CN110617836A - Model-free Doppler log DVL error calibration method - Google Patents
Model-free Doppler log DVL error calibration method Download PDFInfo
- Publication number
- CN110617836A CN110617836A CN201910910599.5A CN201910910599A CN110617836A CN 110617836 A CN110617836 A CN 110617836A CN 201910910599 A CN201910910599 A CN 201910910599A CN 110617836 A CN110617836 A CN 110617836A
- Authority
- CN
- China
- Prior art keywords
- horizontal
- vertical
- predictor
- dvl
- regression
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C25/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C25/00—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass
- G01C25/005—Manufacturing, calibrating, cleaning, or repairing instruments or devices referred to in the other groups of this subclass initial alignment, calibration or starting-up of inertial devices
Abstract
The invention discloses a model-free DVL error calibration method for a Doppler log, which can realize high-precision error calibration without a priori DVL error model. The method comprises the following steps: in the underwater motion process of an underwater vehicle AUV, two-dimensional horizontal velocity information and vertical velocity information of the AUV are respectively obtained, and the horizontal measurement velocity and the vertical measurement velocity of a Doppler log DVL arranged above the AUV are obtained. And constructing a horizontal regression predictor, and training the horizontal regression predictor by using the two-dimensional horizontal velocity information as an output training sample and the horizontal measurement velocity of the DVL as an input training sample to obtain the trained horizontal regression predictor. And obtaining the well-trained vertical regression predictor in the same way. And combining the trained horizontal regression predictor and the trained vertical regression predictor to obtain the target predictor. And collecting DVL speed measurement information in real time as the input of the target predictor, wherein the output of the target predictor is the calibrated DVL measurement speed.
Description
Technical Field
The invention relates to the technical field of navigation systems, in particular to a method for calibrating errors of a model-free Doppler log DVL.
Background
High-precision underwater navigation and positioning are the basis of all ocean development activities and ocean high-tech development, and play an extremely important role in many aspects. The navigation problem of the autonomous underwater vehicle AUV causes a complex working environment and becomes very challenging. Since the drift error of the inertial navigation system INS is unavoidable and can accumulate over time, long-term operation can cause the navigation error to exceed the task accuracy requirement. The high-precision Doppler log DVL can provide accurate speed information and can correct the speed error accumulation of the INS in time, so that the positioning precision is improved. Therefore, the INS/DVL combined navigation is currently the most widely used underwater navigation technique.
In an underwater integrated navigation system, two factors restrict the positioning accuracy of an INS/DVL integrated navigation system: first, the installation error angle between INS/DVL; second is the speed measurement error of the DVL itself.
How to accurately realize the DVL error calibration to construct a high-precision underwater navigation system is a main research direction at present. At present, an automatic calibration method for performing calibration by using acceleration information does not need an additional sensor, but needs a very complex motion track, and the control difficulty of an underwater vehicle is greatly increased. And then, a two-point calibration method is provided, and calibration can be completed only by the starting point information and any point information in the navigation process. In order to realize the self-calibration of the scale factor error and the installation error angle, a calibration method based on system observability analysis can be adopted, and the calibration can be completed without adding any additional sensor. The algorithm has great limitation and is complex to realize in practical application, and in order to solve the problem, a least square calibration method based on SVD is provided.
The above conventional calibration algorithm needs to satisfy the following two assumptions: (1) the DVL error model is known; (2) the installation error angles are small. However, in practical applications, the DVL error model cannot be accurately modeled, because not all error terms can be considered, and these not considered DVL error terms have a great influence on the calibration accuracy. At the same time, it is not possible to ensure that the installation error angle between the INS and the DVL is small. It can be seen that the conventional calibration algorithm has a great limitation, whether in terms of calibration accuracy or algorithm reliability.
Therefore, a method for calibrating the DVL error without a prior model is needed.
Disclosure of Invention
In view of this, the invention provides a model-free doppler log DVL error calibration method, which can realize high-precision error calibration without a priori DVL error model. .
In order to achieve the purpose, the technical scheme of the invention comprises the following steps:
in the underwater motion process of an underwater vehicle AUV, two-dimensional horizontal velocity information and vertical velocity information of the AUV are respectively obtained, and the horizontal measurement velocity and the vertical measurement velocity of a Doppler log DVL arranged above the AUV are obtained.
And constructing a horizontal regression predictor, using the two-dimensional horizontal velocity information as an output training sample of the horizontal regression predictor, using the horizontal measurement velocity of the DVL as an input training sample of the horizontal regression predictor, and training the horizontal regression predictor to obtain the trained horizontal regression predictor.
And constructing a vertical regression predictor, taking the vertical speed information as an output training sample of the vertical regression predictor, taking the vertical measuring speed of the DVL as an input training sample of the vertical regression predictor, and training the vertical regression predictor to obtain the trained vertical regression predictor.
And combining the trained horizontal regression predictor and the trained vertical regression predictor to obtain a target predictor.
And collecting DVL speed measurement information in real time as the input of the target predictor, wherein the output of the target predictor is the calibrated DVL measurement speed.
Further, in the underwater motion process of the underwater vehicle AUV, two-dimensional horizontal velocity information and vertical velocity information of the AUV are respectively obtained, specifically:
in the underwater motion process of an underwater vehicle AUV, an INS/GPS integrated navigation system is adopted to obtain two-dimensional horizontal velocity information of the AUV, and a pressure sensor is adopted to obtain vertical velocity information of the AUV.
Further, acquiring a horizontal measurement speed and a vertical measurement speed acquired by a doppler log DVL arranged on the AUV specifically are:
the Doppler log DVL acquires and obtains three-dimensional speed information of the AUV, wherein the three-dimensional speed information comprises an x-axis speed, a y-axis speed and a z-axis speed; wherein the x-axis, the y-axis and the z-axis are respectively: the center point of the DVL is used as an origin, the horizontal right direction is used as the positive direction of the x axis, the horizontal forward direction is used as the positive direction of the y axis, and the vertical upward direction is used as the positive direction of the z axis.
Dividing the three-dimensional speed information as follows: taking the combination of the x-axis speed and the y-axis speed as the horizontal measuring speed; and taking the z-axis speed as the vertical measuring speed.
Further, a horizontal regression prediction period is established, the two-dimensional horizontal velocity information is used as an output training sample of the horizontal regression predictor, the horizontal measurement velocity of the DVL is used as an input training sample of the horizontal regression predictor, the horizontal regression predictor is trained, and the trained horizontal regression predictor is obtained, and the method specifically comprises the following steps:
and constructing a horizontal regression predictor.
And adopting the two-dimensional horizontal velocity information as an output training sample of the horizontal regression predictor, and adopting the horizontal measurement velocity of the DVL as an input training sample of the horizontal regression predictor.
And carrying out parameter optimization on the horizontal regression predictor by using a parameter optimization algorithm to obtain an optimal parameter, substituting the optimal parameter into a support vector regression algorithm to train the horizontal regression predictor, and obtaining the trained horizontal regression predictor.
Further, a vertical regression predictor is constructed, the vertical speed information is used as an output training sample of the vertical regression predictor, the vertical measurement speed of the DVL is used as an input training sample of the super European regression predictor, the vertical regression predictor is trained, and the trained vertical predictor is obtained, and the method specifically comprises the following steps:
and constructing a vertical regression predictor.
And adopting the vertical speed information as an output training sample of the vertical regression predictor, and adopting the vertical measuring speed of the DVL as an input training sample of the vertical regression predictor.
And carrying out parameter optimization on the vertical regression predictor by using a parameter optimization algorithm to obtain an optimal parameter, substituting the optimal parameter into a support vector regression algorithm to train the vertical regression predictor, and obtaining the trained vertical regression predictor.
Further, the parameter optimization algorithm is a genetic algorithm.
Has the advantages that:
(1) the invention provides a method for calibrating errors of a model-free Doppler log DVL (dynamic Voltage Log) by using a regression predictor to calibrate the errors of the DVL. When the method is applied to the invention, in the training process, the DVL speed measurement with the error is taken as an input training sample, the acquired accurate AUV speed information is taken as an output training sample, and the regression predictor obtained by training carries out error calibration on the DVL speed measurement with the error.
(2) The model-free Doppler log DVL error calibration method provided by the invention has the advantages that when the regression detector is trained, parameter optimization is carried out by using a parameter optimization algorithm, and the obtained optimal parameters are substituted into a support vector regression algorithm for training to obtain a regression predictor. The support vector regression algorithm can complete training only by few training samples, which is very beneficial to the underwater vehicle to perform tasks.
Drawings
FIG. 1 is a flowchart of a method for model-free DVL error calibration of GA-SVR according to an embodiment of the present invention.
Detailed Description
The invention is described in detail below by way of example with reference to the accompanying drawings.
The invention provides a model-free Doppler log DVL error calibration method, the flow principle of which is shown in figure 1, and the method specifically comprises the following steps:
and S1, respectively acquiring two-dimensional horizontal velocity information and vertical velocity information of the AUV during the underwater motion process of the AUV, and acquiring the horizontal measurement velocity and the vertical measurement velocity of a Doppler log DVL arranged above the AUV.
In the embodiment of the invention, in the underwater motion process of the AUV of the underwater carrier, an INS/GPS integrated navigation system is adopted to obtain two-dimensional horizontal velocity information of the AUV, namely the AUV sails on the water surface and acquires GPS information to construct the INS/GPS integrated navigation system, and the INS/GPS integrated navigation system outputs the two-dimensional horizontal velocity information. And acquiring the vertical speed information of the AUV by adopting a pressure sensor.
In the embodiment of the invention, a Doppler log DVL acquires and obtains three-dimensional speed information of an AUV, wherein the three-dimensional speed information comprises an x-axis speed, a y-axis speed and a z-axis speed; wherein the x-axis, the y-axis and the z-axis are respectively: the center point of the DVL is used as an origin, the horizontal right direction is used as the positive direction of the x axis, the horizontal forward direction is used as the positive direction of the y axis, and the vertical upward direction is used as the positive direction of the z axis. The three-dimensional speed information is divided as follows: taking the combination of the x-axis speed and the y-axis speed as a horizontal measuring speed; the z-axis velocity is used as the vertical measurement velocity.
The data collection process in fig. 1 is the present step S1.
S2, constructing a horizontal regression predictor, and training the horizontal regression predictor by using the two-dimensional horizontal velocity information as an output training sample of the horizontal regression predictor and the horizontal measurement velocity of the DVL as an input training sample of the horizontal regression predictor to obtain the trained horizontal regression predictor.
The following steps can be specifically adopted:
constructing a horizontal regression predictor;
two-dimensional horizontal velocity information is used as an output training sample of a horizontal regression predictor, and the horizontal measurement velocity of the DVL is used as an input training sample of the horizontal regression predictor;
and performing parameter optimization on the horizontal regression predictor by using a parameter optimization algorithm to obtain an optimal parameter, substituting the optimal parameter into a support vector regression algorithm to train the horizontal regression predictor, and obtaining the trained horizontal regression predictor.
The horizontal training process in fig. 1 is the present step S2.
S3, constructing a vertical regression predictor, training the vertical regression predictor by using the vertical speed information as an output training sample of the vertical regression predictor and using the vertical measurement speed of the DVL as an input training sample of the vertical regression predictor to obtain the trained vertical regression predictor.
Specifically, the following method can be adopted
Constructing a vertical regression predictor;
the vertical speed information is used as an output training sample of a vertical regression predictor, and the vertical measuring speed of the DVL is used as an input training sample of the vertical regression predictor;
and performing parameter optimization on the vertical regression predictor by using a parameter optimization algorithm to obtain an optimal parameter, substituting the optimal parameter into a support vector regression algorithm to train the vertical regression predictor, and obtaining the trained vertical regression predictor. The parameter optimization algorithm is a genetic algorithm, and can also be an ant colony algorithm or a particle swarm algorithm.
The support vector regression algorithm can complete training only by few training samples, which is very beneficial to the underwater vehicle to perform tasks.
The vertical training process in fig. 1 is the present step S3, and in the implementation process, S2 and S3 may be executed sequentially or simultaneously.
S4, combining the trained horizontal regression predictor and the trained vertical regression predictor to obtain a target predictor;
and S5, collecting DVL speed measurement information in real time as the input of the target predictor, wherein the output of the target predictor is the calibrated DVL measurement speed.
As shown in fig. 1, in the embodiment of the present invention, after the calibrated DVL measurement speed is output, kalman filtering is further performed once, and the filtered data is input into the INS/DVL integrated navigation system.
Experimental results show that the model-free DVL error calibration algorithm provided by the invention not only can greatly improve the calibration precision of the traditional calibration algorithm, but also can realize calibration under the condition of a large installation error angle, which is an advantage that the traditional calibration algorithm does not have.
In summary, the above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (6)
1. A DVL error calibration method of a model-free Doppler log is characterized by comprising the following steps:
respectively acquiring two-dimensional horizontal velocity information and vertical velocity information of an AUV (autonomous Underwater vehicle) in the underwater motion process of the AUV, and acquiring the horizontal measurement velocity and the vertical measurement velocity of a Doppler log DVL (digital video recorder) arranged above the AUV;
constructing a horizontal regression predictor, using the two-dimensional horizontal velocity information as an output training sample of the horizontal regression predictor, using the horizontal measurement velocity of the DVL as an input training sample of the horizontal regression predictor, and training the horizontal regression predictor to obtain a trained horizontal regression predictor;
constructing a vertical regression predictor, taking the vertical speed information as an output training sample of the vertical regression predictor, taking the vertical measuring speed of the DVL as an input training sample of the vertical regression predictor, and training the vertical regression predictor to obtain a well-trained vertical regression predictor;
the trained horizontal regression predictor and the trained vertical regression predictor are combined to obtain a target predictor;
and collecting DVL speed measurement information in real time as the input of the target predictor, wherein the output of the target predictor is the calibrated DVL measurement speed.
2. The method as claimed in claim 1, wherein the two-dimensional horizontal velocity information and vertical velocity information of the underwater vehicle AUV are obtained during the underwater movement of the AUV, respectively, and specifically:
in the underwater motion process of an underwater vehicle AUV, an INS/GPS integrated navigation system is adopted to obtain two-dimensional horizontal velocity information of the AUV, and a pressure sensor is adopted to obtain vertical velocity information of the AUV.
3. The method of claim 1, wherein the obtaining of the horizontal measurement velocity and the vertical measurement velocity collected by the doppler velocity log DVL disposed on the AUV comprises:
the Doppler log DVL acquires and obtains three-dimensional speed information of the AUV, wherein the three-dimensional speed information comprises an x-axis speed, a y-axis speed and a z-axis speed;
wherein the x-axis, the y-axis and the z-axis are respectively: taking the central point of the DVL as an origin, taking the horizontal right direction as the positive direction of an x axis, the horizontal forward direction as the positive direction of a y axis and the vertical upward direction as the positive direction of a z axis;
dividing the three-dimensional speed information as follows:
taking the combination of the x-axis speed and the y-axis speed as the horizontal measuring speed;
and taking the z-axis speed as the vertical measuring speed.
4. The method according to claim 1, wherein the constructing the horizontal regression prediction period includes using the two-dimensional horizontal velocity information as an output training sample of the horizontal regression predictor, using the horizontal measurement velocity of the DVL as an input training sample of the horizontal regression predictor, and training the horizontal regression predictor to obtain a trained horizontal regression predictor, specifically:
constructing a horizontal regression predictor;
adopting the two-dimensional horizontal velocity information as an output training sample of the horizontal regression predictor, and adopting the horizontal measurement velocity of the DVL as an input training sample of the horizontal regression predictor;
and carrying out parameter optimization on the horizontal regression predictor by using a parameter optimization algorithm to obtain an optimal parameter, substituting the optimal parameter into a support vector regression algorithm to train the horizontal regression predictor, and obtaining the trained horizontal regression predictor.
5. The method according to claim 1, wherein the constructing of the vertical regression predictor, the vertical velocity information is used as an output training sample of the vertical regression predictor, the vertical measurement velocity of the DVL is used as an input training sample of the super european regression predictor, and the vertical regression predictor is trained to obtain a trained vertical predictor, specifically:
constructing a vertical regression predictor;
the vertical speed information is used as an output training sample of the vertical regression predictor, and the vertical measuring speed of the DVL is used as an input training sample of the vertical regression predictor;
and carrying out parameter optimization on the vertical regression predictor by using a parameter optimization algorithm to obtain an optimal parameter, substituting the optimal parameter into a support vector regression algorithm to train the vertical regression predictor, and obtaining the trained vertical regression predictor.
6. The method of claim 4 or 5, wherein the parameter-optimizing algorithm is a genetic algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910910599.5A CN110617836B (en) | 2019-09-25 | 2019-09-25 | Model-free Doppler log DVL error calibration method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910910599.5A CN110617836B (en) | 2019-09-25 | 2019-09-25 | Model-free Doppler log DVL error calibration method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110617836A true CN110617836A (en) | 2019-12-27 |
| CN110617836B CN110617836B (en) | 2021-06-01 |
Family
ID=68924034
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910910599.5A Active CN110617836B (en) | 2019-09-25 | 2019-09-25 | Model-free Doppler log DVL error calibration method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110617836B (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102162733A (en) * | 2011-01-11 | 2011-08-24 | 哈尔滨工程大学 | Method for correcting autonomous underwater vehicle (AUV) dead reckoning navigation error in real time based on space vector modulation (SVM) |
| CN103513240A (en) * | 2012-06-20 | 2014-01-15 | 中国科学院声学研究所 | Underwater reverberation suppression method based on vector regression supporting |
| US9304198B1 (en) * | 2013-10-04 | 2016-04-05 | Sandia Corporation | Navigator alignment using radar scan |
| CN106548188A (en) * | 2015-09-17 | 2017-03-29 | 宝钢不锈钢有限公司 | A kind of nickel valency Forecasting Methodology based on sample optimizing support vector machine |
| CN106840150A (en) * | 2017-01-25 | 2017-06-13 | 东南大学 | A kind of mixed processing method for DVL failures in integrated navigation |
| CN108871379A (en) * | 2018-07-04 | 2018-11-23 | 北京理工大学 | A kind of DVL range rate error online calibration method |
| CN109655850A (en) * | 2018-12-20 | 2019-04-19 | 内蒙古工业大学 | Beidou/sustainable navigation system of SINS pine combination and its air navigation aid |
-
2019
- 2019-09-25 CN CN201910910599.5A patent/CN110617836B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102162733A (en) * | 2011-01-11 | 2011-08-24 | 哈尔滨工程大学 | Method for correcting autonomous underwater vehicle (AUV) dead reckoning navigation error in real time based on space vector modulation (SVM) |
| CN103513240A (en) * | 2012-06-20 | 2014-01-15 | 中国科学院声学研究所 | Underwater reverberation suppression method based on vector regression supporting |
| US9304198B1 (en) * | 2013-10-04 | 2016-04-05 | Sandia Corporation | Navigator alignment using radar scan |
| CN106548188A (en) * | 2015-09-17 | 2017-03-29 | 宝钢不锈钢有限公司 | A kind of nickel valency Forecasting Methodology based on sample optimizing support vector machine |
| CN106840150A (en) * | 2017-01-25 | 2017-06-13 | 东南大学 | A kind of mixed processing method for DVL failures in integrated navigation |
| CN108871379A (en) * | 2018-07-04 | 2018-11-23 | 北京理工大学 | A kind of DVL range rate error online calibration method |
| CN109655850A (en) * | 2018-12-20 | 2019-04-19 | 内蒙古工业大学 | Beidou/sustainable navigation system of SINS pine combination and its air navigation aid |
Non-Patent Citations (1)
| Title |
|---|
| 李旻: "基于SINS_DVL与声学定位系统的水下组合导航技术研究", 《舰船电子工程》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110617836B (en) | 2021-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110706279B (en) | Global position and pose estimation method based on information fusion of global map and multiple sensors | |
| CN106767752B (en) | Combined navigation method based on polarization information | |
| CN101285686B (en) | Agricultural machines navigation hierarchical positioning process and system | |
| CN106772524B (en) | A kind of agricultural robot integrated navigation information fusion method based on order filtering | |
| CN103674028A (en) | Positioning test method and positioning test device of indoor advancing track | |
| CN112285676B (en) | Laser radar and IMU external parameter calibration method and device | |
| CN109507706B (en) | GPS signal loss prediction positioning method | |
| Niu et al. | Development and evaluation of GNSS/INS data processing software for position and orientation systems | |
| CN108981687A (en) | A kind of indoor orientation method that vision is merged with inertia | |
| CN109471146A (en) | A kind of self-adapted tolerance GPS/INS Combinated navigation method based on LS-SVM | |
| CN103604430A (en) | Marginalized cubature Kalman filter (CKF)-based gravity aided navigation method | |
| CN102162733A (en) | Method for correcting autonomous underwater vehicle (AUV) dead reckoning navigation error in real time based on space vector modulation (SVM) | |
| CN111595348A (en) | Master-slave mode cooperative positioning method of autonomous underwater vehicle combined navigation system | |
| CN111947644B (en) | Outdoor mobile robot positioning method and system and electronic equipment thereof | |
| CN110160557A (en) | A kind of development machine inertial navigation system two-dimensional position precision calibration method and system | |
| CN110262538B (en) | Map data acquisition method, device, equipment and storage medium | |
| US10323942B2 (en) | User-specific learning for improved pedestrian motion modeling in a mobile device | |
| CN115164936A (en) | Global pose correction method and device for point cloud splicing in high-precision map manufacturing | |
| CN110617836B (en) | Model-free Doppler log DVL error calibration method | |
| CN112098926B (en) | Intelligent angle measurement training sample generation method by using unmanned plane platform | |
| CN204788411U (en) | Wall climbing robot navigational positioning system based on sound localization , INS | |
| CN104913776B (en) | Unmanned plane localization method and device | |
| KR20110114039A (en) | Dr/gps data fusion method | |
| Liu et al. | 3D LiDAR aided GNSS real-time kinematic positioning via coarse-to-fine batch optimization for high accuracy mapping in dense urban canyons | |
| JP2018036151A (en) | Positioning method using landmark |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |