CN113156525A - Underwater multi-magnetic target positioning method based on neural network - Google Patents
Underwater multi-magnetic target positioning method based on neural network Download PDFInfo
- Publication number
- CN113156525A CN113156525A CN202110422632.7A CN202110422632A CN113156525A CN 113156525 A CN113156525 A CN 113156525A CN 202110422632 A CN202110422632 A CN 202110422632A CN 113156525 A CN113156525 A CN 113156525A
- Authority
- CN
- China
- Prior art keywords
- magnetic
- magnetic field
- target
- data
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 20
- 230000005291 magnetic effect Effects 0.000 claims abstract description 258
- 238000003062 neural network model Methods 0.000 claims abstract description 16
- 238000012549 training Methods 0.000 claims description 23
- 239000010410 layer Substances 0.000 claims description 20
- 230000006870 function Effects 0.000 claims description 18
- 230000035699 permeability Effects 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 230000005358 geomagnetic field Effects 0.000 claims description 4
- 238000013507 mapping Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 230000006698 induction Effects 0.000 claims description 3
- 239000011229 interlayer Substances 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000012546 transfer Methods 0.000 claims description 3
- 238000004458 analytical method Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 abstract description 5
- 238000001514 detection method Methods 0.000 description 7
- 238000011160 research Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 4
- 230000004807 localization Effects 0.000 description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000005389 magnetism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 239000013535 sea water Substances 0.000 description 2
- 230000002238 attenuated effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000009795 derivation Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000005294 ferromagnetic effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/38—Processing data, e.g. for analysis, for interpretation, for correction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- Remote Sensing (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention provides an underwater multi-magnetic target positioning method based on a neural network, which comprises the steps of firstly modeling a magnetic field model, respectively modeling each magnetic target by using a single-magnetic target modeling method, and then fusing and adding the magnetic field models to obtain the magnetic field model of the multi-magnetic target; secondly, magnetic field data are manufactured by using a magnetic field model; then, analyzing the magnetic field model, and establishing a proper neural network model according to the magnetic field model; and finally, testing and improving the neural network model by using the manufactured magnetic field data, thereby obtaining the neural network model capable of positioning the underwater multi-magnetic target and realizing the rapid and high-precision positioning of the underwater multi-magnetic target. The scheme of the invention is suitable for positioning underwater multiple magnetic targets, can keep high-precision positioning and simultaneously greatly improve the positioning efficiency of the multiple magnetic targets, and does not need to solve a magnetic field model.
Description
Technical Field
The invention relates to the field of underwater target detection and positioning, in particular to an underwater multi-magnetic target positioning method based on a neural network.
Background
The ocean area of China is vast, and the method has important significance for detecting and positioning ocean resources, underwater vehicles and the like. Due to the limitations of the underwater optical environment and the maneuvering limitations of its work platform, it is relatively difficult to detect and locate underwater targets. Unlike land or air positioning methods, the special environmental conditions of seawater make the positioning means on water have many limitations when used underwater. The research on positioning and navigation of a land or air target object is usually based on radio waves or optical signals, however, when the radio waves propagate in water, the signal intensity can be rapidly attenuated, so that the signal propagation distance can be greatly shortened, and the requirements of underwater operation are difficult to meet; when the optical signal propagates in water, the optical signal not only attenuates rapidly but also scatters, and it is difficult to track the specific position of the signal source. Currently, the localization of underwater objects is typically performed using acoustic detection and localization techniques. However, the traditional acoustic detection positioning system is complex and high in manufacturing cost, and is not very convenient to use or maintain. Acoustic detection is an active detection mode, is significantly affected by the underwater acoustic environment, and requires that the detected object have significant acoustically recognizable characteristics relative to the surrounding environment. In shallow sea or offshore environment, obvious acoustic background noise can seriously affect the positioning of the target, and the seawater salinity and temperature can also reduce the positioning precision of the acoustic detection system on the target.
Locating by taking advantage of the ferromagnetic properties of magnetic targets has its unique advantages over conventional acoustic locating techniques. The method is less influenced, the distribution of the magnetic fields of the geomagnetic field and the magnetic target is relatively stable except for few situations such as magnetic storm, the generated magnetic field is relatively stable even for non-cooperative moving targets, and the magnetic field value can be well measured. For sunken ships and other working machines on the seabed, whether the targets are targets or not is difficult to judge by using an acoustic positioning technology, the targets are easy to mix with the seabed terrain, and the problem can be well solved by positioning the targets by using the magnetism of the targets. Especially for a magnetic target such as a sunken ship or a black clamp covered by the seabed silt, the magnetic positioning of the target is a good choice. Although magnetic detection techniques require the detected object to have distinct independent magnetic characteristics compared to the surrounding environment, this requirement is easily met for underwater objects, which do not interfere with many other magnetic fields on land.
Currently, most of the research on the positioning of magnetic targets is to research a single magnetic target, that is, the target is taken as an ideal single magnetic source. Because the quantity of parameters to be solved can be obviously increased by simultaneously inverting the multiple magnetic targets, and the aliasing magnetic field does not have a standard spatial distribution model, the research difficulty of positioning the multiple magnetic targets is increased, and related research results are fewer. When multiple magnetic targets are positioned, the positioning difficulty and error are larger along with the increase of the number of the targets. According to the magnetic dipole model, the magnetic field intensity is rapidly reduced as the magnetic target is farther away, and the weak magnetic fields can be filtered out as noise during inversion; when positioning is carried out, the existence of ships or underwater vehicles, geomagnetic background fields, other interference magnetic fields and the like can also have great influence on the positioning precision. In real-world environments, however, the ideal single magnetic source is less, and more so if there are multiple magnetic targets, there will be a larger error in locating a single magnetic target without regard to other magnetic objects.
Disclosure of Invention
The invention provides an underwater multi-magnetic target positioning method based on a neural network, which solves the problem that the existing underwater magnetic target positioning technology is limited to positioning a single magnetic target.
In order to achieve the technical effects, the technical scheme of the invention is as follows:
an underwater multi-magnetic target positioning method based on a neural network comprises the following steps:
s1: establishing a multi-magnetic target magnetic field model and generating multi-magnetic target magnetic field data;
s2: establishing a neural network model;
s3: and carrying out inversion positioning on the multiple magnetic targets.
Further, in step S1, a multiple magnetic target magnetic field model is established, and data of the magnetic field strength around the target is generated according to the multiple magnetic target magnetic field model; the multi-magnetic target magnetic field model is used for superposing the single-magnetic target magnetic field models, the magnetic field strength of a certain point around the target is the vector sum of the magnetic field strengths generated by each magnetic target at the point, different amounts of magnetic target magnetic field data such as single-magnetic targets, double-magnetic targets, three-magnetic targets and the like can be generated when the magnetic target magnetic field data are generated, and finally, the magnetic field gradient tensor of the generated magnetic field data is calculated; when the distance from the measuring point to the magnetic target is greater than 2.5 times of the maximum physical size of the target, analyzing and processing by using a magnetic dipole model instead of the magnetic target, wherein the magnetic field is regarded as a superimposed magnetic field generated by a magnetic dipole group, and the specific process of the step S1 is as follows:
the vector expression for the magnetic dipole model is:
wherein the magnetic dipole model represents the magnetic induction intensity, mu, generated by a magnetic target at the origin at a point P (x, y, z) at a distance r0Denotes the magnetic permeability of the magnetic dipole in vacuum, with a value of mu0＝4π×10-7H/m, μ ≈ μ in air0The medium permeability is approximately treated as mu when studying magnetic dipoles0In some studies, 4 pi may be omitted and μ ═ 10 may be taken as it is-7H/m, m is magnetic moment;
let Bx、ByAnd BzIs the component of B in the x, y and z directions, mx、myAnd mzFor the components of m in the x, y and z directions, then B can be represented as a matrix:
the magnetic field gradient tensor represents the rate of spatial change of the three components of the magnetic field vector B in three mutually perpendicular directions in a cartesian coordinate system. For the magnetic field M, the complete magnetic field gradient tensor is expressed as:
because the magnetic field is in the passive space, the divergence and the rotation of the magnetic field are both 0, namely:
thus, it is possible to obtain:
Bxx+Byy+Bzz＝0
Bxy＝Byx
Bxz＝Bzx
Byz＝Bzy
wherein, BxxAnd G11All represent a component BxCalculating the partial derivative of x, and the like;
the components in the magnetic field gradient are represented as:
where i, j represent the three components of a cartesian coordinate system (i.e., i, j is 1,2,3), and when i is j, δ is presentij1, when i ≠ j, δ ij0; the magnetic gradient tensor has 9 components in total, but the number of the independent components is only 5, and when the magnetic field gradient data is generated, only 5 independent magnetic field gradient tensors need to be generated without generating all the magnetic gradient tensor data; as can be seen from the vector of the magnetic dipole model, the relative position (x, y, z) of the magnetic dipole and the measuring point and the magnetic moment (m) are givenx,my,mz) The magnitudes (B) of the three components of the magnetic field can be determinedx,By,Bz) When data are produced, the magnetic field data unit obtained by calculation can be converted into a unit consistent with the geomagnetic field;the permeability coefficient mu may be approximated to be 10-7H/m, magnetic moment (m) set when generating magnetic field intensity datax,my,mz) The magnitude of (100000,200000,300000), the relative position of the magnetic dipole and the measuring point can be randomly generated, after the magnetic field data is generated, the magnetic field gradient tensor is solved according to the gradient tensor formula, and finally the data set of the magnetic field and the gradient thereof is obtained, wherein the specific process of the step S2 is as follows:
establishing a function used by a magnetic target neural network model as a newff function by using matlab, importing generated data by using a load function, and then carrying out normalization processing on training data; the number of nodes in the input layer is generally determined by the number of data to be solved, and since 9 nodes are used for inverting the position of the magnetic target by the magnetic field and the gradient data thereof, the number of nodes in the input layer is 9; the number of the hidden layers is set to be two layers, the number of nodes in the first layer is 20, the number of nodes in the second layer is 40, and the accuracy and the efficiency of inversion are influenced by the number of the hidden layers and the number of the nodes; according to the transmission of data, the interlayer transfer functions are functions of 'tansig', 'logsig' and 'tansig'; the final output is the position of the magnetic target with three components, so the output dimension is set to be 3; creating a training network, wherein a used training function is 'rainlm'; the training result is displayed every 50 steps, the maximum number of training steps is 8000, and the precision of training is 10-4The learning rate is set to 0.01, the more training times and the higher the set precision are, the more accurate the mapping relation formed by the neural network is, and the overlarge learning efficiency influences the inversion efficiency.
Further, in step S3, the generated magnetic field data and the neural network model are used to perform inversion positioning on the single magnetic target, the double magnetic target, and the triple magnetic target, respectively, to obtain a position result of the magnetic target, and the specific process of step S3 is:
respectively inverting the relative positions of the magnetic targets under three conditions of a single magnetic target, a double magnetic target and a triple magnetic target to generate three-dimensional graphs of the relative positions and the real positions of the magnetic targets, calculating the relative error of the generated relative positions of the magnetic targets, expressing the relative error by using gamma, and calculating the relative error by using a calculation formula of the relative error if r is the distance between a measuring point and the magnetic target:
compared with the prior art, the technical scheme of the invention has the beneficial effects that:
firstly, modeling a magnetic field model, respectively modeling each magnetic target by using a single magnetic target modeling method, and then fusing and adding the magnetic targets to obtain a magnetic field model of multiple magnetic targets; secondly, magnetic field data are manufactured by using a magnetic field model; then, analyzing the magnetic field model, and establishing a proper neural network model according to the magnetic field model; and finally, testing and improving the neural network model by using the manufactured magnetic field data, thereby obtaining the neural network model capable of positioning the underwater multi-magnetic target and realizing the rapid and high-precision positioning of the underwater multi-magnetic target. The scheme of the invention is suitable for positioning underwater multiple magnetic targets, can keep high-precision positioning and simultaneously greatly improve the positioning efficiency of the multiple magnetic targets, and does not need to solve a magnetic field model.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a parameter setting of a neural network model constructed according to the present invention;
FIG. 3(a) shows the result of locating a single magnetic target;
FIG. 3(b) shows the dual magnetic target localization result;
FIG. 3(c) shows the results of three magnetic target locations.
Detailed Description
The drawings are for illustrative purposes only and are not to be construed as limiting the patent;
for the purpose of better illustrating the embodiments, certain features of the drawings may be omitted, enlarged or reduced, and do not represent the size of an actual product;
it will be understood by those skilled in the art that certain well-known structures in the drawings and descriptions thereof may be omitted.
The technical solution of the present invention is further described below with reference to the accompanying drawings and examples.
As shown in fig. 1, an underwater multi-magnetic target positioning method based on a neural network includes the following steps:
firstly, establishing a multi-magnetic target magnetic field model and generating multi-magnetic target magnetic field data;
when the distance from the measuring point to the magnetic target is more than 2.5 times of the maximum physical size of the target, the magnetic dipole model is usually used to replace the magnetic target for analysis, which is easy to satisfy in practice, so that the magnetic field can be regarded as a superimposed magnetic field generated by the magnetic dipole group.
The vector expression for the magnetic dipole model is:
wherein the magnetic dipole model represents the magnetic induction intensity generated by the magnetic target at the origin at a point P (x, y, z) at a distance r. Mu.s0Denotes the magnetic permeability of the magnetic dipole in vacuum, with a value of mu0＝4π×10-7H/m, μ ≈ μ in air0The dielectric permeability can also be approximated as mu when studying magnetic dipoles0In some studies, 4 pi may be omitted and μ ═ 10 may be taken as it is-7H/m, m is the magnetic moment.
Let Bx、ByAnd BzIs the component of B in the x, y and z directions, mx、myAnd mzFor the components of m in the x, y and z directions, then B can be represented as a matrix:
the magnetic field gradient tensor represents the rate of spatial change of the three components of the magnetic field vector B in three mutually perpendicular directions in a cartesian coordinate system. For the magnetic field M, the complete magnetic field gradient tensor is expressed as:
because the magnetic field is in the passive space, the divergence and the rotation of the magnetic field are both 0, namely:
thus, it is possible to obtain:
Bxx+Byy+Bzz＝0
Bxy＝Byx
Bxz＝Bzx
Byz＝Bzy
wherein, BxxAnd G11All represent a component BxThe partial derivatives are calculated for x, and the other is similar.
The components in the magnetic field gradient are represented as:
where i, j represent the three components of a cartesian coordinate system (i.e., i, j is 1,2,3), and when i is j, δ is presentij1, when i ≠ j, δij＝0；
From the derivation of the magnetic field gradient tensor, it can be seen that the magnetic gradient tensor has 9 components in total, but only 5 independent components. Therefore, when generating the magnetic field gradient data, only 5 independent magnetic field gradient tensors need to be generated and the entire magnetic gradient tensor data need not be generated.
As can be seen from the vector of the magnetic dipole model, the relative position (x, y, z) of the magnetic dipole and the measuring point and the magnetic moment (m) are givenx,my,mz) Then the magnetism can be obtainedMagnitude of three components of field (B)x,By,Bz). When the data is manufactured, the magnetic field data unit obtained by calculation can be converted into a unit consistent with the geomagnetic field; the permeability coefficient mu may be approximated to be 10-7H/m. Magnetic moment (m) set when generating magnetic field intensity datax,my,mz) Is (100000,200000,300000), the relative positions of the magnetic dipoles and the measuring points can be randomly generated. After the magnetic field data are generated, the magnetic field gradient tensor is solved according to the gradient tensor formula, and finally the data set of the magnetic field and the gradient thereof is obtained.
Secondly, establishing a neural network model;
the matlab neural network tool box is used for establishing a neural network model, the basic idea is to introduce training data to enable the neural network to automatically learn and form a mapping relation from input to output, and then the relative position of a magnetic target is inverted according to test data.
And (3) establishing a function used by the magnetic target neural network model by using matlab as a newff function, importing the generated data by using a load function, and then performing normalization processing on the training data. The number of nodes in the input layer is generally determined by the number of data to be solved, and since 9 nodes are used in total for inverting the position of the magnetic target by the magnetic field and its gradient data, the input layer node is 9. The number of the hidden layers is two, the number of the nodes of the first layer is 20, the number of the nodes of the second layer is 40, and the accuracy and the efficiency of inversion are influenced by the number of the hidden layers and the number of the nodes. The inter-layer transfer functions are "tansig", "logsig" and "tansig" functions, respectively, depending on the transmission of data. The final output is the position of the magnetic target for a total of three components, so the output dimension is given as 3. A training network is created, with the training function "rainlm". The training result is displayed every 50 steps, the maximum number of training steps is 8000, and the precision of training is 10-4The learning rate is set to 0.01. The more training times and the higher the set precision are, the more accurate the mapping relation formed by the neural network is, and the overlarge learning efficiency influences the inversion efficiency. FIG. 2 is a diagram of a neural network of the present invention, with some parameters and time available for display on an interface. Multiple test finding, trainingWhen the data is more than 500 groups and the training times are more than one thousand, the position error of the magnetic target obtained by inversion can be basically controlled within 5 percent. The test data for the single magnetic target, the double magnetic target and the triple magnetic target are all 50 sets, and the relative error between their actual positions and the inverted positions is shown in the attached table 1 (only the first 25 sets of data are taken). As can be seen from the attached table 1, the position error obtained by the method is within 5% no matter whether the target is a single magnetic target or a multi-magnetic target, which shows that the positioning result obtained by the method has higher precision and can meet the actual positioning requirement on the target object.
Thirdly, carrying out inversion positioning on the multiple magnetic targets;
in this step, the relative positions of the magnetic targets are inverted for the three cases of the single magnetic target, the double magnetic target, and the triple magnetic target, respectively, three-dimensional maps of the relative positions and the true positions of the magnetic targets are generated, and the relative errors of the generated relative positions of the magnetic targets are calculated.
The relative error is expressed by gamma, and r is the distance between the measuring point and the magnetic target, the calculation formula of the relative error is as follows:
the obtained positioning results of the single magnetic target, the double magnetic target and the triple magnetic target are shown in fig. 3 and table 1:
TABLE 1 relative error in the position of a magnetic target obtained according to the invention
The same or similar reference numerals correspond to the same or similar parts;
the positional relationships depicted in the drawings are for illustrative purposes only and are not to be construed as limiting the present patent;
it should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.
Claims (9)
1. An underwater multi-magnetic target positioning method based on a neural network is characterized by comprising the following steps:
s1: establishing a multi-magnetic target magnetic field model and generating multi-magnetic target magnetic field data;
s2: establishing a neural network model;
s3: and carrying out inversion positioning on the multiple magnetic targets.
2. The method for positioning underwater multi-magnetic target based on neural network as claimed in claim 1, wherein in step S1, a multi-magnetic target magnetic field model is established, and data of magnetic field intensity around the target is generated according to the multi-magnetic target magnetic field model; the multi-magnetic target magnetic field model is used for superposing the single-magnetic target magnetic field models, the magnetic field strength of a certain point around the target is the vector sum of the magnetic field strengths generated by the magnetic targets at the point, different amounts of magnetic target magnetic field data such as the single-magnetic target, the double-magnetic target, the triple-magnetic target and the like can be generated when the magnetic target magnetic field data are generated, and finally, the magnetic field gradient tensor of the generated magnetic field data is calculated.
3. The method for positioning underwater multi-magnetic target based on neural network as claimed in claim 2, wherein in step S1, when the distance from the measuring point to the magnetic target is greater than 2.5 times of the maximum physical size of the target itself, the magnetic dipole model is used to replace the magnetic target for analysis, and the magnetic field is regarded as the superimposed magnetic field generated by the magnetic dipole group.
4. The method for positioning underwater multi-magnetic target based on neural network as claimed in claim 3, wherein the specific process of step S1 is:
the vector expression for the magnetic dipole model is:
wherein the magnetic dipole model represents the magnetic induction intensity, mu, generated by a magnetic target at the origin at a point P (x, y, z) at a distance r0Denotes the magnetic permeability of the magnetic dipole in vacuum, with a value of mu0＝4π×10-7H/m, μ ≈ μ in air0The medium permeability is approximately treated as mu when studying magnetic dipoles0In some studies, 4 pi may be omitted and μ ═ 10 may be taken as it is-7H/m, m is magnetic moment;
let Bx、ByAnd BzIs the component of B in the x, y and z directions, mx、myAnd mzFor the components of m in the x, y and z directions, then B can be represented as a matrix:
the magnetic field gradient tensor represents the rate of spatial change of the three components of the magnetic field vector B in three mutually perpendicular directions in a cartesian coordinate system. For the magnetic field M, the complete magnetic field gradient tensor is expressed as:
because the magnetic field is in the passive space, the divergence and the rotation of the magnetic field are both 0, namely:
thus, it is possible to obtain:
Bxx+Byy+Bzz＝0
Bxy＝Byx
Bxz＝Bzx
Byz＝Bzy
wherein, BxxAnd G11All represent a component BxCalculating the partial derivative of x, and the like;
the components in the magnetic field gradient are represented as:
where i, j represent the three components of a cartesian coordinate system (i.e., i, j is 1,2,3), and when i is j, δ is presentij1, when i ≠ j, δij＝0。
5. The method of claim 4, wherein the magnetic gradient tensor has 9 components in total, but only 5 independent components, and only 5 independent magnetic gradient tensors are needed to generate the magnetic field gradient data, and not all the magnetic gradient tensor data.
6. The neural network-based underwater multi-magnetic target of claim 5The calibration positioning method is characterized in that the relative position (x, y, z) of the magnetic dipole and the measuring point and the magnetic moment (m) are given as long as the vector of the magnetic dipole model can knowx,my,mz) The magnitudes (B) of the three components of the magnetic field can be determinedx,By,Bz) When data are produced, the magnetic field data unit obtained by calculation can be converted into a unit consistent with the geomagnetic field; the permeability coefficient mu may be approximated to be 10-7H/m, magnetic moment (m) set when generating magnetic field intensity datax,my,mz) The magnitude of (100000,200000,300000), the relative position of the magnetic dipole and the measuring point can be randomly generated, after the magnetic field data is generated, the magnetic field gradient tensor is obtained according to the gradient tensor formula, and finally the data set of the magnetic field and the gradient thereof is obtained.
7. The method for positioning underwater multi-magnetic target based on neural network as claimed in claim 6, wherein the specific process of step S2 is:
establishing a function used by a magnetic target neural network model as a newff function by using matlab, importing generated data by using a load function, and then carrying out normalization processing on training data; the number of nodes in the input layer is generally determined by the number of data to be solved, and since 9 nodes are used for inverting the position of the magnetic target by the magnetic field and the gradient data thereof, the number of nodes in the input layer is 9; the number of the hidden layers is set to be two layers, the number of nodes in the first layer is 20, the number of nodes in the second layer is 40, and the accuracy and the efficiency of inversion are influenced by the number of the hidden layers and the number of the nodes; according to the transmission of data, the interlayer transfer functions are functions of 'tansig', 'logsig' and 'tansig'; the final output is the position of the magnetic target with three components, so the output dimension is set to be 3; creating a training network, wherein a used training function is 'rainlm'; the training result is displayed every 50 steps, the maximum number of training steps is 8000, and the precision of training is 10-4The learning rate is set to 0.01, the more training times and the higher the set precision are, the more accurate the mapping relation formed by the neural network is, and the overlarge learning efficiency influences the inversion efficiency.
8. The method as claimed in claim 7, wherein in step S3, the generated magnetic field data and the neural network model are used to perform inversion positioning on the single magnetic target, the double magnetic target and the triple magnetic target respectively, so as to obtain the position result of the magnetic target.
9. The method for positioning underwater multi-magnetic target based on neural network as claimed in claim 8, wherein the specific process of step S3 is:
respectively inverting the relative positions of the magnetic targets under three conditions of a single magnetic target, a double magnetic target and a triple magnetic target to generate three-dimensional graphs of the relative positions and the real positions of the magnetic targets, calculating the relative error of the generated relative positions of the magnetic targets, expressing the relative error by using gamma, and calculating the relative error by using a calculation formula of the relative error if r is the distance between a measuring point and the magnetic target:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110422632.7A CN113156525A (en) | 2021-04-15 | 2021-04-15 | Underwater multi-magnetic target positioning method based on neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110422632.7A CN113156525A (en) | 2021-04-15 | 2021-04-15 | Underwater multi-magnetic target positioning method based on neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113156525A true CN113156525A (en) | 2021-07-23 |
Family
ID=76868944
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110422632.7A Pending CN113156525A (en) | 2021-04-15 | 2021-04-15 | Underwater multi-magnetic target positioning method based on neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113156525A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113640712A (en) * | 2021-08-06 | 2021-11-12 | 中国人民解放军海军工程大学 | Prediction method for vertical component of vertical induced magnetic field of ship |
| CN115144796A (en) * | 2022-06-27 | 2022-10-04 | 北京航空航天大学 | Modeling and magnetic field estimation method and device for residual magnetism and inductive magnetism of magnet in geomagnetic field |
| TWI783713B (en) * | 2021-10-06 | 2022-11-11 | 瑞昱半導體股份有限公司 | Method and system for building dipole moment model |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202815202U (en) * | 2012-10-22 | 2013-03-20 | 中国船舶重工集团公司第七一〇研究所 | Magnetic torque measuring device for magnetic target based on nine triaxial vector magnetic sensor |
| CN109781093A (en) * | 2018-10-09 | 2019-05-21 | 武昌工学院 | A kind of submarine target magnetic location tracking system and method |
-
2021
- 2021-04-15 CN CN202110422632.7A patent/CN113156525A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202815202U (en) * | 2012-10-22 | 2013-03-20 | 中国船舶重工集团公司第七一〇研究所 | Magnetic torque measuring device for magnetic target based on nine triaxial vector magnetic sensor |
| CN109781093A (en) * | 2018-10-09 | 2019-05-21 | 武昌工学院 | A kind of submarine target magnetic location tracking system and method |
Non-Patent Citations (3)
| Title |
|---|
| 王金根,林春生,龚沈光: "一种基于神经网络的磁性目标定位方法", 海军工程大学学报 * |
| 陈胜之: ""磁性目标定位技术研究"", 《中国优秀硕士学位论文全文数据库 基础科技辑》 * |
| 高俊吉;刘大明;周国华;: "水中非合作运动磁性目标跟踪及参数估计", 哈尔滨工程大学学报 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113640712A (en) * | 2021-08-06 | 2021-11-12 | 中国人民解放军海军工程大学 | Prediction method for vertical component of vertical induced magnetic field of ship |
| CN113640712B (en) * | 2021-08-06 | 2023-11-24 | 中国人民解放军海军工程大学 | Prediction method for vertical component of vertical induction magnetic field of ship |
| TWI783713B (en) * | 2021-10-06 | 2022-11-11 | 瑞昱半導體股份有限公司 | Method and system for building dipole moment model |
| CN115144796A (en) * | 2022-06-27 | 2022-10-04 | 北京航空航天大学 | Modeling and magnetic field estimation method and device for residual magnetism and inductive magnetism of magnet in geomagnetic field |
| CN115144796B (en) * | 2022-06-27 | 2024-06-11 | 北京航空航天大学 | Modeling and magnetic field estimation method and device for residual magnetism and magnetic induction of magnet in geomagnetic field |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113156525A (en) | Underwater multi-magnetic target positioning method based on neural network | |
| CN109975879B (en) | Magnetic dipole target tracking method based on magnetic sensor array | |
| CN110095738B (en) | Parameter design method for magnetic gradiometer | |
| CN111412910B (en) | Ship axis frequency magnetic field positioning method and device based on rotating magnetic dipole | |
| CN111399066B (en) | Method for processing scalar magnetic anomaly gradient signal based on orthogonal basis function | |
| CN111504318B (en) | Ocean navigation auxiliary method based on multi-magnetic dipole inversion | |
| CN106093849B (en) | A kind of Underwater Navigation method based on ranging and neural network algorithm | |
| CN106990440A (en) | A kind of submarine localization method based on magnetic field space gradient at two detecting locations | |
| Tang et al. | Localization of multiple underwater objects with gravity field and gravity gradient tensor | |
| CN109991670B (en) | Underwater target positioning method | |
| Lee et al. | Real-time localization for underwater moving object using precalculated DC electric field template | |
| CN103353612B (en) | A kind of measurement and positioning equipment of underground target object and measurement and positioning method | |
| Tang et al. | Analytical formulas for underwater and aerial object localization by gravitational field and gravitational gradient tensor | |
| Shili et al. | An accurate localization method for subsea pipelines by using external magnetic fields | |
| CN108761384B (en) | Target positioning method for robust sensor network | |
| Yan et al. | Accurate aerial object localization using gravity and gravity gradient anomaly | |
| CN109579845A (en) | A kind of ship tracking localization method based on electric field detecting array | |
| CN112344925A (en) | Underwater target correlation detection positioning method based on normalized magnetic moment | |
| CN109188527B (en) | Method for rapidly establishing three-dimensional offshore bottom speed model in beach and shallow sea area | |
| CN108828509B (en) | Multi-platform multi-radiation source bearing relation judgment method | |
| CN106908058B (en) | Method for determining aperture of geomagnetic positioning array | |
| CN113703050B (en) | Secondary positioning method for deep-sea seismic vertical cable | |
| Liang et al. | Multiple Underwater Magnetic Targets Location Method Based on Neural Network | |
| Gutnik et al. | Data-Driven Underwater Navigation workshop: AUV Close-Range Localization and Guidance Employing an Electro-Magnetic Beacon | |
| Guo et al. | Three-directional analytic signal analysis and interpretation of magnetic gradient tensor |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20210723 |