CN112033406B - Navigation method, device and storage medium based on lightweight network - Google Patents
Navigation method, device and storage medium based on lightweight network Download PDFInfo
- Publication number
- CN112033406B CN112033406B CN202010838295.5A CN202010838295A CN112033406B CN 112033406 B CN112033406 B CN 112033406B CN 202010838295 A CN202010838295 A CN 202010838295A CN 112033406 B CN112033406 B CN 112033406B
- Authority
- CN
- China
- Prior art keywords
- geomagnetic
- information
- matching
- model
- data
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/005—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 with correlation of navigation data from several sources, e.g. map or contour matching
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/02—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by astronomical means
- G01C21/025—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by astronomical means with the use of startrackers
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/04—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by terrestrial means
- G01C21/08—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by terrestrial means involving use of the magnetic field of the earth
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/10—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration
- G01C21/12—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning
- G01C21/16—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation
- G01C21/165—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 by using measurements of speed or acceleration executed aboard the object being navigated; Dead reckoning by integrating acceleration or speed, i.e. inertial navigation combined with non-inertial navigation instruments
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- General Physics & Mathematics (AREA)
- Astronomy & Astrophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Environmental & Geological Engineering (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geology (AREA)
- Navigation (AREA)
Abstract
The invention discloses a navigation method, a navigation device and a storage medium based on a lightweight network, wherein the method comprises the following steps: acquiring satellite magnetic measurement data and solar high-energy particle data; constructing a lightweight multimode geomagnetic map drawing model; enabling the geomagnetic chart drawing model to draw a reference geomagnetic chart by using the input two modal data; acquiring real-time geomagnetic information of the current position of a carrier; reading reference geomagnetic map information, and matching the real-time geomagnetic information with the reference geomagnetic map information; and correcting the position of the inertial navigation system according to the matching result. The model compression is utilized to reduce the calculation amount of the generated reference geomagnetic map, and the reference geomagnetic map generated by the multi-modal model has high precision, so that the navigation is accurate.
Description
Technical Field
The invention relates to the field of navigation, in particular to a navigation method, a navigation device and a storage medium based on a lightweight network.
Background
The geomagnetic aided navigation refers to a process of navigation and positioning by using a geomagnetic map. On the basis of initial information provided by inertial navigation, the measured geomagnetic sequence and a pre-stored geomagnetic map are subjected to correlation matching, so that the real position of the carrier is determined, and high-precision navigation positioning is realized. However, the difficulty in drawing the reference geomagnetic chart seriously affects the development of geomagnetic-aided navigation.
Disclosure of Invention
The present invention is directed to at least one of the technical problems in the prior art, and provides a method, an apparatus and a storage medium for a lightweight network based navigation.
The technical scheme adopted by the invention for solving the problems is as follows:
in a first aspect of the present invention, a method for lightweight network-based navigation includes the following steps:
acquiring satellite magnetic measurement data and solar high-energy particle data;
constructing a lightweight geomagnetic map drawing model by a model compression method;
inputting the satellite magnetic measurement data and the solar high-energy particle data to the geomagnetic chart drawing model, enabling the geomagnetic chart drawing model to combine two modes of a first characteristic of the satellite magnetic measurement data and a second characteristic of the solar high-energy particle data to obtain geomagnetic characteristics, and obtaining geomagnetic indexes according to the geomagnetic characteristics in a classification mode to draw a reference geomagnetic chart;
acquiring real-time geomagnetic information of the current position of a carrier;
reading reference geomagnetic map information, and matching the reference geomagnetic map information with the real-time geomagnetic information;
and correcting the position of the inertial navigation system according to the matching result.
According to the first aspect of the present invention, while inputting the satellite magnetic survey data and the solar energetic particle data to the geomagnetic map drawing model drawing reference geomagnetic map, model weights are adjusted by an online learning method.
According to the first aspect of the present invention, the adjusting the model weight by the online learning method specifically includes: inputting a group of satellite magnetic measurement data and the solar high-energy particle data, and using a loss function and a gradient generated by the satellite magnetic measurement data and the solar high-energy particle data to draw a model iteration on the geomagnetic map; and repeating the steps, and adjusting the model weight of the geomagnetic chart drawing model through multiple iterations of the multiple groups of satellite magnetic measurement data and solar high-energy particle data on the geomagnetic chart drawing model.
According to a first aspect of the invention, the model compression method is knowledge distillation.
According to the first aspect of the present invention, the obtaining of the geomagnetic feature by combining the geomagnetic mapping model with the two modalities, i.e. the first feature of the satellite magnetic measurement data and the second feature of the solar high-energy particle data, comprises the following steps:
extracting a first characteristic of the satellite magnetic measurement data;
extracting a second characteristic of the solar energetic particle data;
inputting the first feature and the second feature to train the bidirectional automatic encoder until the bidirectional automatic encoder converges;
and removing the decoder of the bidirectional automatic encoder, and taking the characteristics output by the encoder of the bidirectional automatic encoder as the geomagnetic characteristics.
According to a first aspect of the invention, said extracting first features of satellite magnetic survey data comprises the steps of:
simultaneously extracting spatial features and time features of a fast-forwarding video segment obtained by satellite magnetic measurement data sampling through a plurality of feature encoders;
inputting the spatial features and the temporal features into a fusion network formed by a discrimination sensor and a generation sensor to obtain geomagnetic features; wherein the discriminant perceptron is configured to perceive subtle differences in the intensity of motion of the geomagnetic lines between adjacent video frames by performing a classification of the sampling intervals; the generating perceptron is reconstructed by the difference values to reduce the details of the motion of the original magnetic wire.
According to the first aspect of the present invention, the extracting the second feature of the solar high energy particle data comprises the following steps:
inputting the solar high-energy particle data to a coding module of the second feature extraction module, and processing the solar high-energy particle data through a plurality of coding sublayers; wherein in the coding sublayer, the input data sequentially passes through a first multi-head self-attention structure and a first full-connection forward network processing;
inputting the output of the coding module into a decoding module of the second feature extraction module, processing the output of the coding module through a plurality of decoding sublayers, and processing through a final linear transformation layer and a softmax function layer to obtain the second feature; wherein in the decoding sublayer, the input data is processed sequentially through a masked multi-headed self-attention structure, a second multi-headed self-attention structure, and a second fully-connected forward network.
According to the first aspect of the present invention, the reading of the reference geomagnetic map information and the matching of the reference geomagnetic map information and the real-time geomagnetic information includes the steps of:
reading the reference geomagnetic map information, and determining a coarse matching index and a fine matching index;
performing rough matching on the real-time geomagnetic information and the reference geomagnetic map information, and reserving a matching point which meets the rough matching index in the real-time geomagnetic information;
removing noise and interference signals of the roughly matched real-time geomagnetic information;
and performing fine matching on the real-time geomagnetic information subjected to noise reduction and the reference geomagnetic map information, and keeping a matching point which meets the fine matching index in the real-time geomagnetic information as a matching result.
In a second aspect of the present invention, a lightweight network-based navigation apparatus includes:
the first input module is used for inputting satellite magnetic measurement data and solar high-energy particle data;
the model construction module is used for constructing a lightweight geomagnetic map drawing model by a model compression method, wherein the geomagnetic map drawing model combines two modes, namely a first feature of satellite magnetic measurement data and a second feature of solar high-energy particle data, to obtain geomagnetic features, and geomagnetic indexes are obtained by classification according to the geomagnetic features so as to draw a reference geomagnetic map;
the second input module is used for inputting the real-time geomagnetic information of the current position of the carrier;
the matching module is used for reading reference geomagnetic map information and matching the reference geomagnetic map information with the real-time geomagnetic information;
and the correction module is used for correcting the position of the inertial navigation system according to the matching result.
In a third aspect of the present invention, a storage medium stores executable instructions that can be executed by a computer to cause the computer to perform the method for lightweight network-based navigation according to the first aspect of the present invention.
The scheme at least has the following beneficial effects: and optimizing the geomagnetic map drawing model by a model compression method, optimizing the geomagnetic map drawing model to a light weight level, reducing the layer number and the calculated amount of each module, and keeping the final generated geomagnetic map effect unchanged. By utilizing the complementarity between the two modes of the first characteristic of the satellite magnetic measurement data and the second characteristic of the solar high-energy particle data, the redundancy between the two modes is eliminated, so that better geomagnetic discrimination characteristic representation is learned, the accuracy of geomagnetic indexes obtained by classifying through a classifier is improved, and the precision of a reference geomagnetic graph is further improved. And finally, matching the real-time geomagnetic information of the current position of the carrier, which is measured by the geomagnetic measuring instrument, with the high-precision reference geomagnetic map, and correcting the position of the inertial navigation system according to the matching result to realize geomagnetic auxiliary navigation.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a diagram of steps of a lightweight network-based navigation method according to an embodiment of the present invention;
FIG. 2 is a schematic structural diagram of a lightweight network-based navigation device according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of the structure of a geomagnetism mapping model;
fig. 4 is a structural diagram of a multimodal deep learning model.
Detailed Description
Reference will now be made in detail to the present preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to like elements throughout.
In the description of the present invention, it should be understood that the orientation or positional relationship referred to in the description of the orientation, such as the upper, lower, front, rear, left, right, etc., is based on the orientation or positional relationship shown in the drawings, and is only for convenience of description and simplification of description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus, should not be construed as limiting the present invention.
In the description of the present invention, the meaning of a plurality of means is one or more, the meaning of a plurality of means is two or more, and larger, smaller, larger, etc. are understood as excluding the number, and larger, smaller, inner, etc. are understood as including the number. If the first and second are described for the purpose of distinguishing technical features, they are not to be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.
In the description of the present invention, unless otherwise explicitly limited, terms such as arrangement, installation, connection and the like should be understood in a broad sense, and those skilled in the art can reasonably determine the specific meanings of the above terms in the present invention in combination with the specific contents of the technical solutions.
Referring to fig. 1, an embodiment of an aspect of the present invention provides a navigation method based on a lightweight network, including the following steps:
s100, acquiring satellite magnetic measurement data and solar high-energy particle data;
s200, constructing a lightweight geomagnetic map drawing model 300 by a model compression method;
step S300, inputting satellite magnetic measurement data and solar high-energy particle data to a geomagnetic map drawing model 300, enabling the geomagnetic map drawing model 300 to combine two modes of a first feature of the satellite magnetic measurement data and a second feature of the solar high-energy particle data to obtain geomagnetic features, and obtaining geomagnetic indexes according to geomagnetic feature classification to draw a reference geomagnetic map;
s400, acquiring real-time geomagnetic information of the current position of the carrier;
step S500, reading reference geomagnetic map information, and matching the reference geomagnetic map information with real-time geomagnetic information;
and S600, correcting the position of the inertial navigation system according to the matching result.
In this embodiment, referring to FIG. 3, the geomagnetism mapping model 300 includes at least the following modules: the system comprises an input module 10, a first feature extraction module 20, a second feature extraction module 30, a multi-modal deep learning module 40 and a classifier 50, wherein the multi-modal deep learning module 40 is specifically a bidirectional automatic encoder, and the classifier 50 is specifically a support vector machine. Referring to fig. 4, in the multi-modal deep learning module 40, the hidden layer 21 of the first feature extraction module 20 and the hidden layer 31 of the second feature extraction module 30 are further connected in series, and a restricted boltzmann machine of an upper layer is constructed above the hidden layer 21 of the first feature extraction module 20 and the hidden layer 31 of the second feature extraction module 30 connected in series. The stacked constrained boltzmann machine will be deployed as a multi-modal deep learning module 40 and learn a shared representation of both modalities from the input first and second features at a feature sharing layer 41.
In step S200, the geomagnetic map drawing model 300 is optimized by a model compression method, the geomagnetic map drawing model 300 is optimized to a light weight level, the number of layers and the calculation amount of each module are reduced, but the final generated geomagnetic map effect is unchanged. In particular, the model compression method is knowledge distillation. Copying a teacher network with the same structure as the geomagnetic drawing model 300 and training the teacher network; under the effect of the high temperature coefficient T, the knowledge of the teacher's network is distilled to the geomagnetic mapping model 300 to optimize the geomagnetic mapping model 300.
By utilizing the complementarity between the two modes of the first characteristic of the satellite magnetic measurement data and the second characteristic of the solar high-energy particle data, the redundancy between the two modes is eliminated, so that better geomagnetic discrimination characteristic representation is learned, the accuracy of the geomagnetic index obtained by classifying by the classifier 50 is improved, and the precision of the reference geomagnetic graph is further improved.
And finally, matching the real-time geomagnetic information of the current position of the carrier, which is measured by the geomagnetic measuring instrument, with the high-precision reference geomagnetic map, and correcting the position of the inertial navigation system according to the matching result to realize geomagnetic auxiliary navigation.
Further, in step S300, while inputting satellite magnetic survey data and solar high-energy particle data to the geomagnetic map drawing model 300 to draw the reference geomagnetic map for training, the model weight is adjusted by an online learning method.
Specifically, the adjusting of the model weight by the online learning method specifically includes: inputting a group of satellite magnetic measurement data and solar high-energy particle data, and iterating a geomagnetic chart drawing model 300 by using a loss function and a gradient which are jointly generated by the satellite magnetic measurement data and the solar high-energy particle data; and repeating the steps, and adjusting the model weight of the geomagnetic chart drawing model 300 through multiple iterations of the multiple groups of satellite magnetic measurement data and solar high-energy particle data on the geomagnetic chart drawing model 300.
In this embodiment, the model weight of the optimized geomagnetic map drawing model 300 is continuously adjusted in the training data and online learning processes, so that the geomagnetic map drawing model 300 is more and more suitable for the geomagnetic map drawing direction, which is beneficial to improving the accuracy of the geomagnetic map drawing.
Further, the step of obtaining the geomagnetic feature by combining the geomagnetic drawing model 300 with the two modes of the first feature of the satellite magnetic measurement data and the second feature of the solar high-energy particle data includes:
extracting a first characteristic of satellite magnetic measurement data;
extracting second characteristics of the solar high-energy particle data;
inputting the first characteristic and the second characteristic to train the bidirectional automatic encoder until the bidirectional automatic encoder converges;
and removing the decoder of the bidirectional automatic encoder, and taking the characteristics output by the encoder of the bidirectional automatic encoder as the geomagnetic characteristics.
Further, the extracting the first feature of the satellite magnetic measurement data comprises the following steps:
simultaneously extracting spatial features and time features of a fast-forwarding video segment obtained by satellite magnetic measurement data sampling through a plurality of feature encoders;
inputting the spatial features and the temporal features into a fusion network formed by a discrimination sensor and a generation sensor to obtain geomagnetic features; wherein the discrimination sensor senses a subtle difference in the intensity of motion of the geomagnetic line between adjacent video frames by performing classification of the sampling interval; the generation perceptron is reconstructed by the difference values to reduce the details of the motion of the local magnetic wire.
Semantic preservation is realized through the distinguishing perceptron and the generating perceptron, which means that the encoded time semantics can be transferred to a downstream single-input multi-output classification task as much as possible. Through self-supervision space-time representation learning, the time resolution characteristics of the video and the inner space are captured, so that accurate extraction of geomagnetic features from geomagnetic intensity videos is facilitated, and further improvement of geomagnetic index prediction effects is facilitated.
Further, the step of extracting the second characteristic of the solar high-energy particle data comprises the following steps:
inputting the solar high-energy particle data to the coding module of the second feature extraction module 30, so that the solar high-energy particle data is processed by a plurality of coding sublayers; in the coding sublayer, the input data sequentially passes through a first multi-head self-attention structure and a first full-connection forward network for processing;
inputting the output of the coding module to the decoding module of the second feature extraction module 30, so that the output of the coding module is processed by a plurality of decoding sublayers, and then processed by a final linear transformation layer and a softmax function layer to obtain a second feature; in the decoding sublayer, the input data is processed by the mask multi-head self-attention structure, the second multi-head self-attention structure and the second full-connection forward network in sequence.
Parallel calculation is realized through a plurality of coding sublayers and a plurality of decoding sublayers, so that the calculation efficiency is improved; the calculation complexity required for calculating the real-time satellite magnetic measurement data and the association between the solar high-energy particle data and the geomagnetic index at the future moment is not increased along with the increase of the data distance, so that the calculation complexity is reduced; each head of the multi-headed self-attention structure can perform different tasks, making the model more interpretable.
Further, for step S400, the acquiring of the real-time geomagnetic information of the current position of the carrier specifically includes: two sets of geomagnetic measurement equipment are adopted to simultaneously measure the real-time geomagnetic information of the current position of the carrier, so that the equipment error is reduced, the reliability of the measured value is ensured, and the measurement principle is as follows: and subtracting the measured values of the two pieces of geomagnetic measurement equipment to obtain the approximate intrinsic error and strategy noise of the equipment, calculating the average value of the measured values of the two pieces of geomagnetic measurement equipment, and then subtracting the approximate intrinsic error and measurement noise of the equipment to obtain more accurate real-time geomagnetic information.
Further, for step S500, reading the reference geomagnetic map information, and matching the reference geomagnetic map information with the real-time geomagnetic information includes the following steps:
reading reference geomagnetic chart information, and determining a coarse matching index and a fine matching index;
carrying out rough matching on the real-time geomagnetic information and the reference geomagnetic map information by using an Npord algorithm, and reserving matching points meeting rough matching indexes in the real-time geomagnetic information; the similarity measurement function of the Nbird algorithm is P (x, y), and the (x, y) is the position of the real-time geomagnetic information on the reference geomagnetic map; in order to find the optimal matching point, the Nprod algorithm must perform similarity matching calculation on each point in the search area in the reference geomagnetic map; the rough matching index is related to noise intensity, geomagnetic characteristics and the like and can be obtained by counting multiple matching data;
removing noise and interference signals of the real-time geomagnetic information after coarse matching by using a noise reduction algorithm; because the real-time geomagnetic information has the influence of solar disturbance brought to the earth by solar wind, measurement noise of a geomagnetic sensor, magnetic interference of surrounding environment and other various errors, the real-time geomagnetic information after rough matching needs to be subjected to noise reduction processing;
carrying out fine matching on the real-time geomagnetic information subjected to noise reduction and the reference geomagnetic map information by using a Hausdorff distance algorithm, and reserving a matching point meeting a fine matching index in the real-time geomagnetic information as a matching result; hausdorff distance algorithmThe matching degree of the real-time geomagnetic information and the reference geomagnetic map can be reflected; when the matching degree meets the fine matching index, the position is the best matching position, namely the matching result x (t)0)。
The matching result has a better solution through two times of matching of coarse matching and fine matching according to the matching result x (t)0) And correcting the position of the inertial navigation system at the corresponding position on the reference geomagnetic map.
Referring to fig. 2, an embodiment of the apparatus of the present invention provides a navigation apparatus based on a lightweight network, to which the navigation method based on a lightweight network as described in the method embodiment is applied. The navigation device includes:
the first input module 100 is used for inputting satellite magnetic measurement data and solar high-energy particle data;
the model construction module 200 is configured to construct a lightweight geomagnetic map drawing model 300 by using a model compression method, where the geomagnetic map drawing model 300 combines two modes, namely a first feature of satellite magnetic measurement data and a second feature of solar high-energy particle data, to obtain geomagnetic features, and classifies the geomagnetic features to obtain geomagnetic indexes according to the geomagnetic features to draw a reference geomagnetic map;
a second input module 400, configured to input real-time geomagnetic information of the current position of the carrier;
a matching module 500, configured to read reference geomagnetic map information and match the reference geomagnetic map information with real-time geomagnetic information;
and a correction module 600 for correcting the position of the inertial navigation system according to the matching result.
In this embodiment, the geomagnetic map drawing model 300 is optimized by a model compression method, the geomagnetic map drawing model 300 is optimized to a light-weight level, the number of layers and the calculation amount of each module are reduced, and the final generated geomagnetic map effect is unchanged. By utilizing the complementarity between the two modes of the first characteristic of the satellite magnetic measurement data and the second characteristic of the solar high-energy particle data, the redundancy between the two modes is eliminated, so that better geomagnetic discrimination characteristic representation is learned, the accuracy of the geomagnetic index obtained by classifying by the classifier 50 is improved, and the precision of the reference geomagnetic graph is further improved. And finally, matching the real-time geomagnetic information of the current position of the carrier, which is measured by the geomagnetic measuring instrument, with the high-precision reference geomagnetic map, and correcting the position of the inertial navigation system according to the matching result to realize geomagnetic auxiliary navigation.
It should be noted that, the navigation apparatus based on the lightweight network, applying the navigation method based on the lightweight network as described in the method embodiment, can execute each step of the navigation method based on the lightweight network through cooperation of each module, and has the same technical effect, and therefore, detailed description is not repeated here.
In another embodiment of the present invention, a storage medium is provided, which stores executable instructions that can be executed by a computer to cause the computer to perform a method for lightweight network-based navigation according to an embodiment of the method of the present invention.
Examples of storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.
The above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiment, and the present invention shall fall within the protection scope of the present invention as long as the technical effects of the present invention are achieved by the same means.
Claims (10)
1. The navigation method based on the lightweight network is characterized by comprising the following steps:
acquiring satellite magnetic measurement data and solar high-energy particle data;
constructing a lightweight geomagnetic map drawing model by a model compression method;
inputting the satellite magnetic measurement data and the solar high-energy particle data to the geomagnetic chart drawing model, enabling the geomagnetic chart drawing model to combine two modes of a first characteristic of the satellite magnetic measurement data and a second characteristic of the solar high-energy particle data to obtain geomagnetic characteristics, and obtaining geomagnetic indexes according to the geomagnetic characteristics in a classification mode to draw a reference geomagnetic chart;
acquiring real-time geomagnetic information of the current position of a carrier;
reading reference geomagnetic map information, and matching the reference geomagnetic map information with the real-time geomagnetic information;
and correcting the position of the inertial navigation system according to the matching result.
2. The lightweight network-based navigation method according to claim 1, wherein model weights are adjusted by an online learning method while inputting the satellite magnetic survey data and the solar energetic particle data to the geomagnetic map drawing model drawing reference geomagnetic map.
3. The navigation method based on the lightweight network according to claim 2, wherein the adjusting the model weight by the online learning method specifically comprises: inputting a group of satellite magnetic measurement data and the solar high-energy particle data, and using a loss function and a gradient generated by the satellite magnetic measurement data and the solar high-energy particle data to draw a model iteration on the geomagnetic map; and repeating the steps, and adjusting the model weight of the geomagnetic chart drawing model through multiple iterations of the multiple groups of satellite magnetic measurement data and solar high-energy particle data on the geomagnetic chart drawing model.
4. The lightweight network based navigation method of claim 1, wherein the model compression method is knowledge distillation.
5. The navigation method based on the lightweight network according to claim 1, wherein the step of combining the geomagnetism mapping model with the first feature of the satellite magnetic measurement data and the second feature of the solar high-energy particle data to obtain the geomagnetic features comprises the following steps:
extracting a first characteristic of the satellite magnetic measurement data;
extracting a second characteristic of the solar energetic particle data;
inputting the first feature and the second feature to train a bidirectional automatic encoder until the bidirectional automatic encoder converges;
and removing the decoder of the bidirectional automatic encoder, and taking the characteristics output by the encoder of the bidirectional automatic encoder as the geomagnetic characteristics.
6. The lightweight network based navigation method according to claim 5, wherein the extracting the first feature of the satellite magnetic survey data comprises the following steps:
simultaneously extracting spatial features and time features of a fast-forwarding video segment obtained by satellite magnetic measurement data sampling through a plurality of feature encoders;
inputting the spatial features and the temporal features into a fusion network formed by a discrimination sensor and a generation sensor to obtain geomagnetic features; wherein the discrimination perceptron perceives subtle differences in the intensity of motion of the geomagnetic lines between adjacent video frames by performing a classification of the sampling intervals; the generating perceptron is reconstructed by the difference values to reduce the details of the motion of the original magnetic wire.
7. The navigation method based on the lightweight network according to claim 5, wherein the step of extracting the second feature of the solar energetic particle data comprises the following steps:
inputting the solar high-energy particle data to a coding module of the second feature extraction module, and processing the solar high-energy particle data through a plurality of coding sublayers; wherein in the coding sublayer, the input data sequentially passes through a first multi-head self-attention structure and a first full-connection forward network processing;
inputting the output of the coding module into a decoding module of the second feature extraction module, processing the output of the coding module through a plurality of decoding sublayers, and processing the output of the coding module through a final linear transformation layer and a softmax function layer to obtain the second feature; wherein in the decoding sublayer, the input data is processed sequentially through a masked multi-headed self-attention structure, a second multi-headed self-attention structure, and a second fully-connected forward network.
8. The navigation method based on the lightweight network according to claim 1, wherein the reading of the reference geomagnetic map information and the matching of the reference geomagnetic map information and the real-time geomagnetic information comprise the following steps:
reading the reference geomagnetic map information, and determining a coarse matching index and a fine matching index;
performing rough matching on the real-time geomagnetic information and the reference geomagnetic map information, and reserving a matching point which meets the rough matching index in the real-time geomagnetic information;
removing noise and interference signals of the roughly matched real-time geomagnetic information;
and performing fine matching on the real-time geomagnetic information subjected to noise reduction and the reference geomagnetic map information, and keeping a matching point which meets the fine matching index in the real-time geomagnetic information as a matching result.
9. A navigation device based on a lightweight network is characterized by comprising:
the first input module is used for inputting satellite magnetic measurement data and solar high-energy particle data;
the model construction module is used for constructing a lightweight geomagnetic map drawing model by a model compression method, wherein the geomagnetic map drawing model combines two modes, namely a first feature of satellite magnetic measurement data and a second feature of solar high-energy particle data, to obtain geomagnetic features, and geomagnetic indexes are obtained by classification according to the geomagnetic features so as to draw a reference geomagnetic map;
the second input module is used for inputting the real-time geomagnetic information of the current position of the carrier;
the matching module is used for reading reference geomagnetic map information and matching the reference geomagnetic map information with the real-time geomagnetic information;
and the correction module is used for correcting the position of the inertial navigation system according to the matching result.
10. Storage medium, characterized in that it stores executable instructions that can be executed by a computer, causing the computer to execute the method of lightweight network based navigation according to any of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010838295.5A CN112033406B (en) | 2020-08-19 | 2020-08-19 | Navigation method, device and storage medium based on lightweight network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010838295.5A CN112033406B (en) | 2020-08-19 | 2020-08-19 | Navigation method, device and storage medium based on lightweight network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112033406A CN112033406A (en) | 2020-12-04 |
| CN112033406B true CN112033406B (en) | 2022-05-17 |
Family
ID=73576986
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010838295.5A Active CN112033406B (en) | 2020-08-19 | 2020-08-19 | Navigation method, device and storage medium based on lightweight network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112033406B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116721316A (en) * | 2023-08-11 | 2023-09-08 | 之江实验室 | Model training and geomagnetic chart optimizing method, device, medium and equipment |
| CN117029838B (en) * | 2023-10-09 | 2024-01-23 | 广东电网有限责任公司阳江供电局 | Navigation control method and system for underwater robot |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110531431A (en) * | 2019-08-23 | 2019-12-03 | 唐山哈船科技有限公司 | A kind of marine geomagnetic day becomes observation device and magnetic survey method |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4033524A (en) * | 1971-07-21 | 1977-07-05 | Brice Neil M | Method and apparatus for controlling earth orbiting satellites and vehicles inter alia |
| US20060229813A1 (en) * | 2005-03-30 | 2006-10-12 | Tobiska William K | Ionospheric forecast system (IFS) |
| RU2322373C2 (en) * | 2005-11-22 | 2008-04-20 | Открытое акционерное общество "Ракетно-космическая корпорация "Энергия" им. С.П. Королева" | Method of control of spacecraft solar battery position and system for realization of this method |
| CN105303258B (en) * | 2015-10-30 | 2019-04-02 | 中国科学院国家天文台 | Solar burst event prediction method based on machine learning techniques forecasting model |
| CN109459027B (en) * | 2018-11-09 | 2022-06-10 | 北京航空航天大学 | Navigation method based on polarization-geomagnetic vector tight combination |
-
2020
- 2020-08-19 CN CN202010838295.5A patent/CN112033406B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110531431A (en) * | 2019-08-23 | 2019-12-03 | 唐山哈船科技有限公司 | A kind of marine geomagnetic day becomes observation device and magnetic survey method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112033406A (en) | 2020-12-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20190361972A1 (en) | Method, apparatus, device for table extraction based on a richly formatted document and medium | |
| CN107833208B (en) | Hyperspectral anomaly detection method based on dynamic weight depth self-encoding | |
| CN112033406B (en) | Navigation method, device and storage medium based on lightweight network | |
| CN106780631B (en) | Robot closed-loop detection method based on deep learning | |
| CN110929080B (en) | Optical remote sensing image retrieval method based on attention and generation countermeasure network | |
| CN109522832A (en) | It is a kind of based on order cloud sheet section matching constraint and track drift optimization winding detection method | |
| CN110245683B (en) | Residual error relation network construction method for less-sample target identification and application | |
| CN110991284B (en) | Optical remote sensing image statement description generation method based on scene pre-classification | |
| CN111753789A (en) | Robot vision SLAM closed loop detection method based on stack type combined self-encoder | |
| CN113838135B (en) | Pose estimation method, system and medium based on LSTM double-flow convolutional neural network | |
| CN116385761A (en) | 3D target detection method integrating RGB and infrared information | |
| CN110569387B (en) | Radar-image cross-modal retrieval method based on depth hash algorithm | |
| CN115641514A (en) | Pseudo visible light cloud map generation method for night sea fog monitoring | |
| CN117213470B (en) | Multi-machine fragment map aggregation updating method and system | |
| CN117041972A (en) | Channel-space-time attention self-coding based anomaly detection method for vehicle networking sensor | |
| CN117152630A (en) | Optical remote sensing image change detection method based on deep learning | |
| CN112069717A (en) | Magnetic storm prediction method and device based on multi-mode representation learning and storage medium | |
| CN115578574A (en) | Three-dimensional point cloud completion method based on deep learning and topology perception | |
| CN113435243A (en) | Hyperspectral true downsampling fuzzy kernel estimation method | |
| CN114140497A (en) | Target vehicle 3D real-time tracking method and system | |
| CN113011597A (en) | Deep learning method and device for regression task | |
| CN113850915A (en) | Vehicle tracking method based on Autoware | |
| CN113487374A (en) | Block E-commerce platform transaction system based on 5G network | |
| CN114693988B (en) | Satellite autonomous pose judging method, system and storage medium | |
| CN115995047B (en) | Star map identification method of double-view-field star sensor based on star template |
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 |