CN104931028B - A kind of three axle magneto-electronic compass error compensation methods based on deep learning - Google Patents
A kind of three axle magneto-electronic compass error compensation methods based on deep learning Download PDFInfo
- Publication number
- CN104931028B CN104931028B CN201510378092.1A CN201510378092A CN104931028B CN 104931028 B CN104931028 B CN 104931028B CN 201510378092 A CN201510378092 A CN 201510378092A CN 104931028 B CN104931028 B CN 104931028B
- Authority
- CN
- China
- Prior art keywords
- error
- magnetic
- training
- electronic compass
- deep learning
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000013135 deep learning Methods 0.000 title claims abstract description 16
- 230000005291 magnetic effect Effects 0.000 claims abstract description 79
- 238000012549 training Methods 0.000 claims abstract description 40
- 238000005259 measurement Methods 0.000 claims abstract description 24
- 230000008569 process Effects 0.000 claims abstract description 6
- 238000004364 calculation method Methods 0.000 claims abstract description 3
- 239000013598 vector Substances 0.000 claims description 19
- 241000701447 unidentified baculovirus Species 0.000 claims description 6
- 230000001133 acceleration Effects 0.000 claims description 5
- 230000006870 function Effects 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 239000006185 dispersion Substances 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 3
- 239000000696 magnetic material Substances 0.000 claims description 3
- 230000002159 abnormal effect Effects 0.000 claims description 2
- 230000005856 abnormality Effects 0.000 claims description 2
- 238000013461 design Methods 0.000 claims description 2
- 239000003302 ferromagnetic material Substances 0.000 claims description 2
- 238000001914 filtration Methods 0.000 claims description 2
- 230000005484 gravity Effects 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims description 2
- 238000011160 research Methods 0.000 claims description 2
- 238000012360 testing method Methods 0.000 claims description 2
- 238000010200 validation analysis Methods 0.000 claims description 2
- 238000013528 artificial neural network Methods 0.000 abstract description 7
- 230000005389 magnetism Effects 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- WKBPZYKAUNRMKP-UHFFFAOYSA-N 1-[2-(2,4-dichlorophenyl)pentyl]1,2,4-triazole Chemical compound C=1C=C(Cl)C=C(Cl)C=1C(CCC)CN1C=NC=N1 WKBPZYKAUNRMKP-UHFFFAOYSA-N 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005553 drilling Methods 0.000 description 1
- 230000005358 geomagnetic field Effects 0.000 description 1
- 229910052500 inorganic mineral Inorganic materials 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 239000011707 mineral Substances 0.000 description 1
- 230000007935 neutral effect Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- GOLXNESZZPUPJE-UHFFFAOYSA-N spiromesifen Chemical compound CC1=CC(C)=CC(C)=C1C(C(O1)=O)=C(OC(=O)CC(C)(C)C)C11CCCC1 GOLXNESZZPUPJE-UHFFFAOYSA-N 0.000 description 1
- 239000011031 topaz Substances 0.000 description 1
- 229910052853 topaz Inorganic materials 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C17/00—Compasses; Devices for ascertaining true or magnetic north for navigation or surveying purposes
- G01C17/38—Testing, calibrating, or compensating of compasses
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Measuring Magnetic Variables (AREA)
Abstract
A kind of three axle magneto-electronic compass error compensation methods based on deep learning, are trained to implicit expression error model, to compensate the nonlinearity erron that magnetic compass measurement exists, improve magnetic compass orientation accuracy;Error model training includes two stages:First stage is pre-training;Second stage is reverse fine setting, and using back-propagation algorithm, to network, all layers are finely adjusted, and reduces model training error;Magnetic compass is demarcated and the process of compensation is exactly the Nonlinear Error Models being obtained using deep learning Algorithm for Training, true magnetic field value is returned in the measurement magnetic field inversion after distortion, thus reducing course angular error calculation;The error training method based on deep learning proposing for magnetic compass nonlinearity erron, for comparing traditional neural network random initializtion, each layer weight can be located at the preferable position of parameter space, is conducive to improving Algorithm Convergence and model training precision, realize magnetic compass high-precision fixed to.
Description
Technical field
The present invention relates to the error compensation of Magnetic Electronic Compass magnetic-field measurement, particularly nonlinearity erron part therein is entered
Row compensates.
Technical background
High Accuracy Magnetic Electronic Compass is not only one of core technology of earth-magnetism navigation, fixed in mineral reserve detection, drilling well, robot
To etc. field apply also widely.Abroad have now been developed the Magnetic Electronic Compass of better performances, but soft magnetism interference and intersecting axle
The nonlinearity erron that effect leads to is well solved not yet.It is additionally, since Magnetic Electronic Compass and there is significantly military potential,
Western countries strictly forbid that high accuracy navigation product exports China always.
Magnetic Electronic Compass utilizes magnetic sensor and acceleration transducer, and measurement earth's magnetic field and gravitational field are in each axle respectively
Component, and magnetic heading angle is calculated by coordinate transform.Magnetic heading angleWith actual heading angleBetween relation beWherein β be measurement place magnetic declination (angle of deviation between magnetic field north and north geographic pole can pass through earth's magnetic field
Model is calculated).Magnetic heading angleFor
WhereinIt is Magnetic Electronic Compass measurement magnetic fieldIn the projection of geographical horizontal plane, attitude can be passed through
Conversion obtains
Wherein θ and γ is respectively the angle of pitch and roll angle, can be solved by (3) formula.
Wherein g is gravitational field amplitude,WithIt is respectively magnetic compass along the gravitational acceleration component of x-axis and y-axis.
The measurement error of Magnetic Sensor and acceleration transducer directly affects magnetic compass orientation accuracy.Magnetic compass is installed on load
When body carries out magnetic-field measurement, the soft magnetic materials on carrier has relatively low coercivity and narrower hysteresis curve, can be compared with weak magnetic
It is magnetized in, produce nonlinear induced field.The size of induced field and direction are with attitude of carrier and carrier in earth's magnetic field
In change in location and change, once external magnetic field disappears, its induced magnetism also disappears therewith, and remanent magnetism is almost nil.And carrier
Soft magnetism shape and distribution also more complicated, around magnetic compass formed non-uniform magnetic-field.In addition, the change of temperature also can cause
The change of signal amplification circuit and modulate circuit parameter in gaussmeter, thus produce nonlinear measurement error.Various non-linear mistakes
Difference mechanism of production is different, and especially soft magnetism interference profile situation is complicated, is difficult to find out rule therein, sets up explicit error mould
Type.
At present, both at home and abroad magneto-electronic compass error is calibrated substantially by the analysis to source of error and mechanism, build
A vertical explicit linearity error model, then using distinct methods, error parameter is estimated and compensated to, such as:Ellipse hypothesis
Method, step calibration method, ellipsoid subjunctive, Filters with Magnitude Constraints method, position inversion method, dot product not political reform etc.[1-6].Certainly, in addition
Also part document is had to think that Magnetic Sensor comprises nonlinearity erron, by neural network implicit expression error model.Document [7,8]
With the deformation of each component of sample data as independent variable, construct the functional-link direct type FLANN network structure of no hidden layer, to measure magnetic
And the true magnetic field amplitude difference of two squares and be cost function, by iterate to calculate estimation difference parameter, magnetic field amplitude after calibration
Error declines 66 times and 10 times respectively, close to ellipsoid subjunctive.Document [9,10] proposes to describe magnetic using three-layer neural network
Non-linear relation between measurement course angle and true course angle under the disturbed condition of field, defeated with reference to training input using outside course
Go out Nonlinear Error Models, compensation precision is up to 0.4 °, relevant with training sample quality.Document [11] utilizes three layers of BP god
Through network, magnetic course angle error is modeled, neural network weight is trained using adaptive differential evolution algorithm, navigate after compensation
Control in the range of ± 0.22 ° to angle error, but do not specify and adopt what type of Magnetic Sensor.Although neutral net can be distinguished
Know nonlinear system, solve complicated modeling problem, but generally existing convergence is slow, the problems such as poor real, training speed becomes
For restricting the main problem of its development and application.In addition traditional neural network there is also " platform " phenomenon in the training process
Or generation over-fitting, lead to network generalization poor.
List of references
[1]Gebre-Egziabher D.Magnetometer autocalibration leveraging
measurement locus constraints[J].Journal of Aircraft,2007,44(4):1361–1368.
[2]Vasconcelos J F,Elkaim G,Silvestre C,et al.Geometric approach to
strapdown magnetometer calibration in sensor frame[J].IEEE Transactions on
Aerospace and Electronic Systems,2011,47(2):1293-1306.
[3]Asl H.G.Pourtakdoust S.H.Samani M.A new non-linear algorithm for
complete pre-flight calibration of magnetometers in the geomagnetic field
domain[J].Proc.Inst.Mech.Eng.G J.Aerosp.Eng.2009,223,729–739.
[4]Bonnet S.Bassompierre C.Godin C.et al.Calibration methods for
inertial and magnetic sensors[J].Sens.ActuatorsAPhys.2009,156,302–311.
[5] Li Xiang, Li Zhi. the dot product not political reform [J] of attitude heading reference system three axis magnetometer correction. Chinese journal of scientific instrument,
2012,33(8):1813-1818.
[6]Yanxia Liu,Xisheng Li,Xiaojuan Zhang,et al.Novel Calibration
Algorithm for a Three-Axis Strapdown Magnetometer[J].Sensors,2014,14,8485-
8504.
[7] Wu Dehui, yellow pine ridge, Zhao Wei. [J] is studied in the three axis magnetometer error correction based on FLANN. instrument and meter
Report, 2009-3,30 (3):449-453.
[8] topaz, Hao Yanling. the gradometer error correction [J] based on FLANN and least square. Chinese journal of scientific instrument,
2012-4,33(4):911-917.
[9]Wang J H,Gao Y.A new magnetic compass calibration algorithm using
neural networks[J].Measurement Science and Technology,2006,17(1):153-160.
[10] Wang Lu, Zhao Zhong, Shao Yumei etc. magnetic compass error analyses and compensation [J]. sensing technology journal 2007,20
(2):439-441.
[11] Yue Haibo, Zhang Shudong, Shi Zhi eat. and evolved based on adaptive differential and the compass error of BP network compensates
[J]. aerospace journal, 2013-12,34 (12):1628-1633.
Content of the invention
The present invention is a kind of method that can be used for Magnetic Electronic Compass nonlinear error compensation it is proposed that a kind of be based on depth
The three axis magnetometer Error Compensation Algorithm practised,
Implicit expression error model
Implicit expression error model is trained, to compensate the nonlinearity erron that Magnetic Electronic Compass measurement exists, improves magnetoelectricity
Sub- compass orientation precision, as shown in Figure 1.
The input of error model is 7 dimensional vectorsWhereinRepresent magnetic respectively
Under electronic compass coordinate system, earth's magnetic field is in x, y, the component of z-axis,Represent gravitational field under Magnetic Electronic Compass coordinate system respectively
In x, y, the component of z-axis, tm p represents temperature;It is output as actual heading anglePut using current research and most widely used depth
Communication network DBN builds many hidden layers (having multiple restriction Boltzmann machine RBM superpositions to constitute);Last layer adopts BP network, receives
From the output characteristic vector of RBM, and the input feature value as BP network.Course angle measurementWith actual heading angle
Between there is nonlinear functional relationship
Wherein β is magnetic declination, and f () represents course angle measurementAnd actual valueBetween nonlinear function mapping close
System.The purpose of deep learning algorithm is exactly that this nonlinear function preferably expressed by training error model.
Error model is trained
Error model training includes two stages.First stage is pre-training, initially with to sdpecific dispersion
(Contrastive Divergence, CD) algorithm is successively trained to RBM.In each step, the front L-1 having trained
Layer is fixing, then increases L layer.Finally with depth network weight that the weights initialisation of the independent training of each layer is whole.Second-order
Section is reverse fine setting, provides more accurate course angle reference first with photoelectric encoder, as the mark of training data.Then
Using back-propagation algorithm, to network, all layers are finely adjusted, and reduce model training error.
Error compensation
Magnetic Electronic Compass is demarcated and the process of compensation is exactly the nonlinearity erron mould being obtained using deep learning Algorithm for Training
Type, returns the measurement magnetic field inversion after distortion to true magnetic field value, thus reducing course angular error calculation.
WhereinIt is the Magnetic Electronic Compass course angle after compensating,For activation primitive, X is input vector, w(L)Represent the Connecting quantity vector between L layer and L+1 layer each unit, a(L)Represent the activation value of L layer each unit, b(L)Represent
It is attached to the bias term of L+1 layer each unit.
Advantage of the present invention
The nonlinearity erron existing for Magnetic Electronic Compass, sets up implicit expression error model it is proposed that based on deep learning
Error training method, for comparing traditional neural network random initializtion, each layer weight can be located at the preferable position of parameter space,
Improve Algorithm Convergence and model training precision it is achieved that Magnetic Electronic Compass high-precision fixed to.
Brief description
Fig. 1 is Nonlinear Error Models Deep Learning network.
Fig. 2 compensates schematic diagram for magneto-electronic compass error.
Specific embodiment
Magneto-electronic compass error compensation method based on deep learning proposed by the present invention in two stages, i.e. non-linear mistake
Error compensation during difference model training and use.
Error model is trained
The first step:As shown in Fig. 2 being fixed on Magnetic Electronic Compass on Baculovirus vectors in experimentation, carrier contains ferromagnetic material
Material and soft magnetic materials, are wound around a circle wire outside Baculovirus vectors, and size of current can adjust and (change soft magnetism interference magnetic field, mould
Intend nonlinearity erron).Then Baculovirus vectors are installed on three axles no magnetic turntable, realize the angle of pitch, roll using no magnetic turntable
Angle, azimuthal variation, and using the output of photoelectric encoder as angle reference.
Second step:Using same place earth's magnetic field and gravity field vector invariant feature, design slip median filter, to survey
Amount data carries out median filter process, rejects earth's magnetic field and gravitational field amplitude variation abnormality data, reduce magnetic-field measurement noise and
Random error is disturbed;Abandon geomagnetic fieldvector and gravitational field inner product of vector abnormal data, reduce gaussmeter and accelerometer measures
Noise and random error interference, obtain quality preferable error model training setAnd mark
Note
3rd step:Using minimax method, initial data is normalized, improves network weight (i.e. error mould
Type) training precision.
4th step:Randomly draw in training set 80% data, utilize to the method such as sdpecific dispersion and reverse fine setting to depth
Network is trained.
5th step:Remaining 20% data is as test set, the training effect of validation error model.Repeat the 4th step and
Five steps, until obtaining more stable training result.
Error compensation
Magnetic Electronic Compass is arranged on suitably local using carrier inside, the magnetic then Magnetic Electronic Compass measurement being obtained
, the data such as acceleration and temperature after filtering with normalized after, input the Nonlinear Error Models training above, root
Course angle output after being compensated according to formula (5)As shown in Figure 2.
Claims (2)
1. a kind of three axle magneto-electronic compass error compensation methods based on deep learning it is characterised in that:
A kind of three axis magnetometer Error Compensation Algorithm based on deep learning of this method,
Implicit expression error model
Implicit expression error model is trained, to compensate the nonlinearity erron that Magnetic Electronic Compass measurement exists, improves magnetic electronic sieve
Disk orientation accuracy;
The input of error model is 7 dimensional vectorsWhereinRepresent magnetic electronic respectively
Under compass coordinate system, earth's magnetic field is in x, y, the component of z-axis,Represent that under Magnetic Electronic Compass coordinate system, gravitational field is in x respectively,
Y, the component of z-axis, tmp represents temperature;It is output as actual heading angleUsing current research and most widely used depth confidence net
Network DBN builds many hidden layers;Last layer adopts BP network, receives the output characteristic vector from RBM, and defeated as BP network
Enter characteristic vector;Course angle measurementWith actual heading angleBetween there is nonlinear functional relationship
Wherein β is magnetic declination, and f () represents course angle measurementWith actual heading angleBetween nonlinear function mapping close
System;The purpose of deep learning algorithm is exactly that this nonlinear function preferably expressed by training error model;
Error model is trained
Error model training includes two stages;First stage be pre-training, initially with sdpecific dispersion algorithm is carried out to RBM by
Layer training;In each step, the front L-1 layer having trained is fixed, then increases L layer;Finally with the independent training of each layer
The whole depth network weight of weights initialisation;Second stage is reverse fine setting, provides more accurate first with photoelectric encoder
Course angle reference, as the mark of training data;Then using back-propagation algorithm, to network, all layers are finely adjusted, and reduce
Model training error;
Error compensation
Magnetic Electronic Compass is demarcated and the process of compensation is exactly the Nonlinear Error Models being obtained using deep learning Algorithm for Training,
True magnetic field value is returned in measurement magnetic field inversion after distortion, thus reducing course angular error calculation;
WhereinIt is the Magnetic Electronic Compass course angle after compensating,For activation primitive, X is input vector, w(L)Table
Show the Connecting quantity vector between L layer and L+1 layer each unit, a(L)Represent the activation value of L layer each unit, b(L)Represent additional
Bias term to L+1 layer each unit.
2. a kind of three axle magneto-electronic compass error compensation methods based on deep learning according to claim 1, its feature
It is:This method propose the magneto-electronic compass error compensation method based on deep learning in two stages, i.e. nonlinearity erron
Error compensation during model training and use;
Error model is trained
The first step:In experimentation, Magnetic Electronic Compass is fixed on Baculovirus vectors, carrier contains ferromagnetic material and soft magnetic materials,
It is wound around a circle wire outside Baculovirus vectors, size of current can be adjusted;Then Baculovirus vectors are installed in three axles no magnetic turntable
On, realize the angle of pitch, roll angle, azimuthal variation using no magnetic turntable, and using the output of photoelectric encoder as angle reference;
Second step:Using same place earth's magnetic field and gravity field vector invariant feature, design slip median filter, to measurement number
According to carrying out median filter process, reject earth's magnetic field and gravitational field amplitude variation abnormality data, reduce magnetic-field measurement noise and random
Error interference;Abandon geomagnetic fieldvector and gravitational field inner product of vector abnormal data, reduce gaussmeter and accelerometer measures noise
With random error interference, obtain quality preferable error model training setAnd mark
3rd step:Using minimax method, initial data is normalized, improves the training precision of network weight;
4th step:Randomly draw in training set 80% data, utilize to the method such as sdpecific dispersion and reverse fine setting to depth network
It is trained;
5th step:Remaining 20% data is as test set, the training effect of validation error model;Repeat the 4th step and the 5th
Step, until obtaining more stable training result;
Error compensation
Magnetic Electronic Compass is arranged on suitably local using carrier inside, then Magnetic Electronic Compass measurement obtain magnetic field,
The data such as acceleration and temperature after filtering with normalized after, input the Nonlinear Error Models that train above, according to
Course angle after formula (2) is compensated exports
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510378092.1A CN104931028B (en) | 2015-06-30 | 2015-06-30 | A kind of three axle magneto-electronic compass error compensation methods based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510378092.1A CN104931028B (en) | 2015-06-30 | 2015-06-30 | A kind of three axle magneto-electronic compass error compensation methods based on deep learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104931028A CN104931028A (en) | 2015-09-23 |
CN104931028B true CN104931028B (en) | 2017-03-01 |
Family
ID=54118306
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510378092.1A Expired - Fee Related CN104931028B (en) | 2015-06-30 | 2015-06-30 | A kind of three axle magneto-electronic compass error compensation methods based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104931028B (en) |
Families Citing this family (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105445764A (en) * | 2015-11-13 | 2016-03-30 | 上海华测导航技术股份有限公司 | Spatial attitude accuracy test method for GNSS receivers |
CN105843243B (en) * | 2016-05-06 | 2019-02-22 | 深圳一电航空技术有限公司 | Control method, device and the multi-rotor aerocraft of multi-rotor aerocraft |
CN106767789B (en) * | 2017-01-12 | 2019-12-24 | 南京航空航天大学 | Pedestrian course optimal fusion method based on adaptive Kalman filtering |
CN106959121B (en) * | 2017-02-28 | 2020-12-29 | 北京联合大学 | Application method of self-contained reverse-optimization-based overrun learning algorithm in magnetic compass error compensation |
CN107132587B (en) * | 2017-04-06 | 2019-03-05 | 中国科学院上海微系统与信息技术研究所 | The full tensor magnetic gradient measurements system mounting error calibration method of aviation superconduction and device |
CN107238847A (en) * | 2017-04-21 | 2017-10-10 | 中国科学院光电研究院 | A kind of satellite navigation interferer signal detection method and system |
CN107084712B (en) * | 2017-04-21 | 2019-11-29 | 北京京东尚科信息技术有限公司 | Data processing method and device and method for calibrating compass and device |
CN109389412B (en) * | 2017-08-02 | 2022-03-04 | 创新先进技术有限公司 | Method and device for training model, service equipment and user equipment |
CN107389049B (en) * | 2017-08-10 | 2019-12-24 | 北京联合大学 | Magnetic compass error real-time compensation method based on quasi-Kalman factor |
CN107655472B (en) * | 2017-08-30 | 2019-11-01 | 杨华军 | A kind of high-precision inertial navigation set error compensating method based on deep learning |
CN108494946B (en) * | 2018-02-09 | 2021-09-28 | 维沃移动通信有限公司 | Method and device for correcting electronic compass in mobile terminal |
CN109188026B (en) * | 2018-10-25 | 2021-03-09 | 北京航空航天大学 | Automatic calibration deep learning method suitable for MEMS accelerometer |
CN109470888B (en) * | 2018-12-05 | 2020-10-27 | 西北工业大学 | Calibration system and calibration method of high-g-value accelerometer based on deep learning |
CN109855623B (en) * | 2019-01-09 | 2020-07-31 | 东南大学 | Online approximation method for geomagnetic model based on L egenderre polynomial and BP neural network |
CN109870153B (en) * | 2019-03-28 | 2020-07-31 | 北京大学 | Magnetometer orthogonality calibration test method and calibration test device |
CN110324091B (en) * | 2019-06-17 | 2022-05-03 | 电子科技大学 | Nonlinear equalizer based on function chain neural network |
CN111273202A (en) * | 2020-02-25 | 2020-06-12 | 中国电子科技集团公司第二十九研究所 | Array-based magnetic sensor compensation method |
CN112284366B (en) * | 2020-10-26 | 2022-04-12 | 中北大学 | Method for correcting course angle error of polarized light compass based on TG-LSTM neural network |
CN113503899A (en) * | 2021-07-12 | 2021-10-15 | 清华大学 | Nonlinear error suppression system and sensor based on neural network |
CN117537792B (en) * | 2024-01-03 | 2024-04-30 | 西南应用磁学研究所(中国电子科技集团公司第九研究所) | Electronic compass self-adaptive azimuth correction method |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP5070428B2 (en) * | 2006-10-23 | 2012-11-14 | アイチ・マイクロ・インテリジェント株式会社 | Electronic compass and direction measurement method |
CN102297687B (en) * | 2011-05-13 | 2012-07-04 | 北京理工大学 | Calibrating method for electronic compass |
-
2015
- 2015-06-30 CN CN201510378092.1A patent/CN104931028B/en not_active Expired - Fee Related
Also Published As
Publication number | Publication date |
---|---|
CN104931028A (en) | 2015-09-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104931028B (en) | A kind of three axle magneto-electronic compass error compensation methods based on deep learning | |
CN107389049B (en) | Magnetic compass error real-time compensation method based on quasi-Kalman factor | |
CN103267520B (en) | A kind of three axle digital compasses | |
CN102252689A (en) | Electronic compass calibration method based on magnetic sensor | |
CN104406610B (en) | A kind of magnetometer real time correction device and method | |
CN109188026B (en) | Automatic calibration deep learning method suitable for MEMS accelerometer | |
CN109696183A (en) | The scaling method and device of Inertial Measurement Unit | |
CN106647791A (en) | Monitoring device for three-dimensional posture, mechanical device and monitoring method for three-dimensional posture | |
CN105891755A (en) | Aircraft hanging-type fluxgate magnetic gradient tensor instrument correction method | |
CN109856689B (en) | Noise suppression processing method and system for superconducting aeromagnetic gradient tensor data | |
CN104166169A (en) | Error compensation method of three-axis magnetic field sensor connected to underwater vehicle in strap-down mode | |
CN111780786A (en) | Online calibration method for three-axis TMR sensor | |
CN104316037B (en) | A kind of bearing calibration of electronic compass and device | |
CN105785477A (en) | Geomagnetic vector measurement error calibration method with combination of element restriction and summation restriction | |
CN106403952A (en) | Method for measuring combined attitudes of Satcom on the move with low cost | |
CN106441301A (en) | Air vehicle launching initial parameter acquiring method and system | |
CN110187393B (en) | Aeromagnetic compensation method based on generalized regression neural network | |
CN108731676A (en) | A kind of posture fusion enhancing measurement method and system based on inertial navigation technology | |
CN106017452A (en) | Dual gyro anti-disturbance north-seeking method | |
CN109814163A (en) | A kind of boat magnetic tensor data noise restraint method and system based on extension compensation model | |
CN108458728A (en) | A kind of Magnetic Sensor on-line calibration method for unmanned plane | |
CN107860382A (en) | A kind of method for measuring posture using AHRS in the case of magnetic anomaly | |
CN105303201B (en) | A kind of method and system carrying out handwriting recognition based on action induction | |
Chen et al. | A novel calibration method for tri-axial magnetometers based on an expanded error model and a two-step total least square algorithm | |
CN107036576B (en) | Real-time resolving method based on difference coefficient method magnetic survey Rotary aircraft roll angle |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170301 |