CN112132346B - Ship navigation track prediction method based on ship type - Google Patents
Ship navigation track prediction method based on ship type Download PDFInfo
- Publication number
- CN112132346B CN112132346B CN202011014416.0A CN202011014416A CN112132346B CN 112132346 B CN112132346 B CN 112132346B CN 202011014416 A CN202011014416 A CN 202011014416A CN 112132346 B CN112132346 B CN 112132346B
- Authority
- CN
- China
- Prior art keywords
- sequence
- ship
- time step
- decoder
- ship type
- 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
- 238000000034 method Methods 0.000 title claims abstract description 28
- 239000013598 vector Substances 0.000 claims abstract description 45
- 239000011159 matrix material Substances 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 5
- 238000005457 optimization Methods 0.000 claims description 3
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 2
- 239000003550 marker Substances 0.000 claims 2
- 238000010586 diagram Methods 0.000 description 3
- 230000002457 bidirectional effect Effects 0.000 description 2
- 235000019580 granularity Nutrition 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 206010000117 Abnormal behaviour Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9537—Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/28—Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- Strategic Management (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Development Economics (AREA)
- Computational Linguistics (AREA)
- Game Theory and Decision Science (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Remote Sensing (AREA)
- Traffic Control Systems (AREA)
Abstract
A ship navigation track prediction method based on ship types. The method comprises the following steps: acquiring AIS data of a ship to be predicted, wherein the AIS data comprises a navigation track point sequence and a ship type of the ship to be predicted; converting the navigation track point sequence into a sea area grid number sequence according to a sea area grid divided in advance; determining a ship type feature vector of the ship to be predicted according to the ship type of the ship to be predicted and a pre-established ship type dictionary matrix; inputting the sea area grid number sequence and the ship type feature matrix into a sequence model obtained by training in advance so as to obtain a predicted result sequence; and determining longitude and latitude coordinates of central points of all sea area grids in the predicted result sequence to obtain a predicted track point sequence of the ship to be predicted. The method and the device can improve the prediction precision of the ship navigation track and generate different track prediction results for different types of ships.
Description
Technical Field
The invention relates to the field of ship navigation, in particular to a ship navigation track prediction method based on ship types.
Background
The ship track prediction, i.e. predicting the future sailing track based on the current track of the ship, has wide application. For example, the track prediction information can be used for ship scheduling, collision among ships is effectively avoided, ship navigation safety is ensured, marine navigation efficiency is improved, abnormal behaviors of the ships can be detected, abnormal ships can be monitored, and further the method is helpful for the shore protection and the side sea protection of China.
In order to predict the ship track, an automatic ship identification system (Automatic Identification System, AIS) is generally needed, and the AIS is a set of digital equipment and navigation equipment system which utilize network, communication and electronic information display technologies, and can provide information including ship number, ship position, draft, speed, bow direction, ship type, ship length and width, and man-carrying number, and the like, which can be used as characteristic attributes of the ship track, so that a data source is provided for realizing ship track prediction.
At present, researches on ship track prediction exist, but due to the variety of ships, different kinds of ships have different sailing characteristics, such as changeable small ship tracks, flexible steering and faster sailing speed, and offshore sailing is often carried out; large ships have monotonous track, slow steering and slow speed, and are often sailed in ocean. The existing method cannot distinguish different types of ships, so that track prediction cannot be performed on different types of ships, and accurate navigation tracks are difficult to obtain.
Disclosure of Invention
The invention aims to provide a ship navigation track prediction method based on a ship type so as to improve the prediction accuracy of the ship navigation track.
In order to achieve the above object, an embodiment of the present invention provides a ship navigation track prediction method based on a ship type, the method including:
Acquiring AIS data of a ship to be predicted, wherein the AIS data comprises a navigation track point sequence and a ship type of the ship to be predicted;
Converting the navigation track point sequence into a sea area grid number sequence according to a sea area grid divided in advance;
Determining a ship type feature vector of the ship to be predicted according to the ship type of the ship to be predicted and a pre-established ship type dictionary matrix;
inputting the sea area grid number sequence and the ship type feature matrix into a sequence model obtained by training in advance so as to obtain a predicted result sequence;
and determining longitude and latitude coordinates of central points of all sea area grids in the predicted result sequence to obtain a predicted track point sequence of the ship to be predicted.
As can be seen from the technical solutions provided by the embodiments of the present disclosure, the ship navigation track prediction method provided by the present invention has at least the following beneficial effects:
The method converts the regression problem of accurate track prediction into the classification problem by dividing the sea area grids, reduces the difficulty of ship navigation track prediction, improves the accuracy, can track and predict the ship in real time, and has high universality and wide application range.
Compared with the traditional LSTM model, the method can predict a plurality of navigation track points of the ship at one time, rather than performing iterative prediction of the result, thereby ensuring that the prediction accuracy is not exponentially decreased.
The invention can focus on representative historical track points by combining the attention mechanism in the sequence-to-sequence model, such as identifying track inflection points, avoiding important features of dangerous reef areas and the like, and overcomes the defect of model precision reduction caused by the increase of the input track sequence.
The invention can distinguish track characteristics of different ship types, so as to generate different track prediction results aiming at different types of ships, thereby realizing personalized prediction.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings may be obtained according to the drawings without inventive effort to those skilled in the art.
FIG. 1 is a flow chart of a ship navigation track prediction method based on a ship type, which is provided by an embodiment of the invention;
FIG. 2 is a flowchart of an overall frame of a ship navigation track prediction method based on a ship type according to an embodiment of the present invention;
FIG. 3 is a block diagram of an algorithm model provided by an embodiment of the present invention;
fig. 4 is a schematic diagram of prediction of a ship navigation track according to an embodiment of the present invention.
Detailed Description
The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and the specific embodiments, it should be understood that these embodiments are only for illustrating the present invention and not for limiting the scope of the present invention, and various modifications of equivalent forms of the present invention will fall within the scope of the appended claims after reading the present invention.
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 invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
Referring to fig. 1, a flowchart of a ship navigation track prediction method based on a ship type according to an embodiment of the disclosure may include the following steps:
s1: and acquiring AIS data of the ship to be predicted, wherein the AIS data comprise a navigation track point sequence and the ship type of the ship to be predicted.
S2: and converting the navigation track point sequence into a sea area grid number sequence according to the sea area grid divided in advance.
Specifically, the sea area grid is obtained in advance through the following steps S21 to S24:
s21: and acquiring historical AIS data, wherein the historical AIS data comprises historical sailing track point sequences of a plurality of ships.
S22: and determining the navigation range of the ship according to the historical navigation track point sequence.
S23: and dividing the sea area in the navigation range into a plurality of sea area grids according to the prediction accuracy requirement.
S24: and numbering the divided sea area grids according to the row number and the column number of each sea area grid.
For example, each sea area grid may be a square sea area with a side length of 2 seas, and the individual sea area grids are numbered according to a two-dimensional coordinate system, e.g., the sea area grid of row 2 and column 10 is denoted by the number [2,10]. There is no intersection between sea area grids, the grid range can be set according to the prediction precision requirement, the grid range is small, the prediction precision is high, but the prediction time is slow, otherwise, the grid range is large, the prediction time is fast, but the precision is correspondingly reduced.
S3: and determining the ship type feature vector of the ship to be predicted according to the ship type of the ship to be predicted and a pre-established ship type dictionary matrix.
Specifically, the ship type dictionary matrix is composed of ship type feature vectors, the ship type feature vector of each ship type is generated according to a coded value corresponding to each ship type, and the coded value is used for determining a line number of the ship type feature vector in the ship type dictionary matrix.
For example, the ship type dictionary matrix is an n×m-dimensional matrix, n is the total number of ship types, m is a super parameter representing the characteristic dimension of each type of ship in the model, and m can be set to 128 or 256, or can be other values, which is not limited by the invention. Each row of the ship type dictionary matrix represents ship type feature vectors of one type of ship, and the numerical values of each row are not identical, so that different types of ships are distinguished.
In addition, according to different user demands and data volumes, ship type division with different granularities can be performed. For example, when the amount of track data per vessel is sufficient and the user needs to make an accurate track prediction, the vessel type classification may be made with the vessel side numbers, one for each specific vessel type. When the data volume is small or the user is in coarse granularity, the data volume can be divided according to the use type of the ship, such as a cruise ship, a fishing ship, a tanker, a cargo ship and the like, or the data volume can be divided in a clustering way according to the characteristics of the ship, such as the sailing speed, the carrying capacity, the voyage and the like.
Specifically, the ship type feature vector of the ship to be predicted may be determined by the following steps S31 to S33:
s31: and determining the coding value corresponding to the ship type of the ship to be predicted.
S32: and determining the line number of the ship type characteristic vector of the ship to be predicted in the ship type dictionary matrix according to the coding value.
S33: and addressing the ship type dictionary matrix according to the row number to acquire the ship type feature vector of the ship to be predicted.
For example, assuming that the ship type dictionary matrix is a 60×128-dimensional matrix, that is, a ship type dictionary matrix in which ship type feature vectors of 60 types of ships are stored, when it is determined that the code value corresponding to the ship type of the ship to be predicted is 5, the ship type feature vector of 1×128 in the 5 th row dimension is acquired from the ship type dictionary matrix.
S4: inputting the sea area grid number sequence and the ship type feature matrix into a sequence model (namely a Seq2Seq model) obtained through training in advance so as to obtain a predicted result sequence.
Specifically, referring to the training phase shown in fig. 2, the Seq2Seq model may be trained in advance by the following steps S41 to S47:
S41: converting a historical navigation track point sequence of a ship into a sea area grid number sequence, and dividing the sea area grid number sequence into an input grid sequence and an output grid sequence.
S42: and determining a ship type characteristic vector of the ship.
S43: the input grid sequence is input into an encoder of a Seq2Seq model to acquire hidden layer states of the encoder at each time step.
S44: and acquiring the coding vector corresponding to the next time step of the decoder according to the hidden layer state of the encoder in each time step and the hidden layer state of the decoder in the current time step.
S45: and determining an output result of the decoder in the next time step based on the ship type feature vector and the coding vector corresponding to the next time step of the decoder.
S46: repeating the steps to obtain the output results of the decoder in a plurality of time steps and generating an output result sequence.
S47: and determining the error of the Seq2Seq model according to the output grid sequence and the output result sequence to adjust and optimize parameters of the Seq2Seq model to obtain the trained Seq2Seq model.
Specifically, considering that the routes of the ship have the distinction of the round trip directions, which causes the round trip routes to have different navigation characteristics, the use of the bidirectional LSTM unit to obtain the characteristics of two directions at the same time affects the accuracy of the result, so that the encoder and the decoder of the Seq2Seq model in the present invention are LSTM units, and do not use the bidirectional LSTM units.
Specifically, an input grid sequence and an output grid sequence of each ship are determined through a sliding window algorithm, each sequence is composed of a plurality of time steps arranged according to a time sequence, wherein the input grid sequence sequentially comprises a grid number sequence covered by a sliding window, a track ending mark < stop > and a sequence ending mark < eos >; the output grid sequence comprises in order a sequence start tag < go >, a grid number sequence following the grid number sequence covered by the sliding window, an arrival port tag < port >, a track end tag < stop >, and a sequence end tag < eos >.
Specifically, referring to the algorithm model structure diagram shown in fig. 3, step S44 may include the following substeps S441 to S443:
s441: and calculating the association degree according to the hidden layer state of the encoder in each time step and the hidden layer state of the decoder in the current time step.
Wherein e t+1 in the above formula is a degree of correlation, A is a correlation operator,S t is the hidden layer state of the encoder at each time step, and s t is the hidden layer state of the decoder at the current time step.
S442: and performing soft maximization operation (i.e. softmax operation) on the association degree, and performing standardization to obtain Attention weights (i.e. Attention weights) corresponding to hidden layers of the encoder in each time step.
αt+1＝soft max(et+1)
Wherein e t+1 in the above formula is a degree of association, and α t+1 is an Attention weight.
S443: multiplying the hidden layer state of the encoder in each time step by the corresponding Attention weight to obtain the coding vector corresponding to the next time step of the decoder.
Wherein c t+1 in the above formula is a coding vector corresponding to a next time step of the decoder, h j is a hidden layer state of the encoder at a j time step, α tj is an Attention weight corresponding to the hidden layer state of the encoder at the j time step, and T x is a total number of time steps included in the input trellis sequence.
Specifically, step S45 may include the following substeps S451 to S452:
S451: and determining the hidden layer state of the decoder in the next time step according to the ship type feature vector, the coding vector corresponding to the next time step in the decoder, the output result of the decoder in the current time step and the hidden layer state of the decoder in the current time step.
st+1＝f(st,[yt,vi],ct+1)
Wherein s t+1 in the above formula is the hidden layer state of the decoder at the next time step, s t is the hidden layer state of the decoder at the current time step, y t is the output result of the decoder at the current time step, v i is the ship type feature vector, c t+1 is the coding vector corresponding to the next time step in the decoder, and [ y t,vi ] represents the concatenation of y t and v i.
S452: and determining an output result of the decoder in the next time step according to the ship type feature vector, the coding vector corresponding to the next time step in the decoder, the output state of the decoder in the current time step and the hidden layer state of the decoder in the next time step.
yt+1＝g([yt,vi],st+1,ct+1)
Wherein y t+1 in the above formula is an output result of the decoder at the next time step, y t is an output result of the decoder at the current time step, v i is a ship type feature vector, s t+1 is a hidden layer state of the decoder at the next time step, and c t+1 is a coding vector corresponding to the next time step in the decoder.
It should also be noted that the hidden layer state s 0 of the first time step of the decoder is defined by the hidden layer state of the last time step of the encoderMultiplying by a weight matrix w, i.e./>The output of the first time step of the decoder is the sequence start tag < go >.
It can be seen that the output result sequence is a predicted track sequence of the Seq2Seq model for the input grid sequence, and the generation error of the model can be calculated by using the predicted track sequence, the output grid sequence (real track sequence) and the loss function, so as to estimate the performance of the model, and the parameters of the model are optimized by a back propagation mode, wherein the loss function can be a cross entropy function, and the back propagation optimization algorithm can be an Adam algorithm. The Seq2Seq model will be able to generate a predicted trajectory sequence that approximates the real trajectory sequence as the model is continually trained and parameter tuning is performed to optimal performance. It should be noted that the predicted track sequence also includes an arrival port tag < port >, a track end tag < stop >, and a sequence end tag < eos >, which are included in the output grid sequence.
S5: and determining longitude and latitude coordinates of central points of all sea area grids in the predicted result sequence to obtain a predicted track point sequence of the ship to be predicted.
Referring to fig. 4, in A specific embodiment, the real-time sailing tracks of the cargo ship A and the tanker B are identical, and the passing sea grid sequences are [1,0], [2,0], [3,0] and [4,1], but the predicted tracks obtained by the sailing track prediction method provided by the invention are not identical, and the grid numbers in the predicted result sequence of the cargo ship A are sequentially [4,2], [4,3], [5,3], [6,3], [7,3], … …, [7,9], and the grid numbers in the predicted result sequence of the tanker B are sequentially [5,2], [6,3], [7,3], … …, [7,9].
The foregoing embodiments in the present specification are all described in a progressive manner, and the same and similar parts of the embodiments are mutually referred to, and each embodiment is mainly described in a different manner from other embodiments.
The foregoing description is only a few embodiments of the present invention, and the embodiments disclosed in the present invention are merely embodiments adopted for the purpose of facilitating understanding of the technical solutions of the present invention, and are not intended to limit the present invention. Any person skilled in the art can make any modification and variation in form and detail of the embodiments without departing from the spirit and scope of the present disclosure, but the scope of the present disclosure is still subject to the scope of the appended claims.
Claims (9)
1. A ship navigation track prediction method based on a ship type, comprising:
Acquiring AIS data of a ship to be predicted, wherein the AIS data comprises a navigation track point sequence and a ship type of the ship to be predicted;
Converting the navigation track point sequence into a sea area grid number sequence according to a sea area grid divided in advance;
determining a ship type feature vector of the ship to be predicted according to the ship type of the ship to be predicted and a pre-established ship type dictionary matrix, wherein the ship type dictionary matrix is composed of ship type feature vectors, the ship type feature vector of each ship type is generated according to a coding value corresponding to each ship type, and the coding value is used for determining a line number of the ship type feature vector in the ship type dictionary matrix;
inputting the sea area grid number sequence and the ship type feature matrix into a sequence model obtained by training in advance so as to obtain a predicted result sequence;
and determining longitude and latitude coordinates of central points of all sea area grids in the predicted result sequence to obtain a predicted track point sequence of the ship to be predicted.
2. The method according to claim 1, wherein the sea grid is obtained in advance by a method comprising:
Acquiring historical AIS data, wherein the historical AIS data comprises historical sailing track point sequences of a plurality of ships;
Determining the navigation range of the ship according to the historical navigation track point sequence;
dividing the sea area in the navigation range into a plurality of sea area grids according to the prediction accuracy requirement;
and numbering the divided sea area grids according to the row number and the column number of each sea area grid.
3. Method according to claim 1, characterized in that the ship type feature vector of the ship to be predicted is determined by:
Determining a coding value corresponding to the ship type of the ship to be predicted;
determining the line number of the ship type feature vector of the ship to be predicted in the ship type dictionary matrix according to the coding value;
and addressing the ship type dictionary matrix according to the row number to acquire the ship type feature vector of the ship to be predicted.
4. The method according to claim 1, wherein the sequence-to-sequence model is trained beforehand by:
Converting a historical navigation track point sequence of a ship into a sea area grid number sequence, and dividing the sea area grid number sequence into an input grid sequence and an output grid sequence;
Determining a ship type feature vector of the ship;
inputting the input grid sequence into an encoder of a sequence model to obtain hidden layer states of the encoder in each time step;
Acquiring a coding vector corresponding to the next time step of the decoder according to the hidden layer state of the encoder in each time step and the hidden layer state of the decoder in the current time step;
Determining an output result of the decoder in the next time step based on the ship type feature vector and the encoding vector corresponding to the next time step of the decoder;
repeating the steps to obtain the output results of the decoder in a plurality of time steps and generating an output result sequence;
And determining the error from the sequence to the sequence model according to the output grid sequence and the output result sequence so as to adjust the parameters of the sequence to the sequence model and obtain the trained sequence to the sequence model.
5. The method of claim 4, wherein the encoder and decoder of the sequence-to-sequence model are LSTM units.
6. The method of claim 4, wherein the input grid sequence and the output grid sequence for each vessel are determined by a sliding window algorithm, wherein the input grid sequence comprises in sequence a grid number sequence covered by a sliding window, a track end marker, and a sequence end marker; the output grid sequence sequentially comprises a sequence start mark, a grid number sequence after the grid number sequence covered by the sliding window, an arrival port mark, a track end mark and a sequence end mark.
7. The method of claim 4, wherein the obtaining the encoded vector corresponding to the next time step of the decoder based on the hidden layer state of the encoder at each time step and the hidden layer state of the decoder at the current time step comprises the steps of:
calculating the association degree according to the hidden layer state of the encoder in each time step and the hidden layer state of the decoder in the current time step;
Performing soft maximization operation on the association degree, and performing standardization to obtain attention weights corresponding to hidden states of the encoder in each time step;
multiplying the hidden layer state of the encoder at each time step by the corresponding attention weight to obtain the encoding vector corresponding to the next time step of the decoder.
8. The method of claim 4, wherein determining the output of the decoder at the next time step based on the ship type feature vector and the encoded vector corresponding to the next time step of the decoder comprises:
Determining the hidden layer state of the decoder in the next time step according to the ship type feature vector, the coding vector corresponding to the next time step in the decoder, the output result of the decoder in the current time step and the hidden layer state of the decoder in the current time step;
And determining an output result of the decoder in the next time step according to the ship type feature vector, the coding vector corresponding to the next time step in the decoder, the output state of the decoder in the current time step and the hidden layer state of the decoder in the next time step.
9. The method of claim 4, wherein the sequence-to-sequence model parameters are optimized by a back-propagation optimization algorithm, the back-propagation optimization algorithm comprising Adam's algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011014416.0A CN112132346B (en) | 2020-09-24 | 2020-09-24 | Ship navigation track prediction method based on ship type |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011014416.0A CN112132346B (en) | 2020-09-24 | 2020-09-24 | Ship navigation track prediction method based on ship type |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112132346A CN112132346A (en) | 2020-12-25 |
| CN112132346B true CN112132346B (en) | 2024-04-19 |
Family
ID=73839585
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011014416.0A Active CN112132346B (en) | 2020-09-24 | 2020-09-24 | Ship navigation track prediction method based on ship type |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112132346B (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112182133B (en) * | 2020-09-29 | 2022-02-15 | 南京北斗创新应用科技研究院有限公司 | AIS data-based ship loitering detection method |
| CN113312438B (en) * | 2021-03-09 | 2023-09-15 | 中南大学 | Marine target position prediction method integrating route extraction and trend judgment |
| CN112947485B (en) * | 2021-03-31 | 2022-10-21 | 亿海蓝（北京）数据技术股份公司 | Path planning method and system |
| CN113190636B (en) * | 2021-04-23 | 2023-08-04 | 亿海蓝（北京）数据技术股份公司 | Offshore road network construction method and system |
| CN113220678B (en) * | 2021-04-23 | 2023-09-08 | 亿海蓝（北京）数据技术股份公司 | Ship leaning event identification method and system |
| CN113221450B (en) * | 2021-04-27 | 2024-03-12 | 中国科学院国家空间科学中心 | Space-time prediction method and system for sparse non-uniform time sequence data |
| CN113240199B (en) * | 2021-06-07 | 2022-10-28 | 广西民族大学 | Port ship track prediction method based on DILATE _ TLSTM |
| CN113434617B (en) * | 2021-06-21 | 2022-05-13 | 武汉理工大学 | Behavior automatic division method and system based on ship track and electronic equipment |
| CN114282625A (en) * | 2021-12-29 | 2022-04-05 | 中国电子科技集团公司电子科学研究院 | Offshore target trajectory prediction method and device based on multi-source data |
| CN114596731B (en) * | 2022-03-18 | 2022-09-30 | 北京国交信通科技发展有限公司 | AIS-based ship navigation process meteorological sea condition data fusion processing system and method |
| CN115187005B (en) * | 2022-06-23 | 2023-04-18 | 中国人民公安大学 | Scheduling method, device, equipment and storage medium |
| CN115660027B (en) * | 2022-10-10 | 2023-05-23 | 北京邮电大学 | Multi-equipment sea area target data generation method and system supporting small samples |
| CN115618251B (en) * | 2022-11-03 | 2024-02-06 | 中国科学院空天信息创新研究院 | Ship track prediction method and device, electronic equipment and storage medium |
| CN115730742B (en) * | 2022-12-01 | 2024-01-16 | 中远海运科技股份有限公司 | On-flight container airliner route prediction method and system |
| CN115953746B (en) * | 2023-03-13 | 2023-06-02 | 中国铁塔股份有限公司 | Ship monitoring method and device |
| CN116911468B (en) * | 2023-09-12 | 2023-12-08 | 中国标准化研究院 | Global shipping information monitoring system based on big data |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106779184A (en) * | 2016-11-30 | 2017-05-31 | 中国矿业大学 | A kind of marine ships method for predicting based on big data modeling |
| KR101951142B1 (en) * | 2018-04-13 | 2019-02-21 | 한국해양과학기술원 | System for estimating loss data of a ship using machine learning |
| CN109817023A (en) * | 2019-03-06 | 2019-05-28 | 国网福建省电力有限公司莆田供电公司 | A kind of novel sea cable waters AIS object detection method |
| CN111401531A (en) * | 2020-04-24 | 2020-07-10 | 中国人民解放军国防科技大学 | Trajectory prediction method and system |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101703906B1 (en) * | 2015-01-22 | 2017-02-08 | 한국전자통신연구원 | Vessel monitoring system and vessel monitoring method thereof |
-
2020
- 2020-09-24 CN CN202011014416.0A patent/CN112132346B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106779184A (en) * | 2016-11-30 | 2017-05-31 | 中国矿业大学 | A kind of marine ships method for predicting based on big data modeling |
| KR101951142B1 (en) * | 2018-04-13 | 2019-02-21 | 한국해양과학기술원 | System for estimating loss data of a ship using machine learning |
| JP2019185785A (en) * | 2018-04-13 | 2019-10-24 | コリア インスティテュート オブ オーシャン サイエンス テクノロジー | Lost track data estimation system of marine vessel using machine learning |
| CN109817023A (en) * | 2019-03-06 | 2019-05-28 | 国网福建省电力有限公司莆田供电公司 | A kind of novel sea cable waters AIS object detection method |
| CN111401531A (en) * | 2020-04-24 | 2020-07-10 | 中国人民解放军国防科技大学 | Trajectory prediction method and system |
Non-Patent Citations (2)
| Title |
|---|
| Modified Minimal Resource Allocating Network for Ship Motion Predictive Control;Jianchuan Yin;《International Conference on Intelligent Control and Information Processing》;全文 * |
| 基于Spark 的船舶航行轨迹聚类方法;彭祥文;《中国航海》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112132346A (en) | 2020-12-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112132346B (en) | Ship navigation track prediction method based on ship type | |
| CN108197736A (en) | A kind of Air Quality Forecast method based on variation self-encoding encoder and extreme learning machine | |
| CN110147900A (en) | A kind of steamer line speed of a ship or plane multitask comprehensive optimization method | |
| CN107525509B (en) | Open water area sailing ship path planning method based on genetic algorithm | |
| JP6867898B2 (en) | Optimal route search method and equipment | |
| CN109489672A (en) | Consider the energy saving A star paths planning method of ocean current and unmanned boat kinetic effect | |
| CN115618251B (en) | Ship track prediction method and device, electronic equipment and storage medium | |
| JP2013134089A (en) | Optimal sailing route calculating apparatus and optimal sailing route calculating method | |
| CN111144666B (en) | Ocean thermocline prediction method based on deep space-time residual error network | |
| CN115131393A (en) | Trajectory prediction method, collision detection method, apparatus, electronic device, and medium | |
| Mehri et al. | A contextual hybrid model for vessel movement prediction | |
| KR102109571B1 (en) | System and method for providing optimized vessel seaway and computer-readable recording medium thereof | |
| CN117232520A (en) | Ship intelligent navigation system and navigation method suitable for offshore navigation | |
| CN112069371A (en) | Marine ship effective rescue searching method based on gridding neighborhood | |
| Anagnostopoulos | Big data techniques for ship performance study | |
| Campbell et al. | Envelope peaks over threshold (EPOT) application and verification | |
| Donandt et al. | Short-term inland vessel trajectory prediction with encoder-decoder models | |
| CN111824357B (en) | Test method, test device, electronic equipment and computer readable storage medium | |
| CN117474184A (en) | Ship track prediction method and system driven by dynamics knowledge | |
| CN117520874A (en) | Multi-mode prediction method and system for ship track | |
| CN115577868B (en) | Method and device for predicting destination port of in-transit ship, readable storage medium and ship | |
| KR102483608B1 (en) | Method and apparatus for generating optimum seaway using sub-map | |
| CN116700291A (en) | Path planning method, device, equipment and computer readable storage medium | |
| Vanem et al. | Collision damage stability of passenger ships: Holistic and risk-based approach | |
| Domala et al. | A Univariate and multivariate machine learning approach for prediction of significant wave height |
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 |