CN115223398A - Nuclear adaptive fractional order complex value neural network AIS interpolation method considering channel constraint - Google Patents
Nuclear adaptive fractional order complex value neural network AIS interpolation method considering channel constraint Download PDFInfo
- Publication number
- CN115223398A CN115223398A CN202210828991.7A CN202210828991A CN115223398A CN 115223398 A CN115223398 A CN 115223398A CN 202210828991 A CN202210828991 A CN 202210828991A CN 115223398 A CN115223398 A CN 115223398A
- Authority
- CN
- China
- Prior art keywords
- neural network
- ais
- ship
- track
- channel
- 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
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G3/00—Traffic control systems for marine craft
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Radar, Positioning & Navigation (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Remote Sensing (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Ocean & Marine Engineering (AREA)
- Feedback Control In General (AREA)
- Complex Calculations (AREA)
Abstract
The invention provides a nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraints, which utilizes the orthogonality of longitude and latitude to construct a complex neural network so as to solve the problem of track interpolation recovery of sensor data of an Automatic Identification System (AIS) of a ship. Wherein, the core self-adapting mechanism is to select the most suitable neural network core for different scenes. The ship is routed to a navigation channel, and when the ship navigates in the navigation channel, the track of the ship needs to be in a polygon formed by the navigation channel, which is one of the constraints of the neural network data. The longitude is used as the real part of the input, and the latitude is used as the imaginary part of the input, so that a complex number input is constructed. Since all longitudes and latitudes on earth are perpendicular to each other. It is thus satisfied that the real and imaginary parts of the complex number form a basis of orthogonal units.
Description
Technical Field
The invention belongs to the technical field of ship navigation track data processing and deep learning, and particularly relates to a nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint.
Background
In the big data era, with the development of new generation information technology, data sources and sensor networks are gradually expanded, and the amount, complexity and complexity of data are rapidly increased. More and more data has both spatiotemporal information, instant spatiotemporal data and spatiotemporal data. There is a pressing need to integrate and analyze spatiotemporal data to extract valuable information.
How to efficiently and accurately utilize AIS to carry out difference on ship tracks belongs to a problem to be solved in the field.
Disclosure of Invention
Aiming at the problems that the traditional space-time interpolation method has subjective influence on model selection and parameter estimation and complex nonlinear relation between space-time distance and space-time weight is difficult to fit, the invention provides a nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint. The method provides a kernel self-adaptive fractional complex value neural network, and a space-time autoregressive neural network interpolation method considering space-time autocorrelation.
The mapping from the time distance and the space distance to the space-time weight is realized by utilizing the self-learning and abstract capabilities of the neural network on the nonlinear relation, and the complicated parameter estimation process and subjective factors are avoided. The complex variable optimization problem with equality constraints can be solved.
The invention constructs a complex number neural network by utilizing the orthogonality of longitude and latitude so as to solve the problem of track interpolation recovery of the sensor data of an Automatic Identification System (AIS) of a ship.
Wherein, the core self-adapting mechanism is to select the most suitable neural network core for different scenes. A navigation channel is defined for a ship, and when the ship navigates in the navigation channel, the track of the ship needs to be in a polygon formed by the navigation channel, which is one of the constraints of the neural network data. The longitude is used as the real part of the input, and the latitude is used as the imaginary part of the input, so that a complex number input is constructed. Since all longitudes and latitudes on earth are perpendicular to each other. It is thus satisfied that the real and imaginary parts of the complex number form a basis of orthogonal units.
The invention specifically adopts the following technical scheme:
a nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint is characterized in that: based on AIS data, taking longitude as a real part of input and latitude as an imaginary part of input to construct complex number input; carrying out track interpolation recovery on the AIS sensor data of the automatic ship identification system by adopting a plurality of neural networks; and (3) determining a navigation channel for the ship, wherein when the ship navigates in the navigation channel, the track of the ship is in a polygon formed by the navigation channel, and the polygon is used as the constraint of the neural network data.
Further, the rule for performing data cleaning on the original AIS data is as follows: the included angle between the ground course COG and the real course in each AIS data record does not exceed 360 degrees; the ground speed SOG in each AIS data record does not exceed 50kn; a ship must navigate in a course without the ship's trajectory crossing the surrounding land and shorelines, islands or other obstacles.
Further, adjacent AIS data records should form an integral whole; data points of the front part and the rear part of the track are not required to be obviously deviated, and a 'jump' deviation is displayed; and the data points before and after the drift point should be considered as a whole.
Further, the training process of the complex neural network comprises the following steps:
step 1: inputting AIS track training samples and a track to be interpolated;
step 2: initializing weights of the neural network model;
and step 3: selecting the most suitable neural network kernel for different scenes;
and 4, step 4: calculating a neural network activation function;
and 5: calculating a neural network output;
step 6: calculating the output error of the neural network;
and 7: updating the weights of the neural network;
and 8: if the termination condition is not met, turning to step 4;
and step 9: performing AIS track interpolation by using the obtained neural network and the test AIS;
the input of the complex neural network is AIS data complex value, and the output is the position value of the interpolation point:
z＝z R +iz I (1)
wherein z is R Is the real part of the input vector; z is a radical of I Is the imaginary part of z;
is a neuron activation function; w is the weight connecting the hidden layer and the output layer, and represents the number of input nodes; w is a R Is the real part of w; w is a I Is the imaginary part of w; o denotes the output of the neural network:
o＝W(n)φ(z(n)) (3)
W(n)＝[w 1 (n),w 2 (n),…,w H (n)] (4)
w(n)＝w R (n)+w I (n) (5)。
further, when the ship navigates in the navigation channel, the track of the ship should be in a polygon formed by the navigation channel, and the constraint that the track is used as the neural network data specifically comprises the following steps:
for a ship sailing in a channel, the track of the ship needs to be in a polygon formed by the channel; if the point P is located within a quadrilateral whose vertices are a, B, C, and D, then the area a, B, C, D of the quadrilateral = the area P, a, B + the area P, B, C + the area P, C, D + the area P, D, a; otherwise it is not in the quadrilateral:
S ABCD ＝S ΔABZ +S ΔADZ +S ΔBCZ +S ΔCDZ (7)
representing a, b and c as the lengths of three sides of the triangle; the method for calculating the triangle area S is as follows:
the Z point is adopted to represent the position of the sailing boat, and the quadrangles of the vertexes A, B, C and D represent the channel boundary; the constraint condition is that the ship sails in the boundary of the channel:
further, in the error calculation, d is defined as the desired output of the neural network; d R Is the real part of d; d is a radical of I Is the imaginary part of d; the squared error function is defined as:
d＝d R +jd I (12)
g R the real part representing the error is:
g R the imaginary part representing the error is:
the first derivative of the real part of the error is:
the first derivative of the imaginary part of the error is:
the second derivative of the real part of the error is:
the second derivative of the imaginary part of the error is:
minimizing a loss function by adjusting the weight;
is denoted by w R The increment of the adjustment of (a) is,
is denoted by w I Adjustment increment of (c):
λ > 0 is a learning coefficient; the update rule of the neural network weight is as follows:
compared with the prior art, the method and the optimal scheme thereof realize the mapping from the time distance and the space distance to the time-space weight by utilizing the self-learning and abstract capabilities of the neural network on the nonlinear relation, and avoid the complicated parameter estimation process and subjective factors. It can solve the problem of complex variable optimization with equality constraints.
Drawings
The invention is described in further detail below with reference to the following figures and detailed description:
FIG. 1 is a schematic diagram of the overall design of an embodiment of the present invention;
FIG. 2 is a schematic diagram of a neural network training process according to an embodiment of the present invention;
FIG. 3 is a diagram of a neural network architecture according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the lane boundary and ship alignment scheme of the present invention;
FIG. 5 is a diagram illustrating a standard method for determining any point in an irregular closed area according to an embodiment of the present invention;
FIG. 6 is a schematic diagram of a sailing track and a channel of a ship according to an embodiment of the present invention.
FIG. 7 is a diagram illustrating interpolation results according to an embodiment of the present invention;
FIG. 8 is a diagram illustrating interpolation results of multiple quadratic kernels according to an embodiment of the present invention;
FIG. 9 is a graph of interpolation results for a Gaussian kernel according to an embodiment of the present invention;
FIG. 10 is a graph of interpolation results for a linear kernel according to an embodiment of the present invention;
FIG. 11 is a graph of interpolation results for cubic kernels, according to an embodiment of the present invention;
FIG. 12 is a graph illustrating the convergence of training errors according to an embodiment of the present invention.
Detailed Description
In order to make the features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in detail as follows:
it should be noted that the following detailed description is exemplary and is intended to provide further explanation of the disclosure. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments according to the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The ship is within the polygon boundary formed by the channel and is regarded as one of the targets for the channel to navigate. As shown in fig. 1, the embodiment takes longitude of the AIS sensor data as a real part of the input and latitude as an imaginary part of the input. A complex input of KAFCVNN is constructed. Since all longitudes and latitudes on earth are perpendicular to each other, this satisfies that the real part and imaginary part of the complex number form an orthogonal unit base. A number of factors are used as inputs to the neural network. It also takes into account the effects of other attributes in the same coordinate location.
The core adaptation mechanism is to select the most appropriate neural network core for different artificial intelligence scenarios. There is some noisy data in the original AIS data that needs to be cleaned up before the trace is restored.
First, the ground heading (COG) and the true heading in each AIS data record must not exceed 360 degrees. The ground Speed (SOG) in each AIS data record must not exceed 50kn.
Second, the vessel must navigate in a course where the vessel trajectory must not cross the surrounding land and shorelines, islands or other obstacles.
Third, the adjacent AIS data records should form an integral whole. The data points in the front and back portions of the trace should not deviate significantly, indicating a "jump" deviation. The data points before and after the drift point should be considered as a whole.
The training process of the neural network of the present embodiment is shown in fig. 2.
Step 1: inputting an AIS track training sample and a track to be interpolated;
step 2: neural network models such as weights are initialized. The structure of the neural network is shown in fig. 3.
Its input is the AIS data complex value, the output is the position value of the interpolation point:
z＝z R +iz I (1)
wherein z is R Is the real part of the input vector; z is a radical of formula I Is the imaginary part of z.
Is a neuron activation function. w is the weight connecting the hidden layer and the output layer, and represents the number of input nodes; w is a R Is the real part of w. w is a I Is the imaginary part of w. o denotes the output of the neural network:
o＝W(n)φ(z(n)) (3)
W(n)＝[w 1 (n),w 2 (n),…,w H (n)] (4)
w(n)＝w R (n)+w I (n) (5)
and step 3: selecting the most suitable neural network kernel for different scenes; as shown in fig. 8-11, selecting different cores has different effects, and the most suitable core can be confirmed through experiments;
and 4, step 4: calculating a neural network activation function;
and 5: calculating a neural network output;
step 6: calculating the output error of the neural network;
in this embodiment, the ship routing system is an important component of the ship traffic management system. When a ship is sailing in certain areas of the sea, the shore-based department specifies routes, routes or traffic separations in the form of regulations or recommendations. The scenario of the ship routing system is shown in fig. 4.
To determine whether a point is located in an irregular occlusion region, any rays that pass through the polygon may be drawn. If the ray passes through an odd segment, the point is inside the polygon. If the ray passes through an even segment, the point is outside the polygon. The criteria for judging any point in the irregular occlusion region are shown in fig. 5.
For a ship to travel within a channel, its trajectory needs to be within the polygon formed by the channel. If the point P is located within a quadrilateral whose vertices are a, B, C, and D, the area of the quadrilateral (a, B, C, D) = area (P, a, B) + area (P, B, C) + area (P, C, D) + area (P, D, a). Otherwise it is not in the quadrilateral.
S ABCD ＝S ΔABZ +S ΔADZ +S ΔBCZ +S ΔCDZ (7)
A, b and c are represented as the lengths of the three sides of the triangle. The method for calculating the triangle area S is as follows:
the schematic diagram of the ship's sailing track and channel is shown in fig. 6, the Z point represents the position of the sailing ship, and the quadrangle of the vertexes (i.e., a, B, C, and D) represents the channel boundary. The constraint condition is that the ship sails in the boundary of the channel:
in the error calculation, d is defined as the desired output of the neural network. d R Is the real part of d. d I Is the imaginary part of d. The squared error function is defined as:
d＝d R +jd I (12)
g R the real part representing the error is:
g R the imaginary part representing the error is:
the first derivative of the real part of the error is:
the first derivative of the imaginary part of the error is:
the second derivative of the real part of the error is:
the second derivative of the imaginary part of the error is:
by adjusting the weight, the loss function can be minimized.
Indicated as an adjustment increment of wR and,
denoted as adjustment increments of wI.
λ > 0 is a learning coefficient. The update rule of the neural network weight is as follows:
and 7: updating the weight of the neural network;
and step 8: if the termination condition is not met, turning to step 4;
and step 9: using the obtained neural network and the test AIS, AIS trajectory interpolation was performed, and the result is shown in fig. 7, and the error convergence curve is shown in fig. 12.
The present invention is not limited to the above preferred embodiments, and other various types of the core adaptive fractional order complex valued neural network AIS interpolation methods considering the channel constraint can be obtained by anyone who can derive the present invention, and all the equivalent changes and modifications made according to the claims of the present invention shall fall within the scope of the present invention.
Claims (6)
1. A nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint is characterized in that: constructing a complex number input by taking longitude as a real part of the input and latitude as an imaginary part of the input based on the AIS data; carrying out track interpolation recovery on the AIS sensor data of the automatic ship identification system by adopting a plurality of neural networks; and (3) determining a navigation channel for the ship, wherein when the ship navigates in the navigation channel, the track of the ship is in a polygon formed by the navigation channel, and the polygon is used as the constraint of the neural network data.
2. The AIS interpolation method for the nuclear adaptive fractional order complex valued neural network considering channel constraints as claimed in claim 1 is characterized in that: the rule for cleaning the data of the original AIS data is as follows: the included angle between the ground course COG and the real course in each AIS data record does not exceed 360 degrees; the ground speed SOG in each AIS data record does not exceed 50kn; a ship must navigate in a runway without the ship's trajectory crossing the surrounding land and shorelines, islands or other obstacles.
3. The AIS interpolation method for the nuclear adaptive fractional order complex valued neural network considering channel constraints as claimed in claim 2, wherein: the adjacent AIS data records should form an integral whole; data points of the front part and the rear part of the track are not required to be obviously deviated, and a 'jump' deviation is displayed; and the data points before and after the drift point should be considered as a whole.
4. The AIS interpolation method for the nuclear adaptive fractional order complex valued neural network considering channel constraints as claimed in claim 1, wherein: the training process of the complex neural network comprises the following steps:
step 1: inputting an AIS track training sample and a track to be interpolated;
step 2: initializing weights of the neural network model;
and step 3: selecting the most suitable neural network kernel for different scenes;
and 4, step 4: calculating a neural network activation function;
and 5: calculating the neural network output;
step 6: calculating the output error of the neural network;
and 7: updating the weights of the neural network;
and 8: if the termination condition is not met, turning to step 4;
and step 9: performing AIS track interpolation by using the obtained neural network and the test AIS;
the input of the complex neural network is AIS data complex value, and the output is the position value of the interpolation point:
z＝z R +iz I (1)
wherein z is R Is the real part of the input vector; z is a radical of I Is the imaginary part of z;
is a neuron activation function; w is the weight connecting the hidden layer and the output layer, and represents the number of input nodes; w is a R Is the real part of w; w is a I Is the imaginary part of w; o denotes the output of the neural network:
o＝W(n)φ(z(n)) (3)
W(n)＝[w 1 (n),w 2 (n),…,w H (n)] (4)
w(n)＝w R (n)+w I (n) (5)。
5. the AIS interpolation method for the nuclear adaptive fractional order complex valued neural network considering channel constraints as claimed in claim 4, wherein: when a ship navigates in a navigation channel, the track of the ship is in a polygon formed by the navigation channel, and the constraint taking the track as the neural network data specifically comprises the following steps:
for a ship sailing in a channel, the track of the ship needs to be in a polygon formed by the channel; if the point P is located within a quadrilateral whose vertices are a, B, C, and D, then the area a, B, C, D of the quadrilateral = the area P, a, B + the area P, B, C + the area P, C, D + the area P, D, a; otherwise it is not in the quadrilateral:
S ABCD ＝S ΔABZ +S ΔADZ +S ΔBCZ +S ΔCDZ (7)
representing a, b and c as the lengths of three sides of the triangle; the method for calculating the triangle area S is as follows:
the Z point is adopted to represent the position of the sailing boat, and the quadrangles of the vertexes A, B, C and D represent the channel boundary; the constraint is that the vessel travels within the corridor boundary:
6. the AIS interpolation method for the nuclear adaptive fractional order complex valued neural network considering channel constraints as claimed in claim 4, wherein:
in the error calculation, d is defined as the desired output of the neural network; d R Is the real part of d; d is a radical of I Is the imaginary part of d; the squared error function is defined as:
d＝d R +jd I (12)
g R the real part representing the error is:
g R the imaginary part representing the error is:
the first derivative of the real part of the error is:
the first derivative of the imaginary part of the error is:
the second derivative of the real part of the error is:
the second derivative of the imaginary part of the error is:
minimizing a loss function by adjusting the weights;
is denoted by w R The adjustment of (c) is increased by an amount,
is denoted by w I Adjustment increment of (c):
λ > 0 is a learning coefficient; the update rule of the neural network weight is as follows:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210828991.7A CN115223398B (en) | 2022-07-14 | 2022-07-14 | Nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210828991.7A CN115223398B (en) | 2022-07-14 | 2022-07-14 | Nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115223398A true CN115223398A (en) | 2022-10-21 |
| CN115223398B CN115223398B (en) | 2023-09-19 |
Family
ID=83611236
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210828991.7A Active CN115223398B (en) | 2022-07-14 | 2022-07-14 | Nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115223398B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170303079A1 (en) * | 2015-07-24 | 2017-10-19 | Hitachi, Ltd. | Information distribution apparatus and method |
| CN109214107A (en) * | 2018-09-26 | 2019-01-15 | 大连海事大学 | A kind of ship's navigation behavior on-line prediction method |
| CN109948207A (en) * | 2019-03-06 | 2019-06-28 | 西安交通大学 | A kind of aircraft engine high pressure rotor rigging error prediction technique |
| CN111611300A (en) * | 2020-05-27 | 2020-09-01 | 中航信移动科技有限公司 | Flight air hover state judgment method |
| CN112561148A (en) * | 2020-12-09 | 2021-03-26 | 北京京航计算通讯研究所 | Ship track prediction method and system based on one-dimensional convolutional neural network and LSTM |
| CN113537020A (en) * | 2021-07-07 | 2021-10-22 | 中国人民解放军国防科技大学 | Complex SAR image target identification method based on improved neural network |
-
2022
- 2022-07-14 CN CN202210828991.7A patent/CN115223398B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20170303079A1 (en) * | 2015-07-24 | 2017-10-19 | Hitachi, Ltd. | Information distribution apparatus and method |
| CN109214107A (en) * | 2018-09-26 | 2019-01-15 | 大连海事大学 | A kind of ship's navigation behavior on-line prediction method |
| CN109948207A (en) * | 2019-03-06 | 2019-06-28 | 西安交通大学 | A kind of aircraft engine high pressure rotor rigging error prediction technique |
| CN111611300A (en) * | 2020-05-27 | 2020-09-01 | 中航信移动科技有限公司 | Flight air hover state judgment method |
| CN112561148A (en) * | 2020-12-09 | 2021-03-26 | 北京京航计算通讯研究所 | Ship track prediction method and system based on one-dimensional convolutional neural network and LSTM |
| CN113537020A (en) * | 2021-07-07 | 2021-10-22 | 中国人民解放军国防科技大学 | Complex SAR image target identification method based on improved neural network |
Non-Patent Citations (2)
| Title |
|---|
| 何正伟;杨帆;刘力荣;: "基于AIS数据的船舶安全航行水深参考图", 交通运输工程学报, vol. 18, no. 4, pages 171 - 181 * |
| 张雷 等: "基于复值神经网络的分式规划问题", 《西南师范大学学报(自然科学版)》, vol. 44, no. 5, pages 1 - 6 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115223398B (en) | 2023-09-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110210551B (en) | Visual target tracking method based on adaptive subject sensitivity | |
| Li et al. | GPS/INS/Odometer integrated system using fuzzy neural network for land vehicle navigation applications | |
| Li et al. | Autonomous underwater vehicle optimal path planning method for seabed terrain matching navigation | |
| CN107504972A (en) | A kind of aircraft's flight track method and device for planning based on dove group's algorithm | |
| CN111681178B (en) | Knowledge distillation-based image defogging method | |
| JP6865364B2 (en) | A learning method and learning device for improving segmentation performance used in detecting events including pedestrian events, automobile events, falling events, and fallen events using edge loss, and a test method and test device using the learning method and learning device. | |
| CN110108292B (en) | Vehicle navigation path planning method, system, device and medium | |
| CN111340868B (en) | Unmanned underwater vehicle autonomous decision control method based on visual depth estimation | |
| Zheng et al. | A Decision‐Making Method for Ship Collision Avoidance Based on Improved Cultural Particle Swarm | |
| CN109858137B (en) | Complex maneuvering aircraft track estimation method based on learnable extended Kalman filtering | |
| CN111507906B (en) | Method and apparatus for debouncing with neural networks for fault tolerance and surge robustness | |
| CN114199248B (en) | AUV co-location method for optimizing ANFIS based on mixed element heuristic algorithm | |
| CN111507459A (en) | Method and apparatus for reducing annotation cost of neural networks | |
| US20230243658A1 (en) | Systems, Methods and Devices for Map-Based Object's Localization Deep Learning and Object's Motion Trajectories on Geospatial Maps Using Neural Network | |
| CN110444022A (en) | The construction method and device of traffic flow data analysis model | |
| CN108444479A (en) | Gravity Matching method based on ADAPTIVE ROBUST Unscented kalman filtering | |
| CN113190636A (en) | Offshore road network construction method and system | |
| CN114152257A (en) | Ship prediction navigation method based on attention mechanism and environment perception LSTM | |
| CN115601657A (en) | Method for detecting and identifying ship target in severe weather | |
| Liu et al. | Application of dijkstra algorithm in path planning for geomagnetic navigation | |
| CN108415445A (en) | A kind of submarine navigation device positioning robust Optimal methods based on switch constraint | |
| Lin et al. | Research on UUV obstacle avoiding method based on recurrent neural networks | |
| CN105825239A (en) | Multi-sensor track fusion method based on sparse expression | |
| CN115223398B (en) | Nuclear self-adaptive fractional order complex value neural network AIS interpolation method considering channel constraint | |
| Shi et al. | Cooperative flow field estimation using multiple AUVs |
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 |