CN109001722B - Ship track data fusion method based on LSTM model - Google Patents

Ship track data fusion method based on LSTM model Download PDF

Info

Publication number
CN109001722B
CN109001722B CN201810537118.6A CN201810537118A CN109001722B CN 109001722 B CN109001722 B CN 109001722B CN 201810537118 A CN201810537118 A CN 201810537118A CN 109001722 B CN109001722 B CN 109001722B
Authority
CN
China
Prior art keywords
time
coordinate information
ship
collection
ais
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
Application number
CN201810537118.6A
Other languages
Chinese (zh)
Other versions
CN109001722A (en
Inventor
谢磊
陆楠楠
张笛
张金奋
包竹
薛双飞
夏文涛
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University of Technology WUT
Original Assignee
Wuhan University of Technology WUT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Wuhan University of Technology WUT filed Critical Wuhan University of Technology WUT
Priority to CN201810537118.6A priority Critical patent/CN109001722B/en
Publication of CN109001722A publication Critical patent/CN109001722A/en
Application granted granted Critical
Publication of CN109001722B publication Critical patent/CN109001722B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/02—Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems
    • G01S13/50—Systems of measurement based on relative movement of target
    • G01S13/58—Velocity or trajectory determination systems; Sense-of-movement determination systems
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/86—Combinations of radar systems with non-radar systems, e.g. sonar, direction finder
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
    • G01S19/38—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system
    • G01S19/39—Determining a navigation solution using signals transmitted by a satellite radio beacon positioning system the satellite radio beacon positioning system transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
    • G01S19/42—Determining position
    • G01S19/45—Determining position by combining measurements of signals from the satellite radio beacon positioning system with a supplementary measurement

Landscapes

  • Engineering & Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Radar Systems Or Details Thereof (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention provides a ship track data fusion method based on an LSTM model. According to the invention, ship navigation tracks acquired by an AIS system, a shipborne radar and a high-precision GPS (global positioning system) in one navigation of a ship are collected as historical data, dynamic longitude and latitude coordinate information of the ship in one section of navigation process is acquired by the AIS system, and dynamic image coordinate information of the ship in one section of navigation process is acquired by the shipborne radar and is used as data to be trained; respectively normalizing the radar longitude and latitude coordinate information, the AIS coordinate information and the high-precision GPS longitude and latitude coordinate information; grouping the ship navigation track coordinate information collected by the normalized ship-borne radar, the AIS system and the high-precision GPS to obtain grouped coordinate information; and performing preliminary training on the LSTM model by using the radar and AIS coordinate information after the grouping normalization as a training set, and then adjusting the weight parameters of the LSTM model by using the grouped and normalized high-precision GPS coordinate information. The invention has the advantage of improving the integration precision of the ship track.

Description

Ship track data fusion method based on LSTM model
Technical Field
The invention relates to the field of ship information perception and fusion, in particular to a ship track data fusion method based on an LSTM model.
Background
The world enters the development stage of 'industry 4.0', and key descriptions are made on high-technology ships, intelligent ship manufacturing and the like. Nowadays, economy develops rapidly, marine ships develop rapidly towards the direction of science and technology and rapidity, the quantity of ships, particularly the quantity of ships for transporting dangerous goods, increases continuously, the density of ships increases continuously, the potential safety hazard of marine traffic is increased rapidly directly due to the reasons, sea damage accidents occur frequently, life and property are threatened greatly, and marine environmental pollution is caused to a great extent. In order to ensure the safety of marine navigation and improve the efficiency of shipping management, the requirements on communication navigation equipment are continuously improved.
The modern Radar technology ARPA (Automatic Radar Plotting instruments) or TT (Target Tracking) can automatically record and track a Target, can provide dynamic information of the Target such as ship position, navigational speed, course and the like, and can display a relatively comprehensive traffic situation image so as to obtain the distance between ships. In practical application, the radar carries out target detection by analyzing the echo signal information of ships, and a certain detection blind area exists. In addition, because the microwave target operates in the microwave band, the microwave target is easily interfered by terrain environment and meteorological conditions, and the precision is limited, and target missing detection and false target phenomena can occur. The AIS provides rich information and can provide real-time dynamic information such as ship position, speed, heading, UTC (universal coordinated time), etc. AIS adopts radio wave propagation, works in very High frequency VHF (very High frequency) maritime frequency band, has a long working distance due to strong wave diffraction capability, and works without being influenced by distance and position due to positioning data from a Global Navigation Satellite System (GNSS), more importantly, the positioning data is hardly limited by weather, and provides a large amount of information and High precision. However, AIS also has the disadvantage that the device cannot actively acquire specific target information because it is characterized by passive reception of information; the AIS provides a ship position as a coordinate point, which cannot acquire video information and surrounding environment information of a target.
Therefore, the AIS cannot be considered as the only navigation collision avoidance device during the flight. The AIS equipment is used for assisting the radar, and information after the AIS equipment and the AIS equipment are fused is displayed on the chart, so that not only can a very complex display picture be avoided, the automatic identification of a radar target be realized, the identification difficulty of an operator be solved, but also data after the AIS equipment and the chart are fused are more accurate, and the reliability is higher. The advantages of the radar and the AIS information are complementary, so that the maritime traffic safety is guaranteed to a greater extent.
Disclosure of Invention
In order to solve the technical problem, the invention provides a ship track data fusion method based on an LSTM model.
The technical scheme of the invention is a ship track data fusion method based on an LSTM model, which comprises the following steps:
step 1, collecting ship navigation tracks collected by an AIS system, a shipborne radar and a high-precision GPS (global positioning system) of a ship in one navigation as historical data, collecting dynamic longitude and latitude coordinate information of the ship in a section of navigation process through the AIS system, and collecting dynamic image coordinate information of the ship in a section of navigation process through the shipborne radar as data to be trained;
step 2: respectively normalizing the radar longitude and latitude coordinate information, the AIS coordinate information and the high-precision GPS longitude and latitude coordinate information in the step 1;
step 3, grouping coordinate information is obtained by grouping the ship navigation track coordinate information acquired by the ship-borne radar, the AIS system and the high-precision GPS after normalization in the step 2 into one group every five time points according to the time sequence;
and 4, step 4: and performing preliminary training on the LSTM model by using the normalized grouped radar and AIS coordinate information as a training set, and then adjusting the weight parameters of the LSTM model by using the grouped and normalized high-precision GPS coordinate information as a test set through error reverse transmission to obtain the trained LSTM model.
Preferably, in the step 1, the dynamic longitude and latitude coordinate information of the ship during sailing is collected through the AIS, meanwhile, image information collected by the radar needs to be converted into coordinate information matched with the AIS information, and the track coordinate information is arranged according to a time sequence;
in the step 1, the AIS system acquires dynamic position information of a ship as dynamic longitude and latitude coordinate information:
Stime+i*Δt＝(longtime+i*Δt,lattime+i*Δt)i∈[0,N]
wherein time is the collection start time, delta t is the collection interval, N is the number of collection points, Stime+i*ΔtCollecting alternate dynamic latitude and longitude coordinate information for ith, Longtime+i*ΔtCollecting the longitude coordinate information of interval for ith, lattime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
in the step 1, the dynamic position information of the ship-borne radar collected ship is image information which is converted into longitude and latitude coordinate information matched with the AIS dynamic position information through the inkcard support projection:
Ltime+i*Δt＝(xtime+i*Δt,ytime+i*Δt)i∈[0,N]
wherein, time is the collection start time, delta t is the collection interval, N is the collection point number, Ltime+i*ΔtDynamic image coordinate information, x, for the ith acquisition intervaltime+i*ΔtCollecting interval longitude coordinate information for ithtime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
in the step 1, the high-precision GPS acquires dynamic position information of the ship as high-precision longitude and latitude coordinate information:
Gtime+i*Δt＝(ptime+i*Δt,qtime+i*Δt)i∈[0,N]
wherein time is the collection start time, delta t is the collection interval, N is the number of collection points, Gtime+i*ΔtHigh precision motion picture coordinate information for ith acquisition interval, ptime+i*ΔtCollecting longitude coordinate information of interval for ith, qtime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
preferably, the longitude and latitude coordinate information of the grouped radar in the step 2 is normalized as follows:
Figure BDA0001678448830000031
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000032
acquiring the radar coordinate information of the interval for the ith normalized acquisition;
and step 2, normalizing the AIS coordinate information into:
Figure BDA0001678448830000033
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000034
the normalized AIS coordinate information of the ith acquisition interval is acquired;
and 2, normalizing the grouped high-precision GPS coordinate information into:
Figure BDA0001678448830000035
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000041
the coordinate information acquired by the high-precision GPS at the ith acquisition interval after normalization;
preferably, the normalized coordinate information collected by the radar in step 3 is grouped into one group at every five time points according to the time sequence:
Figure BDA0001678448830000042
wherein, tim is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000043
for the nth coordinate information in the jth group, WLjThe normalized j-th grouping coordinate information is obtained;
the coordinate information collected by the normalized AIS system in the step 3 is grouped into a group according to five time points in the time sequence:
Figure BDA0001678448830000044
wherein, tim is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000045
for the nth coordinate information in the jth group, WSjThe normalized j-th grouping coordinate information is obtained;
in step 3, the normalized high-precision GPS coordinate information is grouped into a group of every five time points according to the time sequence:
Figure BDA0001678448830000046
wherein, tim is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000047
for the nth coordinate information in the jth group, WGjThe normalized j-th grouping coordinate information is obtained;
preferably, the LSTM model described in step 4 is a structure having α input layers, β hidden layers, and γ output layers, wherein the LSTM learning process focuses on the hidden layers using LSTM-specific cell units having an input gate, a forgetting gate, and an output gate;
the input gate indicates whether new data information is allowed to be added into the current hidden layer node, if the value is 1, input is allowed, and if the value is 0, input is not allowed;
the forgetting gate represents whether historical data stored by the current hidden layer node is reserved or not, if the value is 1, the gate is opened, the historical data is reserved, and if the value is 0, the gate is closed, the historical data stored by the current node is emptied;
the output gate represents whether the output value of the current node is output to the next layer, namely the next hidden layer or the output layer, if the output value is 1, namely the gate is opened, the output value of the current node acts on the next layer, and if the output value is 0, namely the gate is closed, the output value of the current node is not output;
the LSTM model in step 4 has the key formula as follows:
xtime+(i+1)*Δt＝[WLj,WSj
htime+i*Δt＝[WLj,WSj,WGj
ftime+(i+1)*Δt＝σ(Wf·[Ctime+i*Δt,htime+i*Δt,xtime+(i+1)*Δt]+bf)
itime+(i+1)*Δt＝σ(Wi·[Ctime+i*Δt,htime+i*Δt,xtime+(i+1)*Δt]+bi)
Otime+(i+1)*Δt＝σ(Wo·[Ctime+(i+1)*Δt,htime+i*Δt,xtime+(i+1)*Δt]+bo)
Ctime+(i+1)*Δt＝ftime+(i+1)*Δt*Ctime+i*Δt+itime+(i+1)*Δt
*tanh(Wc[htime+i*Δt,xtime+(i+1)*Δt]+bc)
htime+(i+1)*Δt＝Otime+(i+1)*Δttanh(Ctime+(i+1)*Δt)
wherein, the concrete meaning of the formula is as follows, f represents a forgetting gate, i represents an input gate, O represents an output gate, C represents a unit activation vector, h represents a hidden layer unit, W represents a hidden layer unitfIs a weight matrix between the forgetting gate and other input vectors, WiIs a weight matrix between the input gate and other vectors receiving input data, WoIs a weight matrix between the output gate and the data before the output gate, WcActivating a weight matrix among other vectors for a cellThe activation function is tanh, bf、bi、bo、bcRespectively, the deviation values of the forgetting gate, the input gate, the output gate and the unit activation vector, and different sampling moments are indicated by subscript t, wherein htime+(i+1)*ΔtIs given by the hidden layer h at the last momenttime+i*ΔtWith new input xtime+(i+1)*ΔtAnd the self state C of the cell at the last momenttime+i*ΔtCompounded by gate control, at the last instant htime+(i+1)*ΔtI.e. the output.
Compared with the prior art, the method has the advantages that the information loss is minimum, the processed information amount is maximum, and the fusion performance is best, so that the real track of the ship motion is better determined.
Drawings
FIG. 1: the invention is a principle flow chart;
FIG. 2: adding a cell data relation graph in LSTM of a peepole structure;
FIG. 3: schematic diagram of cell unit structure in LSTM.
Detailed Description
In order to facilitate the understanding and implementation of the present invention for those of ordinary skill in the art, the present invention is further described in detail with reference to the accompanying drawings and examples, it is to be understood that the embodiments described herein are merely illustrative and explanatory of the present invention and are not restrictive thereof.
An embodiment of the present invention is described below with reference to fig. 1 to 3, and specifically includes the following steps:
step 1, collecting ship navigation tracks collected by an AIS system, a shipborne radar and a high-precision GPS (global positioning system) of a ship in one navigation as historical data, collecting dynamic longitude and latitude coordinate information of the ship in a section of navigation process through the AIS system, and collecting dynamic image coordinate information of the ship in a section of navigation process through the shipborne radar as data to be trained;
acquiring dynamic longitude and latitude coordinate information of a ship in navigation through the AIS in the step 1, converting image information acquired by a radar into coordinate information matched with the AIS information, and arranging track coordinate information according to a time sequence;
in the step 1, the AIS system acquires dynamic position information of a ship as dynamic longitude and latitude coordinate information:
Stime+i*Δt＝(longtime+i*Δt,lattime+i*Δt)i∈[0,N]
wherein time is the collection start time, delta t is the collection interval, N is the number of collection points, Stime+i*ΔtCollecting alternate dynamic latitude and longitude coordinate information for ith, Longtime+i*ΔtCollecting the longitude coordinate information of interval for ith, lattime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
in the step 1, the dynamic position information of the ship-borne radar collected ship is image information which is converted into longitude and latitude coordinate information matched with the AIS dynamic position information through the inkcard support projection:
Ltime+i*Δt＝(xtime+i*Δt,ytime+i*Δt)i∈[0,N]
wherein, time is the collection start time, delta t is the collection interval, N is the collection point number, Ltime+i*ΔtDynamic image coordinate information, x, for the ith acquisition intervaltime+i*ΔtCollecting interval longitude coordinate information for ithtime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
in the step 1, the high-precision GPS acquires dynamic position information of the ship as high-precision longitude and latitude coordinate information:
Gtime+i*Δt＝(ptime+i*Δt,qtime+i*Δt)i∈[0,N]
wherein time is the collection start time, delta t is the collection interval, N is the number of collection points, Gtime+i*ΔtHigh precision motion picture coordinate information for ith acquisition interval, ptime+i*ΔtCollecting longitude coordinate information of interval for ith, qtime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
step 2: respectively normalizing the radar longitude and latitude coordinate information, the AIS coordinate information and the high-precision GPS longitude and latitude coordinate information in the step 1;
and 2, normalizing the longitude and latitude coordinate information of the grouped radar into:
Figure BDA0001678448830000071
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000072
acquiring the radar coordinate information of the interval for the ith normalized acquisition;
and step 2, normalizing the AIS coordinate information into:
Figure BDA0001678448830000073
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000074
the normalized AIS coordinate information of the ith acquisition interval is acquired;
and 2, normalizing the grouped high-precision GPS coordinate information into:
Figure BDA0001678448830000081
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000082
the coordinate information acquired by the high-precision GPS at the ith acquisition interval after normalization;
step 3, grouping coordinate information is obtained by grouping the ship navigation track coordinate information acquired by the ship-borne radar, the AIS system and the high-precision GPS after normalization in the step 2 into one group every five time points according to the time sequence;
in step 3, the normalized coordinate information collected by the radar is grouped into a group according to five time points in the time sequence:
Figure BDA0001678448830000083
wherein, tim is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000084
for the nth coordinate information in the jth group, WLjThe normalized j-th grouping coordinate information is obtained;
the coordinate information collected by the normalized AIS system in the step 3 is grouped into a group according to five time points in the time sequence:
Figure BDA0001678448830000085
wherein, tim is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000086
for the nth coordinate information in the jth group, WSjThe normalized j-th grouping coordinate information is obtained;
in step 3, the normalized high-precision GPS coordinate information is grouped into a group of every five time points according to the time sequence:
Figure BDA0001678448830000087
wherein, tim is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure BDA0001678448830000091
for the nth coordinate information in the jth group, WGjIs made ofNormalized j-th grouping coordinate information;
and 4, step 4: by normalizing the grouped radar and AIS coordinate information WLj，WSjThe LSTM model is initially trained as a training set and then the high-precision GPS coordinate information WG subjected to grouping normalization is usedjAs a test set, adjusting the weight parameters of the LSTM model through error reverse transfer to obtain a trained LSTM model;
the LSTM model described in step 4 is a structure having α ═ 12 input layers, β ═ 18 hidden layers, and γ ═ 4 output layers, where the LSTM learning process focuses on the hidden layers using LSTM-specific cell units having input gates, forgetting gates, and output gates;
the input gate indicates whether new data information is allowed to be added into the current hidden layer node, if the value is 1, input is allowed, and if the value is 0, input is not allowed;
the forgetting gate represents whether historical data stored by the current hidden layer node is reserved or not, if the value is 1, the gate is opened, the historical data is reserved, and if the value is 0, the gate is closed, the historical data stored by the current node is emptied;
the output gate represents whether the output value of the current node is output to the next layer, namely the next hidden layer or the output layer, if the output value is 1, namely the gate is opened, the output value of the current node acts on the next layer, and if the output value is 0, namely the gate is closed, the output value of the current node is not output;
the LSTM model in step 4 has the key formula as follows:
xtime+(i+1)*Δt＝[WLj,WSj]
htime+i*Δt＝[WLj,WSj,WGj]
ftime+(i+1)*Δt＝σ(Wf·[Ctime+i*Δt,htime+i*Δt,xtime+(i+1)*Δt]+bf)
itime+(i+1)*Δt＝σ(Wi·[Ctime+i*Δt,htime+i*Δt,xtime+(i+1)*Δt]+bi)
Otime+(i+1)*Δt＝σ(Wo·[Ctime+(i+1)*Δt,htime+i*Δt,xtime+(i+1)*Δt]+bo)
Ctime+(i+1)*Δt＝ftime+(i+1)*Δt*Ctime+i*Δt+itime+(i+1)*Δt
*tanh(Wc[htime+i*Δt,xtime+(i+1)*Δt]+bc)
htime+(i+1)*Δt＝Otime+(i+1)*Δttanh(Ctime+(i+1)*Δt)
wherein, the concrete meaning of the formula is as follows, f represents a forgetting gate, i represents an input gate, O represents an output gate, C represents a unit activation vector, h represents a hidden layer unit, W represents a hidden layer unitfIs a weight matrix between the forgetting gate and other input vectors, WiIs a weight matrix between the input gate and other vectors receiving input data, WoIs a weight matrix between the output gate and the data before the output gate, WcFor a unit, the activation function is tanh, bf、bi、bo、bcRespectively, the deviation values of the forgetting gate, the input gate, the output gate and the unit activation vector, and different sampling moments are indicated by subscript t, wherein htime+(i+1)*ΔtIs given by the hidden layer h at the last momenttime+i*ΔtWith new input xtime+(i+1)*ΔtAnd the self state C of the cell at the last momenttime+i*ΔtCompounded by gate control, at the last instant htime+(i+1)*ΔtI.e. the output.
And inputting ship navigation track data acquired by the AIS system to be fused, the shipborne radar and the high-precision GPS into the LSTM model after training, and learning the data to be fused by the LSTM model after training to finally obtain the actual ship motion track after fusion.
It should be understood that parts of the specification not set forth in detail are well within the prior art.
It should be understood that the above description of the preferred embodiments is given for clarity and not for any purpose of limitation, and that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the invention as defined by the appended claims.

Claims (4)

1. A ship track data fusion method based on an LSTM model is characterized by comprising the following steps:
step 1, collecting ship navigation tracks collected by an AIS system, a shipborne radar and a high-precision GPS (global positioning system) of a ship in one navigation as historical data, collecting dynamic longitude and latitude coordinate information of the ship in a section of navigation process through the AIS system, and collecting dynamic image coordinate information of the ship in a section of navigation process through the shipborne radar as data to be trained;
step 2: respectively normalizing the radar longitude and latitude coordinate information, the AIS coordinate information and the high-precision GPS longitude and latitude coordinate information in the step 1;
step 3, grouping coordinate information is obtained by grouping the ship navigation track coordinate information acquired by the ship-borne radar, the AIS system and the high-precision GPS after normalization in the step 2 into one group every five time points according to the time sequence;
and 4, step 4: carrying out preliminary training on the LSTM model by using the radar and AIS coordinate information after the grouping normalization as a training set, and then adjusting the weight parameters of the LSTM model by using the high-precision GPS coordinate information after the grouping normalization as a testing set through error reverse transmission to obtain the trained LSTM model;
and 5: and inputting ship navigation track data acquired by the AIS system to be fused, the shipborne radar and the high-precision GPS into the LSTM model after training, and learning the data to be fused by the LSTM model after training to finally obtain the actual ship motion track after fusion.
2. The LSTM model-based ship track data fusion method of claim 1, wherein in step 1, the AIS is used to collect dynamic longitude and latitude coordinate information of ship during navigation, and image information collected by radar is converted into coordinate information matching with AIS information, and the track coordinate information is arranged according to time sequence;
in the step 1, the AIS system acquires dynamic position information of a ship as dynamic longitude and latitude coordinate information:
Stime+i*Δt＝(longtime+i*Δt,lattime+i*Δt)i∈[0,N]
wherein time is the collection start time, delta t is the collection interval, N is the number of collection points, Stime+i*ΔtCollecting alternate dynamic latitude and longitude coordinate information for ith, Longtime+i*ΔtCollecting the longitude coordinate information of interval for ith, lattime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
in the step 1, the dynamic position information of the ship-borne radar collected ship is image information which is converted into longitude and latitude coordinate information matched with the AIS dynamic position information through the inkcard support projection:
Ltime+i*Δt＝(xtime+i*Δt,ytime+i*Δt)i∈[0,N]
wherein, time is the collection start time, delta t is the collection interval, N is the collection point number, Ltime+i*ΔtDynamic image coordinate information, x, for the ith acquisition intervaltime+i*ΔtCollecting interval longitude coordinate information for ithtime+i*ΔtAcquiring alternate latitude coordinate information for the ith;
in the step 1, the high-precision GPS acquires dynamic position information of the ship as high-precision longitude and latitude coordinate information:
Gtime+i*Δt＝(ptime+i*Δt,qtime+i*Δt)i∈[0,N]
wherein time is the collection start time, delta t is the collection interval, N is the number of collection points, Gtime+i*ΔtHigh precision motion picture coordinate information for ith acquisition interval, ptime+i*ΔtCollecting longitude coordinate information of interval for ith, qtime+i*ΔtAnd acquiring alternate latitude coordinate information for the ith.
3. The LSTM model-based ship track data fusion method of claim 1, wherein the grouped radar longitude and latitude coordinate information is normalized to:
Figure FDA0003484528180000021
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure FDA0003484528180000022
acquiring the radar coordinate information of the interval for the ith normalized acquisition;
and step 2, normalizing the AIS coordinate information into:
Figure FDA0003484528180000023
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure FDA0003484528180000024
the normalized AIS coordinate information of the ith acquisition interval is acquired;
and 2, normalizing the grouped high-precision GPS coordinate information into:
Figure FDA0003484528180000025
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure FDA0003484528180000031
and the coordinate information acquired by the high-precision GPS at the normalized ith acquisition interval is acquired.
4. The LSTM model-based ship track data fusion method of claim 1, wherein the normalized radar-collected coordinate information in step 3 is grouped every five time points in time order as:
Figure FDA0003484528180000032
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure FDA0003484528180000033
for the nth coordinate information in the jth group, WLjThe normalized j-th grouping coordinate information is obtained;
the coordinate information collected by the normalized AIS system in the step 3 is grouped into a group according to five time points in the time sequence:
Figure FDA0003484528180000034
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure FDA0003484528180000035
for the nth coordinate information in the jth group, WSjThe normalized j-th grouping coordinate information is obtained;
in step 3, the normalized high-precision GPS coordinate information is grouped into a group of every five time points according to the time sequence:
Figure FDA0003484528180000036
wherein time is the collection starting time, delta t is the collection interval, N is the collection point number,
Figure FDA0003484528180000037
for the nth coordinate information in the jth group, WGjIs normalized j-th grouping coordinate information.
CN201810537118.6A 2018-05-30 2018-05-30 Ship track data fusion method based on LSTM model Active CN109001722B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810537118.6A CN109001722B (en) 2018-05-30 2018-05-30 Ship track data fusion method based on LSTM model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810537118.6A CN109001722B (en) 2018-05-30 2018-05-30 Ship track data fusion method based on LSTM model

Publications (2)

Publication Number Publication Date
CN109001722A CN109001722A (en) 2018-12-14
CN109001722B true CN109001722B (en) 2022-03-15

Family

ID=64573237

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810537118.6A Active CN109001722B (en) 2018-05-30 2018-05-30 Ship track data fusion method based on LSTM model

Country Status (1)

Country Link
CN (1) CN109001722B (en)

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109711022B (en) * 2018-12-17 2022-11-18 哈尔滨工程大学 Submarine anti-sinking system based on deep learning
CN109726355B (en) * 2019-01-04 2022-11-25 重庆邮电大学 Ship track repairing method based on vector interpolation
CN109740742A (en) * 2019-01-14 2019-05-10 哈尔滨工程大学 A kind of method for tracking target based on LSTM neural network
CN110174690A (en) * 2019-05-30 2019-08-27 杭州中科微电子有限公司 A kind of satellite positioning method based on shot and long term memory network auxiliary
CN110290466B (en) * 2019-06-14 2021-07-09 中国移动通信集团黑龙江有限公司 Floor distinguishing method, device, equipment and computer storage medium
CN110519693B (en) * 2019-09-29 2020-11-03 东北大学 Fusion positioning method for intelligent mobile terminal
CN110990504B (en) * 2019-11-14 2023-01-17 中国船舶重工集团公司第七0七研究所 Ship track compression method based on course and speed change rate
CN110864440B (en) * 2019-11-20 2020-10-30 珠海格力电器股份有限公司 Air supply method, air supply device and air conditioner
CN111257914A (en) * 2020-01-14 2020-06-09 杭州电子科技大学 Marine fishing boat track prediction method and system based on Beidou and AIS data fusion
CN111563072A (en) * 2020-04-15 2020-08-21 交通运输部水运科学研究所 AIS information-based ship real-time accurate position acquisition method
CN112364119B (en) * 2020-12-01 2023-04-18 国家海洋信息中心 Ocean buoy trajectory prediction method based on LSTM coding and decoding model
CN112268564B (en) * 2020-12-25 2021-03-02 中国人民解放军国防科技大学 Unmanned aerial vehicle landing space position and attitude end-to-end estimation method
CN113516321B (en) * 2021-09-14 2021-11-16 中国电子科技集团公司第十五研究所 Model training method and device for predicting ship track
CN115346399B (en) * 2022-07-23 2024-01-19 交通运输部规划研究院 Bridge ship collision prevention early warning system based on phased array radar, AIS and LSTM network

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103353756A (en) * 2013-05-27 2013-10-16 武汉理工大学 Method for monitoring underway ship in real time based on AIS and VTS information integration
CN105244020A (en) * 2015-09-24 2016-01-13 百度在线网络技术（北京）有限公司 Prosodic hierarchy model training method, text-to-speech method and text-to-speech device
US9760806B1 (en) * 2016-05-11 2017-09-12 TCL Research America Inc. Method and system for vision-centric deep-learning-based road situation analysis
CN108022012A (en) * 2017-12-01 2018-05-11 兰州大学 Vehicle location Forecasting Methodology based on deep learning

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103353756A (en) * 2013-05-27 2013-10-16 武汉理工大学 Method for monitoring underway ship in real time based on AIS and VTS information integration
CN105244020A (en) * 2015-09-24 2016-01-13 百度在线网络技术（北京）有限公司 Prosodic hierarchy model training method, text-to-speech method and text-to-speech device
US9760806B1 (en) * 2016-05-11 2017-09-12 TCL Research America Inc. Method and system for vision-centric deep-learning-based road situation analysis
CN107368890A (en) * 2016-05-11 2017-11-21 Tcl集团股份有限公司 A kind of road condition analyzing method and system based on deep learning centered on vision
CN108022012A (en) * 2017-12-01 2018-05-11 兰州大学 Vehicle location Forecasting Methodology based on deep learning

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
基于LSTM的船舶航迹预测模型;权波等;《计算机科学》;20181115;全文 *
海上目标多源轨迹数据关联综述;鹿强等;《地球信息科学学报》;20180524(第05期);全文 *

Also Published As

Publication number Publication date
CN109001722A (en) 2018-12-14

Similar Documents

Publication Publication Date Title
CN109001722B (en) Ship track data fusion method based on LSTM model
CN109084747B (en) Waterborne traffic panoramic three-dimensional navigation system and method based on universal three-dimensional engine
CN103730031B (en) Inland river bridge district Shipborne navigation is collision avoidance system and collision prevention method initiatively
Stone et al. Search for the wreckage of Air France Flight AF 447
CN112116031B (en) Target fusion method, system, vehicle and storage medium based on road side equipment
CN113325422B (en) Space-based rain radar target positioning and rainfall information three-dimensional processing method and system
KR100807822B1 (en) Electric vessel traffic service method and system for harbor and narrow channel
CN109801217B (en) Full-automatic orthographic image splicing method based on GPS ground control point
CN111536962A (en) Intelligent ship route planning method and device, storage medium and computer equipment
CN105116390A (en) Marine radar calibration-oriented measured value and AIS truth value dot pair construction method
Yang et al. An approach to ship behavior prediction based on AIS and RNN optimization model
Wang et al. Remote sensing monitoring method based on BDS-based maritime joint positioning model
Wu et al. A new multi-sensor fusion approach for integrated ship motion perception in inland waterways
CN108917767B (en) Electronic chart-based navigation motion plotting calculation method
Jie et al. A novel estimation algorithm for interpolating ship motion
CN104237862B (en) Probability hypothesis density filter radar system error fusion estimation method based on ADS-B
CN115346399A (en) Bridge ship collision prevention early warning system based on phased array radar, AIS and LSTM network
KR101830544B1 (en) Integrated Navigation Data Management System and Method based on IEC61162-450
Simdiankin et al. A method of vehicle positioning using a non-satellite navigation system
CN115542317A (en) Sea surface target comprehensive identification method and system based on multiple information sources
Pan et al. Research on Ship Arrival Law Based on Route Matching and Deep Learning
Terpsidi et al. Maritime industry revival through systems digitalization
CN116310689B (en) Unmanned intelligent multi-mode information fusion and target perception system and operation method
Roy et al. A deep learning framework for enhancing maritime coastal security
Liu et al. Towards intelligent navigation in future autonomous surface vessels: developments, challenges and strategies

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