize the mean square error of the total field and the optical pump by three axes, and the mean square root error is used as the fitness of the particle. As shown in equation (1):

where n is the total length of the data to be corrected in this segment, H_ix，H_iyAnd H_izRespectively, the output of the measured superconducting triaxial magnetometer, i.e. 3 component values, O_x，O_yAnd O_zIs 3 offset values; h_eThe total field value of the optical pump corresponding to the magnetic field, and the fitness is the fitness.

Specifically, the particle update speed and position method is performed according to the rule of formula (2):

x_i＝x_i-1+v_i

v_i＝w·v_i-1+c₁·rand()·(p_best-x_i-1)+c₂·rand()·(g_best-x_i-1)

wherein x is_iAnd x_i-1Respectively the current position after particle update and the current position before update, v_iAnd v_i-₁The current velocity after particle update and the current velocity before particle update, b_bestAnd g_bestIs the optimal position of the particle and the global optimal position, and rand () is a random number. In particular, w, c₁And c₂The inertia factor and the global and local search capabilities of the particle, respectively, need to be set in advance, which in this example may be set to 0.5, 0.2, respectively.

S204: if the update times meet the preset iteration times, go to step S205; otherwise, go to step 203.

If the update times of the current speed and the current position do not meet the preset iteration times, the step of 203 is repeated, the fitness of each particle under the current speed and the current position is calculated in an iterative mode, and then the current speed and the current position of each particle are continuously updated … …

S205: and calculating the fitness of each particle under the current speed and the current position by taking the total field value of the optical pump as a basis.

If the update times of the current speed and the current position satisfy the preset iteration times, for example, 1000 times, referring to formula (1) in step S203, the fitness of each particle at the current speed and the current position is obtained.

S206: and selecting the fitness with the minimum value from the fitness of each particle.

S207: and taking the offset corresponding to the fitness as a target offset.

Each particle has a fitness and an offset under the current iteration number, and step 206 and step S207 are to find the most suitable offset as the target offset.

S208: and determining corrected component values according to the target offset and the component values.

The target offset comprises 3 offsets, and the basic operation is performed on the component value and the target offset, for example, the 3 component values are respectively subtracted by the corresponding offsets to obtain the corrected component value B_x，B_yAnd B_z。

S107: and obtaining a direction cosine characteristic matrix by using the magnetic interference compensation model, the corrected component values and the total field value of the optical pump.

In an alternative embodiment, the magnetic interference compensation model may be a 16-term compensation model, and may also be a 21-term compensation model. Step S107 will be described below by taking the 21-way compensation model as an example.

The application introduces a method how to obtain a directional cosine feature matrix. Fig. 3 is a schematic flowchart of a method for determining a direction cosine feature matrix according to an embodiment of the present application, where as shown in fig. 3, the method includes:

s301: and obtaining the direction cosine angle under the carrier coordinate system according to the quotient of the corrected component value and the total field value of the optical pump.

The device may be based on the corrected component value B_x，B_yAnd B_zAnd total field value H of the optical pump_eSolving direction cosine angle u of magnetic field under carrier coordinate₁，u₂And u₃. The following formula (3):

s302: and determining permanent magnetic interference, inductive interference and eddy current interference based on the magnetic interference compensation model, the direction cosine angle and the total field value of the optical pump.

Optionally, a machine body coordinate system modeling is established by taking the mounting point of the optical pump probe as an origin, and magnetic interference is divided into permanent magnetic interference H according to the magnetic interference characteristic and the 21-term compensation model_pInduced interference H_iEddy current disturbance H_eddyAs shown in equation (4):

wherein p is_iIs a compensation coefficient corresponding to the permanent magnetic interference, a_ijIs a compensation coefficient corresponding to the induced interference, b_ijAnd c_ijIs a compensation coefficient, H ', corresponding to the eddy current disturbance'_eIs the total field value H of the optical pump_eU 'derived from time'_jIs u_jDerived from time.

S303: and determining a total magnetic interference formula according to the magnetic field interference, the induction interference and the eddy current interference.

The total magnetic interference can be composed of the three parts, and the expression of the total magnetic interference of the system is further deduced to be shown in the formula (5):

s304: and extracting a direction cosine feature matrix from the total magnetic interference formula.

Thus, the influence magnetic interference H can be extracted from the magnetic interference formula_podThe resulting 21 term feature matrix R is as in equation (6):

R＝[H_eu₁H_eu₂H_eu₃H_eu₁u₁H_eu₁u₂H_eu₁u₃H_eu₂u₂H_eu₂u₃H_eu₁u₁' H_eu₂u₁' H_eu₃u₁'H_eu₁u₂' H_eu₂u₂' H_eu₃u₂' H_eu₁u₃' H_eu₂u₃' H_e'u₁u₁H_e'u₁u₂H_e'u₁u₃H_e'u₂u₂H_e'u₂u₃]

in this way, the device can obtain a directional cosine feature matrix.

S109: attitude information is acquired through an inertial navigation sensor, and an attitude characteristic matrix is determined based on the attitude information.

In the embodiment of the present application, the attitude information includes a roll angle, a pitch angle, and a yaw angle.

Fig. 4 is a schematic flowchart of a method for determining an attitude feature matrix according to an embodiment of the present application, and as shown in fig. 4, the method includes:

s401: and respectively converting the roll angle, the pitch angle and the yaw angle into a first radian corresponding to the roll angle, a second radian corresponding to the pitch angle and a third radian corresponding to the yaw angle.

S402: and solving the first radian, the second radian and the third radian to obtain a sine value and a cosine value corresponding to the first radian, a sine value and a cosine value corresponding to the second radian and a sine value and a cosine value corresponding to the third radian.

S403: and generating an attitude characteristic matrix according to the sine value and the cosine value corresponding to the first radian, the sine value and the cosine value corresponding to the second radian and the sine value and the cosine value corresponding to the third radian.

In fact, according to the coordinate transformation formula, the direction cosine and the three attitude angles can be transformed with each other, and the direction cosine is a function formed by the sine and cosine of the attitude angles, but the formula cannot be completely expressed due to errors such as installation. Therefore, an attitude feature matrix influencing magnetic interference can be generated from the sine and cosine values obtained from the attitude angle, and nonlinear combination is performed by using a neural network to finally obtain 6 attitude feature matrices, wherein the attitude feature matrices can be shown as formula (7):

P＝[sin(r) cos(r) sin(p) cos(p) sin(h) cos(h)]

wherein r, p and h are respectively roll angle, pitch angle and yaw angle, sin (r) and cos (r) are respectively sine and cosine values corresponding to the roll angle, sin (p) and cos (p) are respectively sine and cosine values corresponding to the pitch angle, and sin (h) and cos (h) are respectively sine and cosine values corresponding to the yaw angle.

In summary, the device may obtain a direction cosine feature matrix and an attitude feature matrix.

S111: and filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix.

In an optional implementation manner of filtering processing, a high-pass filter may be used to perform filtering processing on the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix.

Specifically, 27 columns of signals of 21 directional cosine feature matrixes and 6 attitude feature matrixes are processedSpectral analysis, based on the spectrogram, respectively estimating the main frequency (f) of each signal in the range of attitude frequency (0.01-10Hz)₀,f₁...f₂₈) Selecting the smallest frequency value f from the 28 sets of frequency values_minAnd at 0.5f_minAnd (3) constructing a high-pass FIR filter with the frequency as the cut-off frequency, finally filtering 27 columns of signals of the cosine feature matrix and the attitude feature matrix through the constructed high-pass FIR filter, and storing the direction cosine feature matrix after filtering and the attitude feature matrix after filtering.

S113, inputting the filtered direction cosine feature matrix and the filtered attitude feature matrix into the trained magnetic field compensation model to obtain a compensated magnetic interference value.

In an alternative embodiment, as shown in fig. 5, the magnetic field compensation model may be a neural network model, a recursive network model, or the like, and the magnetic field compensation model may include a first sub-model, a second sub-model, and an output layer, where the first sub-model corresponds to the filtered direction cosine feature matrix, and the second sub-model corresponds to the filtered attitude feature matrix. Optionally, the first submodel and the second submodel are arranged in parallel and are connected to the output layer respectively. Optionally, the first submodel and the second submodel have the same structure and respectively include a first hidden layer and a second hidden layer.

Specifically, the magnetic field compensation model comprises an upper branch, namely a first submodel, wherein the first submodel mainly takes a filtered direction cosine feature matrix as an input, and the first submodel comprises a first hidden layer and a second hidden layer. Wherein the first hidden layer and the second hidden layer respectively comprise 64 neurons. The magnetic field compensation model further comprises a lower branch, namely a second submodel, wherein the second submodel mainly takes the filtered attitude characteristic matrix as input, and the second submodel comprises a first hidden layer and a second hidden layer. Wherein the first hidden layer and the second hidden layer respectively comprise 64 neurons. The first submodel and the second submodel are respectively connected to an output layer comprising a neuron, and the upper and lower branch outputs are connected to the output layer together to be summed to be a total output.

The activation functions of the input layer and the hidden layer adopt a Re L u function, and the expression is shown in formula (8):

the output layer adopts a linear activation function, and the expression is shown as the formula (9):

g(x)＝∑_ix_iw_i

in an alternative embodiment, the acquired sample data set may be used as a training sample, and training is performed to obtain the magnetic field compensation model. The embodiment of the present application further provides a training method of a magnetic field compensation model, as shown in fig. 6, including:

s601: acquiring a sample data set, wherein the sample data set comprises a filtered direction cosine feature matrix, a filtered attitude feature matrix and reference magnetic interference corresponding to each sample magnetic field in a plurality of sample magnetic fields; the reference magnetic interference is obtained by filtering the total field value of the reference optical pump corresponding to the reference magnetic interference.

For determining the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field in step S601, reference may be made to the contents corresponding to steps S101 to S111 in the foregoing, which is not described herein again.

The specific steps of acquiring the reference magnetic interference corresponding to each sample magnetic field are as follows: the optical pump total field value of each sample magnetic field is obtained through the optical pump sensor, and then the optical pump total field value of each sample magnetic field may be filtered by using the high-pass filter of step S111, so as to obtain the reference magnetic interference corresponding to each sample magnetic field.

The frequency of the magnetic interference related signal is in a fixed range, but the frequency of the signal actually measured by the sensor is complex. The following explains the reason why the directional cosine feature matrix and the attitude feature matrix need to be filtered: 1. the filtering is to filter out the signal component of the characteristic signal near the magnetic interference frequency and the fluctuation (direct current component) of the geomagnetic field in the optical pump sensor signal to extract the magnetic interference (i.e. the above reference magnetic interference) as much as possible, and then to perform fitting solution to improve the fitting accuracy. 2. If not, firstly, the output of the optical pump sensor comprises two parts of magnetic interference and the magnitude of the earth magnetic field, wherein the earth magnetic field 50uT is thousands times of the magnetic interference (nT level), and the filtering is to extract the magnetic interference instead of the output of the sum of the earth magnetic field and the magnetic interference.

S602: and constructing a preset machine learning model, and determining the preset machine learning model as the current machine learning model.

S603: and based on the current machine learning model, performing magnetic interference prediction operation on the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field, and determining predicted magnetic interference corresponding to the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field.

S604: and determining a loss value based on the filtered direction cosine feature matrix corresponding to each sample magnetic field and the predicted magnetic interference and the reference magnetic interference corresponding to the filtered attitude feature matrix.

The loss value may be a root mean square error, as shown in equation (10):

wherein, MSE represents the root mean square error, which is used as the loss value of the network. y is_iThe predicted magnetic interference is represented, and the magnetic interference output through the neural network training based on the 27 features is the predicted value. Bopm _ h_iThe reference magnetic interference is represented and is an actual value.

S605: and when the loss value is greater than the preset threshold value, the step is switched to the step S606, otherwise, the step is switched to the step S607.

S606: performing back propagation on the basis of the loss value, updating the current machine learning model to obtain an updated machine learning model, and determining the updated machine learning model as the current machine learning model again; go to step S603.

S607: and determining that the loss value is less than or equal to a preset threshold value, and determining the current machine learning model as a magnetic field compensation model.

In an alternative embodiment, the acquired sample data set may be divided into training samples and verification samples, and training is performed to obtain the magnetic field compensation model. The training sample is used for training to obtain a magnetic field compensation model, and the verification sample is used for verifying the accuracy of the magnetic field compensation model.

In order to reduce overfitting, a dropout optimization method is added into two hidden layers, part of neurons are abandoned according to certain probability to prevent overfitting, finally, different parameters are continuously adjusted, and corresponding parameters when a group of small MSEs are found out are final parameters (MSE is 0.000110, and the adam setting parameters are β)₁＝0.9，β₂＝0.99，∈＝10e^-11Dropout is 0.2), the neural network at this time is the final magnetic field compensation model, and the model is stored.

S115: and obtaining the total field value of the target magnetic field according to the total field value of the optical pump and the compensated magnetic interference value.

The device can obtain the compensated target magnetic field total field value according to the optical pump total field value and the magnetic interference value predicted by the compensation model, that is, the optical pump total field value and the compensated magnetic interference value are subjected to appropriate operation, such as addition or subtraction, to obtain the target magnetic field total field value, and then the magnetic compensation is completed.

In conclusion, in the embodiments of the present application, to better illustrate the advantages of the total field compensation method based on the combination of superconducting triaxial magnetometer and inertial navigation, the actual detection data can be obtained by building a magnetic detection system based on the superconducting triaxial magnetometer, the inertial navigation system and the optical pump sensor, and the actually measured data of the superconducting triaxial magnetometer, the inertial navigation system and the optical pump sensor can be substituted into the technical scheme of the above example for further compensation, and the compensation result is shown in fig. 7, the dotted line is residual magnetic interference before compensation after filtering output by the optical pump sensor, the dash-dot line is residual magnetic interference after compensation by a 16-term linear compensation model in the prior art, and the solid line is magnetic interference after compensation by the method, so that compared with the traditional compensation method, the novel method can further suppress motion-related noise. And statistics of the compensation accuracy and improvement ratio before and after compensation are shown in the following table (1):

from the table, the standard deviation (precision) after the compensation of the compensation method based on the combination of the superconducting triaxial magnetometer and the inertial navigation is improved from 0.22nT to 0.067pT, the improvement ratio is improved from 6.5 times to 21, and the effect is obvious.

In summary, the application provides a magnetic compensation method based on a combination of a superconducting triaxial magnetometer and inertial navigation, which can effectively solve the problem that motion-related magnetic interference still exists after compensation is performed only by using magnetic sensors such as a fluxgate as attitude information for modeling, so that high-precision measurement of a total field is realized. Therefore, the present invention effectively overcomes various disadvantages of the prior art and has a high industrial utility value.

An embodiment of the present application further provides a compensation apparatus for a total magnetic field, fig. 8 is a schematic structural diagram of the compensation apparatus for a total magnetic field provided in the embodiment of the present application, and as shown in fig. 8, the compensation apparatus includes:

the component value acquisition module 801 is used for acquiring component values of the magnetic field in a coordinate system through the superconducting triaxial magnetometer;

the optical pump total field value obtaining module 802 is configured to obtain an optical pump total field value of the magnetic field through the optical pump sensor;

the component value correcting module 803 is configured to correct the component value to obtain a corrected component value;

the feature matrix obtaining module 804 is configured to obtain a direction cosine feature matrix by using the magnetic interference compensation model, the corrected component values, and the optical pump total field value; acquiring attitude information through an inertial navigation sensor, and determining an attitude characteristic matrix based on the attitude information;

the filtering module 805 is configured to perform filtering processing on the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix;

the magnetic field compensation determining module 806 is configured to input the filtered direction cosine feature matrix and the filtered attitude feature matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value;

the total field value determining module 807 is configured to obtain a target magnetic field total field value according to the optical pump total field value and the compensated magnetic interference value.

In an optional embodiment, the apparatus further comprises a training module configured to:

acquiring a sample data set, wherein the sample data set comprises a filtered direction cosine feature matrix, a filtered attitude feature matrix and reference magnetic interference corresponding to each sample magnetic field in a plurality of sample magnetic fields; the reference magnetic interference is obtained by filtering the total field value of the reference optical pump corresponding to the reference magnetic interference;

constructing a preset machine learning model, and determining the preset machine learning model as a current machine learning model;

based on a current machine learning model, performing magnetic interference prediction operation on the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field, and determining predicted magnetic interference corresponding to the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field;

determining a loss value based on the filtered direction cosine feature matrix corresponding to each sample magnetic field and the predicted magnetic interference and the reference magnetic interference corresponding to the filtered attitude feature matrix;

when the loss value is larger than the preset threshold value, performing back propagation based on the loss value, updating the current machine learning model to obtain an updated machine learning model, and re-determining the updated machine learning model as the current machine learning model; repeating the steps: based on a current machine learning model, performing magnetic interference prediction operation on the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field, and determining predicted magnetic interference corresponding to the filtered direction cosine feature matrix and the filtered attitude feature matrix corresponding to each sample magnetic field;

and when the loss value is less than or equal to a preset threshold value, determining the current machine learning model as a magnetic field compensation model.

In an alternative embodiment, the apparatus further comprises:

the component value correction module is used for correcting the component values by utilizing a particle swarm optimization algorithm and an optical pump total field value to obtain corrected component values.

In an alternative embodiment, the apparatus further comprises:

the characteristic matrix acquisition module is used for acquiring a direction cosine angle under a carrier coordinate system according to the quotient of the corrected component value and the total field value of the optical pump;

determining permanent magnetic interference, inductive interference and eddy current interference based on a magnetic interference compensation model, a direction cosine angle and an optical pump total field value;

determining a total magnetic interference formula according to the magnetic field interference, the induction interference and the eddy current interference;

and extracting a direction cosine feature matrix from the total magnetic interference formula.

In an alternative embodiment, the apparatus further comprises:

the characteristic matrix acquisition module is used for converting the roll angle, the pitch angle and the yaw angle into a first radian corresponding to the roll angle, a second radian corresponding to the pitch angle and a third radian corresponding to the yaw angle respectively;

solving the first radian, the second radian and the third radian to obtain a sine value and a cosine value corresponding to the first radian, a sine value and a cosine value corresponding to the second radian and a sine value and a cosine value corresponding to the third radian;

and generating an attitude characteristic matrix according to the sine value and the cosine value corresponding to the first radian, the sine value and the cosine value corresponding to the second radian and the sine value and the cosine value corresponding to the third radian.

In an alternative embodiment, the apparatus further comprises:

the filtering module is used for filtering the direction cosine feature matrix and the attitude feature matrix by using a high-pass filter to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix.

The device and method embodiments in the embodiments of the present application are based on the same application concept.

For example, when running on a server, fig. 9 is a hardware block diagram of a server of a method for compensating for a magnetic field total field provided by an embodiment of the present invention, as shown in fig. 9, the server 900 may have a relatively large difference due to different configurations or performances, and may include one or more Central Processing Units (CPUs) 910 (the processor 910 may include but is not limited to a Processing device such as a microprocessor NCU or a programmable logic device FPGA), a memory 930 for storing data, one or more storage media 920 for storing applications 923 or data 922 (e.g., one or more storage devices), a program stored in the storage media 930 and 920 may be stored temporarily or permanently, and a program stored in the storage media 920 may include one or more modules, each of which may include a series of operations on a server, the Central Processing Units (CPU) 910 may be configured to communicate with the storage media 920 or permanently with a computer terminal, a server, or a similar computing device, and may further include one or more wireless network interfaces ttt 900, tn, ttxn &, ttx.

The input/output interface 940 may be used to receive or transmit data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the server 900. In one embodiment, the input/output Interface 940 includes a Network adapter (NIC) that can be connected to other Network devices through a base station to communicate with the internet. In one example, the input/output interface 940 may be a Radio Frequency (RF) module, which is used for communicating with the internet in a wireless manner.

It will be understood by those skilled in the art that the structure shown in fig. 9 is merely illustrative and is not intended to limit the structure of the electronic device. For example, server 900 may also include more or fewer components than shown in FIG. 9, or have a different configuration than shown in FIG. 9.

Embodiments of the present application also provide a storage medium that can be disposed in a server to store at least one instruction, at least one program, a code set, or a set of instructions related to implementing a method for compensating for a total field of a magnetic field in method embodiments, where the at least one instruction, the at least one program, the code set, or the set of instructions are loaded and executed by the processor to implement the method for compensating for a total field of a magnetic field.

Alternatively, in this embodiment, the storage medium may be located in at least one network server of a plurality of network servers of a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to: various media capable of storing program codes, such as a usb disk, a Read-Only-memory (RON), a random Access memory (RAN), a removable hard disk, a magnetic disk, or an optical disk.

The device provided by the application acquires a component value of a magnetic field in a coordinate system through the superconducting triaxial magnetometer; acquiring an optical pump total field value of a magnetic field through an optical pump sensor; correcting the component value to obtain a corrected component value; obtaining a direction cosine characteristic matrix by using the magnetic interference compensation model, the corrected component values and the total field value of the optical pump; acquiring attitude information through an inertial navigation sensor, and determining an attitude characteristic matrix based on the attitude information; filtering the direction cosine feature matrix and the attitude feature matrix to obtain a filtered direction cosine feature matrix and a filtered attitude feature matrix; and inputting the filtered direction cosine characteristic matrix and the filtered attitude characteristic matrix into a trained magnetic field compensation model to obtain a compensated magnetic interference value, and obtaining a target magnetic field total field value according to the optical pump total field value and the compensated magnetic interference value. Therefore, noise related to the motion attitude in the compensated magnetic field total field value can be reduced and even removed, so that the compensation precision can be improved, and the magnetic field total field value with higher precision can be obtained.

It should be noted that: the sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And specific embodiments thereof have been described above. Other embodiments are within the scope of the following claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In some embodiments, multitasking and parallel processing may also be possible or may be advantageous.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus embodiment, since it is substantially similar to the method embodiment, the description is relatively simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

It will be understood by those skilled in the art that all or part of the steps for implementing the above embodiments may be implemented by hardware, or may be implemented by a program instructing relevant hardware, where the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disk, etc.

The above description is only exemplary of the present application and should not be taken as limiting the present application, as any modification, equivalent replacement, or improvement made within the spirit and principle of the present application should be included in the protection scope of the present application.