CN115409160B - Full sea depth temperature profile inversion method and system based on depth data - Google Patents

Full sea depth temperature profile inversion method and system based on depth data Download PDF

Info

Publication number
CN115409160B
CN115409160B CN202211200508.7A CN202211200508A CN115409160B CN 115409160 B CN115409160 B CN 115409160B CN 202211200508 A CN202211200508 A CN 202211200508A CN 115409160 B CN115409160 B CN 115409160B
Authority
CN
China
Prior art keywords
depth
temperature
temperature profile
order
eof
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
CN202211200508.7A
Other languages
Chinese (zh)
Other versions
CN115409160A (en
Inventor
秦继兴
李倩倩
严娴
顾怡鸣
王海斌
王梦圆
吴双林
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.)
Institute of Acoustics CAS
Original Assignee
Institute of Acoustics CAS
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 Institute of Acoustics CAS filed Critical Institute of Acoustics CAS
Priority to CN202211200508.7A priority Critical patent/CN115409160B/en
Priority to PCT/CN2022/125403 priority patent/WO2024065888A1/en
Publication of CN115409160A publication Critical patent/CN115409160A/en
Application granted granted Critical
Publication of CN115409160B publication Critical patent/CN115409160B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • 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
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Testing Or Calibration Of Command Recording Devices (AREA)

Abstract

The invention discloses a full sea depth temperature profile inversion method and system based on depth setting data, wherein the method comprises the following steps: acquiring ocean temperatures at two fixed depths in real time; inputting the acquired temperature value into a pre-established and trained neural network model to obtain a p-order EOF coefficient; and extracting an average temperature profile and a first p-order EOF basis function from the historical hydrologic data, and calculating the full sea depth temperature profile by using p-order EOF coefficients. The method does not need to arrange a depth vertical temperature chain, has low system complexity, is easy to arrange and operate, and can be applied to a larger area; the AUV is used for acquiring the deep sea temperature, has good maneuverability, can be deployed according to task requirements, and realizes the modeling of the three-dimensional temperature field in a large area.

Description

Full sea depth temperature profile inversion method and system based on depth data
Technical Field
The invention belongs to the technical fields of underwater sound engineering, ocean engineering and sonar, and particularly relates to a full sea depth temperature profile inversion method and system based on depth data.
Background
The most important acoustic parameter in seawater is the propagation velocity of sound waves, and the vertical distribution of the sound velocity profile of seawater is one of the main factors affecting the characteristics of the underwater sound field, which determines the refraction and propagation path of sound waves in the ocean. Seawater temperature is the factor that affects the speed of sound the greatest, and therefore it is important to acquire full-depth temperature profiles in near real time.
Generally, temperature data is often sampled on site through CTD, XBT or an anchor temperature chain, but the method is time-consuming and labor-consuming, and cannot synchronously acquire large-area data, so scientists prefer to adopt an inversion method. The satellite remote sensing means can acquire sea surface temperature data in a large area or even in the whole world in real time, and has higher spatial resolution. The regression statistical analysis method can construct an empirical regression model of sea surface temperature abnormality, sea surface height abnormality and full sea depth temperature profile, and further inversion is carried out to obtain a three-dimensional ocean temperature field, but the accuracy is lower near the thermocline. The acoustic velocity profile can be estimated by using the artificial neural network method by using sea surface parameters, however, the information obtained by remote sensing satellite data only stays on the sea surface layer or near the surface layer, the inversion accuracy is lower by using the average steady-state field of the surface layer information inversion and the temperature profile, see reference [1] ("Estimation of Sound Speed Profiles Using Artificial Neural Networks",2006 in IEEE Geoscience & Remote Sensing Letters, 3 rd phase, initial page number is 467). Marine acoustic tomography is also a common inversion method, where acoustic methods can be used to monitor the mesoscale process of the ocean, however the high power consumption of the transmitter-receiver has been the limiting factor for active acoustic tomography. Researchers reconstruct sound velocity profiles in the full sea depth range using historical hydrographic data in combination with temperatures measured directly in the limited depth range (20 m-40 m), however, the problem of requiring the minimum depth range of the measured data for reconstructing the full sea depth sound velocity profile has not been studied, see reference [2] ("reconstructing the full sea depth sound velocity profile using limited depth sound velocity data", published in acoustic technology, 5 th phase, initial page number 106, 2008).
With the continuous development of the technological level, autonomous underwater robots (Autonomous Underwater Vehicle, AUV) are increasingly being applied to marine feature observations, and the rapid development of AUVs makes it possible to directly measure the hydrological parameters at any depth. The AUV can be provided with a temperature sensor, and temperature data at a designated depth can be obtained by adjusting the head angle and the pitch angle after the buoyancy is basically balanced.
Disclosure of Invention
The invention aims to overcome the defect that the three-dimensional ocean temperature field obtained by inversion in the prior art is low in precision near a thermocline.
In order to achieve the above purpose, the invention provides a full sea depth temperature profile inversion method based on depth setting data, which comprises the following steps:
step S1: acquiring ocean temperatures at two fixed depths in real time;
step S2: inputting the acquired temperature value into a pre-established and trained neural network model to obtain a p-order EOF coefficient;
step S3: and extracting an average temperature profile and a first p-order EOF basis function from the historical hydrologic data, and calculating the full sea depth temperature profile by using p-order EOF coefficients.
As an improvement of the method, the neural network model adopts a BP neural network, and training of the neural network is completed through temperature at a fixed depth and historical hydrologic data of the previous p-order EOF coefficient.
The training process of the BP neural network is divided into two stages of forward propagation and backward propagation;
firstly, forward propagation, namely 2 pieces of depth-setting temperature data are input, are input through an input layer, are processed layer by layer through an hidden layer and then reach an output layer, and a network calculation result is obtained; according to the loss function loss= |y out -y| calculate the error between the network calculation and the actual output, where y out The network output is realized, and y is the actual output; if the loss error exceeds the set threshold, entering a counter-propagation stage;
the back propagation is to reversely transfer the loss error to the input layer through the hidden layer, and to evenly distribute the error to all units of each layer, and to correct the connection weight of each unit according to the gradient descent method;
and (3) forward and backward propagation is circularly carried out, so that the connection weight is continuously adjusted until the loss error is within the threshold value range, and model training is completed.
As an improvement of the above method, the step S1 specifically includes: according to the corresponding depths of two extreme points of the p-th order EOF basis function extracted from the historical hydrologic data, two AUVs are distributed at two fixed depths of a designated sea area, temperature sensors are carried on the AUVs, and after buoyancy is basically balanced, temperature data at the two fixed depths are obtained in real time by adjusting a head angle and a pitch angle.
As an improvement of the above method, the step S3 specifically includes:
obtaining sampling values of temperature profiles at M time points in the historical water level data, wherein each temperature profile is subjected to layering treatment and has N values in depth, and M temperature profile samples are expressed in a matrix form:
Figure BDA0003872269360000021
wherein ,tM (z N ) Representing the value of the temperature profile at the Nth depth of the Mth time point, calculating the average temperature of the M temperature profiles at each layer to obtain the average temperature profile
Figure BDA0003872269360000031
Figure BDA0003872269360000032
wherein ,[]T The representation transposes the vector; subtracting the average temperature profile from the temperature matrix T
Figure BDA0003872269360000033
Obtaining temperature disturbance:
Figure BDA0003872269360000034
singular value decomposition is carried out on the delta T to obtain:
ΔT T ＝UΣV T
wherein v= [ V ] 1 ,…v N ]∈R N×N Is a matrix DeltaDeltaT T ∈R N×N Is the experience to be extractedAn orthogonal function; u= [ U ] 1 ,…u M ]∈R M×M Is a matrix DeltaT T ΔT∈R M×M Sigma=diag ([ λ) 1 …λ N ])∈R M×N Representing a characteristic value matrix; the feature value corresponding to each feature vector represents the weight of the feature vector; the cumulative variance contribution rate of the first m-order modal functions is expressed as:
Figure BDA0003872269360000035
wherein ,
Figure BDA0003872269360000036
λ k representing a kth eigenvalue;
by contribution rate E m Reconstructing a temperature profile by using the top p-order EOF basis function of which the temperature is greater than 95% to obtain a full sea depth temperature profile matrix:
Figure BDA0003872269360000037
wherein ,[α1M α 2M … α pM ] T Representing the coefficient matrix U sigma= [ alpha ] 1 … α N ]∈R M×N P-th order of (2);
Figure BDA0003872269360000038
representation matrix v= [ V 1 ,…,v N ]∈R N×N P-th order of (2);
wherein, the i-th depth temperature value in the temperature profile is:
Figure BDA0003872269360000039
wherein i is N; alpha p Represents the p-th order EOF coefficient, v p Representing the p-th order EOF basis function,
Figure BDA00038722693600000310
represents the i-th depth average temperature of the historical water level data.
As a modification of the above process, said p is preferably 2.
The invention also provides a full sea depth temperature profile inversion system based on the depth setting temperature data, which comprises:
the temperature acquisition module is used for acquiring the ocean temperatures at two fixed depths in real time;
the EOF coefficient calculating module inputs the acquired temperature value into a pre-established and trained neural network model to obtain a p-order EOF coefficient;
and the inversion full sea depth temperature module is used for extracting an average temperature profile and a p-order EOF basis function from historical hydrologic data, and calculating the full sea depth temperature profile by using the p-order EOF coefficient.
As an improvement of the above system, the system further comprises:
and the neural network training module is used for completing training of the neural network through the temperature at the fixed depth and the historical hydrologic data of the previous p-order EOF coefficient.
The neural network adopts a BP neural network, and the training process is divided into two stages of forward propagation and backward propagation;
firstly, forward propagation, namely 2 pieces of depth-setting temperature data are input, are input through an input layer, are processed layer by layer through an hidden layer and then reach an output layer, and a network calculation result is obtained; according to the loss function loss= |y out -y| calculate the error between the network calculation and the actual output, where y out The network output is realized, and y is the actual output; if the loss error exceeds the set threshold, entering a counter-propagation stage;
the back propagation is to reversely transfer the loss error to the input layer through the hidden layer, and to evenly distribute the error to all units of each layer, and to correct the connection weight of each unit according to the gradient descent method;
and (3) forward and backward propagation is circularly carried out, so that the connection weight is continuously adjusted until the loss error is within the threshold value range, and model training is completed.
The invention also provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing a method as claimed in any one of the preceding claims when executing the computer program.
The invention also provides a computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform a method as claimed in any one of the preceding claims.
Compared with the prior art, the invention has the advantages that:
according to the method, the full sea depth temperature profile can be inverted in real time only by acquiring limited 1-2 fixed-depth temperature data by using AUV carrying temperature sensor data. Compared with the traditional observation modes such as fixed-point observation, cruising observation, anchorage buoy array observation and the like, the method can realize mobile observation, and has the advantages of high autonomy, strong load capacity, good maneuverability and high intelligent degree, and meets the accuracy requirement of measured data while meeting the economical efficiency.
Drawings
FIG. 1 is a flow chart of neural network model training and inversion based on a method for inverting full sea depth temperature profile with fixed depth data;
FIG. 2 is a schematic diagram showing the error of water temperature data at different depths for inverting a full sea depth temperature profile;
FIG. 3 is a graph of cumulative variance contribution of the top 15 th order EOF;
FIG. 4 is a graph showing the relationship between the first two-order EOF coefficients and the 24℃isotherms and the temperature gradients; FIG. 4 (a) shows normalized magnitudes of the first-order EOF coefficients and 24℃isotherms; FIG. 4 (b) shows the normalized magnitude of the second order EOF coefficient and temperature gradient;
FIG. 5 shows a scatter plot and a fitted line plot of the first two-order EOF coefficients with 24℃isotherms and temperature gradients; FIG. 5 (a) shows a scatter plot of the first order EOF coefficients versus 24℃isotherms and a fitted line; FIG. 5 (b) shows a scatter plot of second order EOF coefficients versus temperature gradient and a straight line fit;
FIG. 6 shows the value in z 2 、z 3 Inversion of water temperature information at depth to full sea deep temperatureError diagram of degree profile;
FIG. 7 is a schematic diagram of the AUV fixed coordinate system and motion coordinate system;
FIG. 8 shows a BP neural network topology;
FIG. 9 is a flow chart of a method for inverting a full sea depth temperature profile based on depth data.
Detailed Description
According to the invention, the temperature at 1-2 fixed depths is extracted by using AUV investigation data, and a model for inverting the full sea depth temperature profile by using the depth-fixed temperature data is established based on the back propagation neural network, so that the problems that the traditional observation platform cannot acquire temperature data in real time and rapidly and is high in cost can be solved.
The technical scheme of the invention is described in detail below with reference to the accompanying drawings.
The invention provides a full sea depth temperature profile inversion method based on depth setting data, which is particularly suitable for the situation that internal waves exist. The prior researches show that under the internal wave environment, the first two-order EOF mode can reconstruct any temperature profile more accurately, and the information at the depth of the extreme point of the first two-order EOF base function can reflect thermocline information to the greatest extent. On the basis, a method for inverting the full sea depth temperature profile by using limited depth setting temperature data is provided. Firstly, establishing a mathematical model between depth setting temperature data and the first two-order EOF coefficients through a BP neural network, and then realizing inversion of the temperature profile by combining an average temperature profile extracted from historical hydrologic data and an EOF basis function, wherein a flow chart is shown in figure 1.
Step 1: and analyzing the influence of the depth setting temperature data selection on the temperature profile inversion result.
And using measured temperatures at all depths in the historical hydrological data to replace water temperature values at any depth measured by autonomous underwater vehicles such as AUV and the like, and analyzing the effects of inversion temperature profiles of water temperatures at different depths. In order to determine the optimal selection criteria for the fixed depth, the invention uses the measurement data of the temperature chain to select from shallow to deep at equal intervals (1 meter) one by one. I.e. training the BP neural network with temperatures at a fixed depth (from 10m to 87m, with 1m intervals). Fig. 2 plot with plotThe normalized average root mean square error of the temperature profile is inverted by the water temperature at each depth, the solid curve is used for drawing a normalized first-order EOF base function, the dotted curve is used for drawing the absolute value of a normalized second-order EOF base function, and the horizontal axis scales corresponding to the three dotted lines respectively represent z 1 、z 2 、z 3 Depth, where z 1 For the depth corresponding to the extreme point of the first-order EOF base function, z 2 and z3 The corresponding depths at the two extreme points of the second-order EOF basis function are respectively. It can be seen in the figure that the inversion accuracy approximately appears "smile" as a function of depth, i.e. in z 2 The inversion error is greatly reduced along with the water depth by shallow depth; in z 3 The inversion error greatly rises with the water depth; and at z 2 -z 3 In the depth range, the inversion error is small and remains substantially unchanged. As can be seen, the temperature change at a discrete depth can reflect the overall morphology of the full sea depth temperature profile to some extent, however, the water temperature at different depths contains different amounts of information, where z 2 -z 3 The water temperature in the depth range contains the most information.
Step 2: and (3) carrying out EOF decomposition on the historical hydrologic data, and analyzing the physical significance of the EOF base functions and the EOF coefficients of the first two orders.
If the sea water temperature profile is estimated directly, the variables to be estimated are too many, the operation amount is quite large, and the performance requirement on the algorithm is quite high, so that the sea water temperature profile is expressed by adopting an empirical orthogonal function. The empirical orthogonal function is a feature vector extracted from a certain number of sample data, and researches show that the first few orders of orthogonal functions can effectively reconstruct a temperature profile. The temperature profiles are sampled at M time points, each temperature profile having N values in depth through layering, the M temperature profile samples being represented in the form of a matrix:
Figure BDA0003872269360000061
wherein ,ti (z j ) Representing the value of the temperature profile at the j-th depth of the i-th sampling point, and calculating M temperaturesThe average temperature of each layer is obtained by the temperature profile
Figure BDA0003872269360000062
Figure BDA0003872269360000063
Where the symbol T denotes the transposition of the vector. Subtracting the average temperature profile from the temperature matrix T
Figure BDA0003872269360000065
Obtaining temperature disturbance:
Figure BDA0003872269360000064
singular value decomposition of Δt can be performed to obtain:
ΔT T ＝UΣV T (4)
wherein V= [ V ] 1 ,…v N ]∈R N×N Is matrix a=ΔtΔt T ∈R N×N Is the empirical orthogonal function to be extracted. U= [ U ] 1 ,…u M ]∈R M×M Is matrix b=Δt T ΔT∈R M×M Is defined, and Σ=diag ([ λ) 1 … λ N ])∈R M×N Is a characteristic value matrix. X=u Σ e R M×N Is a coefficient matrix.
Figure BDA0003872269360000071
And Ω= [ λ ] 1 ,…λ N ]Is the eigenvalue of matrix a. The feature value corresponding to each feature vector represents the weight of the feature vector, and the smaller the feature value is, the less information the corresponding feature vector (empirical orthogonal function) contains. The cumulative variance contribution rate of the first m-th order modal function can be expressed as: />
Figure BDA0003872269360000072
The contribution rate of the first-order mode is the largest, and the higher the order is, the smaller the contribution rate of the mode function is.
By contribution rate E m More than 95% of the first p-th order EOF basis functions reconstruct a temperature profile, the reconstructed temperature profile matrix being expressed as:
Figure BDA0003872269360000073
wherein ,[α1M α 2M … α pM ] T Is the coefficient matrix U sigma = [ alpha ] 1 … α N ]∈R M×N P-th order of (2);
Figure BDA0003872269360000074
representation matrix v= [ V 1 ,…,v N ]∈R N×N Is the first p-th order of (c).
The accumulated variance contribution rate of the 15 th-order EOF basis functions is calculated by performing EOF decomposition on the historical hydrologic data, and as shown in figure 3, the accumulated variance contribution rate of the two-order EOF modes in front of the south sea area can reach more than 95%.
Thus, the temperature profile is reconstructed using the first two-order EOF basis functions, and the ith temperature profile can be approximated as:
Figure BDA0003872269360000075
wherein ,αi Represents the i-th order EOF coefficient, v i Represents the ith order EOF basis function, t mean Represents the i-th sampling time average ocean temperature of the historical water level data.
And (3) analyzing the physical significance of the EOF base functions and the EOF coefficients of the first two orders according to the conclusion obtained in the step (1). There have been studies giving a physical explanation of the EOF coefficient, the first order EOF coefficient representing the vertical displacement of the thermocline, α 1 The larger the thermocline, the shallower the thermocline. While the second order EOF coefficient represents the change in temperature gradient, and α 2 The larger the thermoclineThe more severe the variation, the more this is verified below. As can be seen from FIG. 2, the depth corresponding to the extreme point of the first-order EOF basis function is z 1 =55m, at this depth, the average temperature is
Figure BDA0003872269360000076
Fig. 4 (a) depicts a normalized 24 ℃ isotherm of a temperature profile in a training set with a normalized first order EOF coefficient, where the isotherm is highly correlated with the trend of the first order projection coefficient, and the correlation coefficient is about 0.98. FIG. 5 (a) is a scattergram and a fitting straight line of the corresponding depth of 24℃isotherms and the first-order EOF coefficient>
Figure BDA0003872269360000077
z 24℃ Is the depth corresponding to the isotherm at 24 ℃. The first-order EOF coefficient is similar to the trend of the temperature of the seawater layer, so it can be also explained that the first-order EOF coefficient can approximately reflect the trend of the seawater layer. As can be seen from FIG. 2, the depth corresponding to the two extreme points of the second-order EOF base function is z 2 ＝49m,z 3 =62 m, the temperature gradient of the training set temperature profile in this depth range is calculated. Fig. 4 (b) shows a normalized temperature gradient and a normalized second-order EOF coefficient, and as can be seen from the graph, the temperature gradient and the trend of the second-order projection coefficient are highly correlated, the correlation coefficient reaches 0.90, and fig. 5 (b) shows a scatter diagram and a fitting straight line of the temperature gradient and the second-order EOF coefficient:
Figure BDA0003872269360000081
the above conclusion is consistent with the existing study, z 2 and z3 Respectively represent the upper and lower boundaries of the thermocline, thus z 2 -z 3 The water temperature data in the depth range can reflect thermocline information to the greatest extent, so that the accuracy is higher when the temperature profile is inverted. This also explains the conclusion reached in step 1. The analysis provides a reference basis for the measurement depth selection of the depth setting data.
Step 3: in z 2 、z 3 The water temperature information at the depth replaces the constant-depth temperature data acquired by the AUV to invert the full-sea-depth temperature profile.
It is conceivable that as the information of the input layer increases, the inversion accuracy of the temperature profile in the training set increases, and the result in fig. 2 shows that the inversion accuracy depends on the measured depth of the depth-determining data, while the temperature profile mainly depends on the characteristics of the thermocline, and step 2 proves that z 2 、z 3 The depth of the upper boundary and the lower boundary of the thermocline are respectively represented, the water temperatures at the two depths can basically determine the basic structure of the temperature profile, the water temperature information at the two depths is used for replacing the constant-depth temperature data acquired by the AUV to invert the full-sea-depth temperature profile, the inversion result is shown in figure 6, the root mean square error of the inversion of the temperature profile is basically below 0.2 ℃, and the mean value of the root mean square error is 0.1137 ℃.
Step 4: in combination with analysis of historical hydrologic data, two AUVs are utilized to measure the depth-setting temperature data of the upper and lower boundaries of the thermocline.
For autonomous underwater robots, fixed depth navigation is a common navigational state. The two AUVs are required to carry temperature sensors to cooperatively operate, and the temperature data is acquired in the designated sea area by constant-depth navigation. After the buoyancy is basically balanced, when the AUV has a certain pitching speed, the constant-depth sailing is realized by adjusting the head rocking angle and the pitching rocking angle of the AUV. In a multiple AUV system, a fixed coordinate system needs to be established to describe the change in position of each AUV. The earth is generally used as a reference system to select a fixed coordinate system E-zeta eta zeta, the origin E of the coordinate system can be any point on the ground or the sea, the zeta axis is kept horizontal, and the main heading of AUV is always the forward direction of Ezeta; eeta and Eζ axes are perpendicular to each other and in a horizontal plane, optionally with the direction of Eζ axis perpendicular to the Eζ coordinate plane, which is directed forward toward the earth center. In order to describe the mutual distance relationship between the two moving AUVs conveniently, a moving coordinate system needs to be established, and the moving coordinate system can be displaced along with the movement of the AUVs. In the motion coordinate system, the Ox axis is taken on a straight line formed by two measurement AUV position points. The Oz axis is oriented in the bottom direction in the longitudinal mid-plane perpendicular to the water plane as shown in fig. 7. In order to clearly observe the real position and navigation law of the underwater AUV, the motion coordinate system needs to be converted into a solid oneAnd the fixed coordinate system, and the related experimental data extracted under the fixed coordinate system more accurately and intuitively reflect the movement process of the underwater AUV. Let the origin of the motion coordinate system have a coordinate value (O) x ,O y ,O z ) The rotation angle of the coordinate axis is
Figure BDA0003872269360000091
The AUV has a coordinate (X 1 ,Y 1 ,Z 1 ) The coordinates in the fixed coordinate system are (X 2 ,Y 2 ,Z 2 ) Any point in the motion coordinate system can be represented in the fixed coordinate system as:
Figure BDA0003872269360000092
step 5: and inverting the full sea depth temperature profile in real time according to the two depth setting temperature data obtained by AUV measurement.
Establishing z through BP neural network with historical hydrologic data 2 、z 3 The network model between the temperature at the depth and the EOF coefficients of the first two orders is shown in fig. 8, after the AUV actually measures and obtains the water temperatures of the two depths, the water temperatures are input into the trained network model, and the temperature profile of the full sea depth is obtained through real-time inversion by combining the historical average temperature profile and the EOF basis functions of the first two orders.
The training process of the BP neural network is divided into two stages of forward propagation and backward propagation. Firstly, forward transmission, namely, input data (2 pieces of depth temperature data) are transmitted in through an input layer, are processed layer by layer through an hidden layer, then reach an output layer, and a network calculation result is obtained. According to the loss function loss= |y out -y| can calculate the error between the network calculation and the actual output, where y out For network output, y is the actual output. If the loss error exceeds the set threshold, the backward propagation phase is entered. The back propagation is to transfer the lost error back to the input layer through the hidden layer, and to distribute the error to all the units of each layer, and to correct the connection weight of each unit according to the gradient descent method. Forward and reverse propagation cycles are performed such thatThe connection weights can be continuously adjusted until the loss error is within a threshold range, indicating that model training is complete.
According to the above test procedure, the invention provides a full sea depth temperature profile inversion method (as shown in fig. 9) based on depth setting data, which comprises the following steps:
step S1: acquiring ocean temperatures at two fixed depths in real time;
according to the corresponding depths of two extreme points of the p-th order EOF basis function extracted from the historical hydrologic data, two AUVs are distributed at two fixed depths of a designated sea area, temperature sensors are carried on the AUVs, and after buoyancy is basically balanced, temperature data at the two fixed depths are obtained in real time by adjusting a head angle and a pitch angle.
Step S2: inputting the acquired temperature value into a pre-established and trained neural network model to obtain a p-order EOF coefficient;
step S3: extracting an average temperature profile and a first p-order EOF basis function from historical hydrologic data, and calculating a full sea depth temperature profile by using p-order EOF coefficients:
obtaining sampling values of temperature profiles at M time points in the historical water level data, wherein each temperature profile is subjected to layering treatment and has N values in depth, and M temperature profile samples are expressed in a matrix form:
Figure BDA0003872269360000101
wherein ,tM (z N ) Representing the value of the temperature profile at the Nth depth of the Mth time point, calculating the average temperature of the M temperature profiles at each layer to obtain the average temperature profile
Figure BDA0003872269360000102
Figure BDA0003872269360000103
wherein ,[]T The representation transposes the vectorThe method comprises the steps of carrying out a first treatment on the surface of the Subtracting the average temperature profile from the temperature matrix T
Figure BDA0003872269360000104
Obtaining temperature disturbance:
Figure BDA0003872269360000105
singular value decomposition is carried out on the delta T to obtain:
ΔT T ＝UΣV T
wherein v= [ V ] 1 ,…v N ]∈R N×N Is a matrix DeltaDeltaT T ∈R N×N The feature vector of (2) is the empirical orthogonal function to be extracted; u= [ U ] 1 ,…u M ]∈R M×M Is a matrix DeltaT T ΔT∈R M×M Sigma=diag ([ λ) 1 … λ N ])∈R M×N Representing a characteristic value matrix; the feature value corresponding to each feature vector represents the weight of the feature vector; the cumulative variance contribution rate of the first m-order modal functions is expressed as:
Figure BDA0003872269360000106
wherein ,
Figure BDA0003872269360000107
λ k representing a kth eigenvalue;
by contribution rate E m Reconstructing a temperature profile by using the first p-order EOF basis functions larger than a set value to obtain a full sea depth temperature profile matrix:
Figure BDA0003872269360000108
/>
wherein ,[α1M α 2M … α pM ] T Representing the coefficient matrix U sigma= [ alpha ] 1 … α N ]∈R M×N P-th order of (2);
Figure BDA0003872269360000111
representation matrix v= [ V 1 ,…,v N ]∈R N×N P-th order of (2);
wherein, the i-th depth temperature value in the temperature profile is:
Figure BDA0003872269360000112
wherein i is N; alpha p Represents the p-th order EOF coefficient, v p Representing the p-th order EOF basis function,
Figure BDA0003872269360000113
represents the i-th depth average temperature of the historical water level data. p is preferably 2.
The neural network model adopts a BP neural network, and training of the neural network is completed through temperature at a fixed depth and historical hydrologic data of the previous p-order EOF coefficient.
The invention also provides a full sea depth temperature profile inversion system based on the depth setting temperature data, which comprises:
the temperature acquisition module is used for acquiring the ocean temperatures at two fixed depths in real time;
the EOF coefficient calculating module inputs the acquired temperature value into a pre-established and trained neural network model to obtain a p-order EOF coefficient;
and the inversion full sea depth temperature module is used for extracting an average temperature profile and a p-order EOF basis function from historical hydrologic data, and calculating the full sea depth temperature profile by using the p-order EOF coefficient.
The system further comprises:
and the neural network training module is used for completing training of the neural network through the temperature at the fixed depth and the historical hydrologic data of the previous p-order EOF coefficient.
The present invention may also provide a computer apparatus comprising: at least one processor, memory, at least one network interface, and a user interface. The various components in the device are coupled together by a bus system. It will be appreciated that a bus system is used to enable connected communications between these components. The bus system includes a power bus, a control bus, and a status signal bus in addition to the data bus.
The user interface may include, among other things, a display, a keyboard, or a pointing device (e.g., a mouse, track ball, touch pad, or touch screen, etc.).
It is to be understood that the memory in the embodiments disclosed herein may be either volatile memory or nonvolatile memory, or may include both volatile and nonvolatile memory. The nonvolatile Memory may be a Read-Only Memory (ROM), a Programmable ROM (PROM), an Erasable PROM (EPROM), an Electrically Erasable EPROM (EEPROM), or a flash Memory. The volatile memory may be random access memory (Random Access Memory, RAM) which acts as an external cache. By way of example, and not limitation, many forms of RAM are available, such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (Double Data Rate SDRAM), enhanced SDRAM (ESDRAM), synchronous DRAM (SLDRAM), and Direct RAM (DRRAM). The memory described herein is intended to comprise, without being limited to, these and any other suitable types of memory.
In some implementations, the memory stores the following elements, executable modules or data structures, or a subset thereof, or an extended set thereof: an operating system and application programs.
The operating system includes various system programs, such as a framework layer, a core library layer, a driving layer, and the like, and is used for realizing various basic services and processing hardware-based tasks. Applications, including various applications such as Media Player (Media Player), browser (Browser), etc., are used to implement various application services. The program implementing the method of the embodiment of the present disclosure may be contained in an application program.
In the above embodiment, the processor may be further configured to call a program or an instruction stored in the memory, specifically, may be a program or an instruction stored in an application program:
the steps of the above method are performed.
The method described above may be applied in a processor or implemented by a processor. The processor may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in a processor or by instructions in the form of software. The processor may be a general purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a field programmable gate array (Field Programmable Gate Array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. The methods, steps and logic blocks disclosed above may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method as disclosed above may be embodied directly in hardware for execution by a decoding processor, or in a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory, and the processor reads the information in the memory and, in combination with its hardware, performs the steps of the above method.
It is to be understood that the embodiments described herein may be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For a hardware implementation, the processing units may be implemented within one or more application specific integrated circuits (Application Specific Integrated Circuits, ASIC), digital signal processors (Digital Signal Processing, DSP), digital signal processing devices (DSP devices, DSPD), programmable logic devices (Programmable Logic Device, PLD), field programmable gate arrays (Field-Programmable Gate Array, FPGA), general purpose processors, controllers, microcontrollers, microprocessors, other electronic units configured to perform the functions described herein, or a combination thereof.
For a software implementation, the inventive techniques may be implemented with functional modules (e.g., procedures, functions, and so on) that perform the inventive functions. The software codes may be stored in a memory and executed by a processor. The memory may be implemented within the processor or external to the processor.
The present invention may also provide a non-volatile storage medium for storing a computer program. The steps of the above-described method embodiments may be implemented when the computer program is executed by a processor.
Finally, it should be noted that the above embodiments are only for illustrating the technical solution of the present invention and are not limiting. Although the present invention has been described in detail with reference to the embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the present invention, which is intended to be covered by the appended claims.

Claims (8)

1. A full sea depth temperature profile inversion method based on depth data, the method comprising:
step S1: acquiring ocean temperatures at two fixed depths in real time;
step S2: inputting the acquired temperature value into a pre-established and trained neural network model to obtain a p-order EOF coefficient;
step S3: extracting an average temperature profile and a first p-order EOF basis function from historical hydrologic data, and calculating a full sea depth temperature profile by using p-order EOF coefficients;
the step S3 specifically comprises the following steps:
obtaining sampling values of temperature profiles at M time points in the historical water level data, wherein each temperature profile is subjected to layering treatment and has N values in depth, and M temperature profile samples are expressed in a matrix form:
Figure FDA0004141332890000011
wherein ,tM (z N ) Representing the value of the temperature profile at the Nth depth of the Mth time point, calculating the average temperature of the M temperature profiles at each layer to obtain the average temperature profile
Figure FDA0004141332890000016
Figure FDA0004141332890000012
wherein ,[]T The representation transposes the vector; subtracting the average temperature profile from the temperature matrix T
Figure FDA0004141332890000013
Obtaining temperature disturbance:
Figure FDA0004141332890000014
singular value decomposition is carried out on the delta T to obtain:
ΔT T ＝UΣV T
wherein v= [ V ] 1 ,…v N ]∈R N×N Is a matrix DeltaDeltaT T ∈R N×N The feature vector of (2) is the empirical orthogonal function to be extracted; u= [ U ] 1 ,…u M ]∈R M×M Is a matrix DeltaT T ΔT∈R M×M Sigma=diag ([ λ) 1 …λ N ])∈R M×N Representing a characteristic value matrix; the feature value corresponding to each feature vector represents the weight of the feature vector; the cumulative variance contribution rate of the first m-order modal functions is expressed as:
Figure FDA0004141332890000015
wherein ,
Figure FDA0004141332890000021
λ k representing a kth eigenvalue;
by contribution rate E m Reconstructing a temperature profile by using the first p-order EOF basis functions larger than a set value to obtain a full sea depth temperature profile matrix:
Figure FDA0004141332890000022
wherein ,[α1M α 2M ...α pM ] T Representing the coefficient matrix U sigma= [ alpha ] 1 …α N ]∈R M×N P-th order of (2);
Figure FDA0004141332890000023
representation matrix v= [ V 1 ,…,v N ]∈R N×N P-th order of (2);
wherein, the i-th depth temperature value in the temperature profile is:
Figure FDA0004141332890000024
wherein i is N; alpha p Represents the p-th order EOF coefficient, v p Representing the p-th order EOF basis function,
Figure FDA0004141332890000025
represents the i-th depth average temperature of the historical water level data.
2. The full sea depth temperature profile inversion method based on depth setting data according to claim 1, wherein the neural network model adopts a BP neural network, and training of the neural network is completed through temperature at a fixed depth and historical hydrologic data of a previous p-order EOF coefficient;
the training process of the BP neural network is divided into two stages of forward propagation and backward propagation;
firstly, forward propagation, namely 2 pieces of depth-setting temperature data are input, are input through an input layer, are processed layer by layer through an hidden layer and then reach an output layer, and a network calculation result is obtained; according to the loss function loss= |y out -y| calculate the error between the network calculation and the actual output, where y out The network output is realized, and y is the actual output; if the loss error exceeds the set threshold, entering a counter-propagation stage;
the back propagation is to reversely transfer the loss error to the input layer through the hidden layer, and to evenly distribute the error to all units of each layer, and to correct the connection weight of each unit according to the gradient descent method;
and (3) forward and backward propagation is circularly carried out, so that the connection weight is continuously adjusted until the loss error is within the threshold value range, and model training is completed.
3. The full sea depth temperature profile inversion method based on the depth setting data according to claim 1, wherein the step S1 specifically comprises: according to the corresponding depths of two extreme points of the p-th order EOF basis function extracted from the historical hydrologic data, two AUVs are distributed at two fixed depths of a designated sea area, temperature sensors are carried on the AUVs, and after buoyancy is basically balanced, temperature data at the two fixed depths are obtained in real time by adjusting a head angle and a pitch angle.
4. The full sea depth temperature profile inversion method based on depth setting data according to claim 1, wherein the p value is 2.
5. A full sea depth temperature profile inversion system based on depth data, the system comprising:
the temperature acquisition module is used for acquiring the ocean temperatures at two fixed depths in real time;
the EOF coefficient calculating module inputs the acquired temperature value into a pre-established and trained neural network model to obtain a p-order EOF coefficient;
the inversion full sea depth temperature module is used for extracting an average temperature profile and a p-order EOF basis function from historical hydrologic data, and calculating the full sea depth temperature profile by using p-order EOF coefficients;
the specific calculation method of the inversion full sea depth temperature module comprises the following steps:
obtaining sampling values of temperature profiles at M time points in the historical water level data, wherein each temperature profile is subjected to layering treatment and has N values in depth, and M temperature profile samples are expressed in a matrix form:
Figure FDA0004141332890000031
wherein ,tM (z N ) Representing the value of the temperature profile at the Nth depth of the Mth time point, calculating the average temperature of the M temperature profiles at each layer to obtain the average temperature profile
Figure FDA0004141332890000035
Figure FDA0004141332890000032
wherein ,[]T The representation transposes the vector; subtracting the average temperature profile from the temperature matrix T
Figure FDA0004141332890000033
Obtaining temperature disturbance: />
Figure FDA0004141332890000034
Singular value decomposition is carried out on the delta T to obtain:
ΔT T ＝UΣV T
wherein v= [ V ] 1 ,…v N ]∈R N×N Is a matrix DeltaDeltaT T ∈R N×N The feature vector of (2) is the empirical orthogonal function to be extracted; u= [ U ] 1 ,…u M ]∈R M×M Is a matrix DeltaT T ΔT∈R M×M Sigma=diag ([ λ) 1 ...λ N ])∈R M×N Representing a characteristic value matrix; the feature value corresponding to each feature vector represents the weight of the feature vector; the cumulative variance contribution rate of the first m-order modal functions is expressed as:
Figure FDA0004141332890000041
wherein ,
Figure FDA0004141332890000042
λ k representing a kth eigenvalue;
by contribution rate E m Reconstructing a temperature profile by using the first p-order EOF basis functions larger than a set value to obtain a full sea depth temperature profile matrix:
Figure FDA0004141332890000043
wherein ,[α1M α 2M ...α pM ] T Representing the coefficient matrix U sigma= [ alpha ] 1 ...α N ]∈R M×N P-th order of (2);
Figure FDA0004141332890000044
representation matrix v= [ V 1 ,…,v N ]∈R N×N P-th order of (2);
wherein, the i-th depth temperature value in the temperature profile is:
Figure FDA0004141332890000045
wherein i is N; alpha p Represents the p-th order EOF coefficient, v p Representing the p-th order EOF basis function,
Figure FDA0004141332890000046
represents the i-th depth average temperature of the historical water level data.
6. The full sea depth temperature profile inversion system based on depth setting data of claim 5, further comprising:
the neural network training module is used for completing training of the neural network through the temperature at the fixed depth and the historical hydrologic data of the previous p-order EOF coefficient;
the neural network adopts a BP neural network, and the training process is divided into two stages of forward propagation and backward propagation;
firstly, forward propagation, namely 2 pieces of depth-setting temperature data are input, are input through an input layer, are processed layer by layer through an hidden layer and then reach an output layer, and a network calculation result is obtained; according to the loss function loss= |y out -y| calculate the error between the network calculation and the actual output, where y out The network output is realized, and y is the actual output; if the loss error exceeds the set threshold, entering a counter-propagation stage;
the back propagation is to reversely transfer the loss error to the input layer through the hidden layer, and to evenly distribute the error to all units of each layer, and to correct the connection weight of each unit according to the gradient descent method;
and (3) forward and backward propagation is circularly carried out, so that the connection weight is continuously adjusted until the loss error is within the threshold value range, and model training is completed.
7. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the method of any one of claims 1 to 4 when executing the computer program.
8. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program which, when executed by a processor, causes the processor to perform the method of any of claims 1 to 4.
CN202211200508.7A 2022-09-29 2022-09-29 Full sea depth temperature profile inversion method and system based on depth data Active CN115409160B (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202211200508.7A CN115409160B (en) 2022-09-29 2022-09-29 Full sea depth temperature profile inversion method and system based on depth data
PCT/CN2022/125403 WO2024065888A1 (en) 2022-09-29 2022-10-14 Deep-sea temperature profile inversion method and system based on depth-keeping data

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211200508.7A CN115409160B (en) 2022-09-29 2022-09-29 Full sea depth temperature profile inversion method and system based on depth data

Publications (2)

Publication Number Publication Date
CN115409160A CN115409160A (en) 2022-11-29
CN115409160B true CN115409160B (en) 2023-04-28

Family

ID=84167175

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211200508.7A Active CN115409160B (en) 2022-09-29 2022-09-29 Full sea depth temperature profile inversion method and system based on depth data

Country Status (2)

Country Link
CN (1) CN115409160B (en)
WO (1) WO2024065888A1 (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108981957A (en) * 2018-05-31 2018-12-11 西北工业大学 Submarine temperatures field reconstructing method based on self organizing neural network and Empirical Orthogonal Function
CN110057365A (en) * 2019-05-05 2019-07-26 哈尔滨工程大学 A kind of depth AUV dive localization method latent greatly
CN112598113A (en) * 2020-12-15 2021-04-02 广东海洋大学 Ocean sound velocity profile acquisition method based on self-organizing competitive neural network
CN112596412A (en) * 2020-12-11 2021-04-02 中国科学院沈阳自动化研究所 Multi-AUV simulation platform
CN113191087A (en) * 2021-02-02 2021-07-30 中国人民解放军海军大连舰艇学院 Navigation type depth measurement data profile sound velocity correction method combining historical profile sound velocity and actually measured surface layer sound velocity
CN113218493A (en) * 2021-04-08 2021-08-06 中国人民解放军国防科技大学 Sound velocity profile inversion method based on empirical orthogonal function method
CN114818232A (en) * 2021-01-19 2022-07-29 中国科学院声学研究所 Reconstruction method and system of seawater space temperature profile

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000298069A (en) * 1999-04-14 2000-10-24 Oki Electric Ind Co Ltd Ocean acoustic tomography-data processing and display device
CN111259943A (en) * 2020-01-10 2020-06-09 天津大学 Thermocline prediction method based on machine learning
CN111307266B (en) * 2020-02-21 2021-06-29 山东大学 Sound velocity obtaining method and global ocean sound velocity field construction method based on same

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108981957A (en) * 2018-05-31 2018-12-11 西北工业大学 Submarine temperatures field reconstructing method based on self organizing neural network and Empirical Orthogonal Function
CN110057365A (en) * 2019-05-05 2019-07-26 哈尔滨工程大学 A kind of depth AUV dive localization method latent greatly
CN112596412A (en) * 2020-12-11 2021-04-02 中国科学院沈阳自动化研究所 Multi-AUV simulation platform
CN112598113A (en) * 2020-12-15 2021-04-02 广东海洋大学 Ocean sound velocity profile acquisition method based on self-organizing competitive neural network
CN114818232A (en) * 2021-01-19 2022-07-29 中国科学院声学研究所 Reconstruction method and system of seawater space temperature profile
CN113191087A (en) * 2021-02-02 2021-07-30 中国人民解放军海军大连舰艇学院 Navigation type depth measurement data profile sound velocity correction method combining historical profile sound velocity and actually measured surface layer sound velocity
CN113218493A (en) * 2021-04-08 2021-08-06 中国人民解放军国防科技大学 Sound velocity profile inversion method based on empirical orthogonal function method

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
胡军 ; 肖业伟 ; 张东波 ; 冷龙龙 ; .声速剖面反演预测方法.海洋科学进展.2019,(第02期), *

Also Published As

Publication number Publication date
CN115409160A (en) 2022-11-29
WO2024065888A1 (en) 2024-04-04

Similar Documents

Publication Publication Date Title
CA3067573A1 (en) Target tracking systems and methods for uuv
CN102749622B (en) Multiwave beam-based depth-sounding joint inversion method for sound velocity profile and seafloor topography
CN107589749B (en) Underwater robot autonomous positioning and node map construction method
CN112598113A (en) Ocean sound velocity profile acquisition method based on self-organizing competitive neural network
Lermusiaux et al. Optimal planning and sampling predictions for autonomous and Lagrangian platforms and sensors in the northern Arabian Sea
CN102323586B (en) UUV (unmanned underwater vehicle) aided navigation method based on current profile
CN109579850B (en) Deepwater intelligent navigation method based on auxiliary inertial navigation to water velocity
CN116306318B (en) Three-dimensional ocean thermal salt field forecasting method, system and equipment based on deep learning
CN113486574A (en) Sound velocity profile completion method and device based on historical data and machine learning
CN111076728A (en) DR/USBL-based deep submersible vehicle combined navigation method
Liu et al. Research into the integrated navigation of a deep-sea towed vehicle with USBL/DVL and pressure gauge
CN115859116A (en) Marine environment field reconstruction method based on radial basis function regression interpolation method
CN115409160B (en) Full sea depth temperature profile inversion method and system based on depth data
CN116805028B (en) Wave surface inversion method and system based on floating body motion response
CN110908404B (en) AUV intelligent observation motion method based on data driving
CN117104452A (en) Method and system for inverting waves by ship-following swinging motion based on artificial neural network
CN115469314A (en) Uniform circular array steady underwater target azimuth tracking method and system
Hajizadeh et al. Determination of ship maneuvering hydrodynamic coe cients using system identi cation technique based on free-running model test
Shishkin et al. A multi-model system of intelligent unmanned surface vehicles for environmental monitoring
CN112085779A (en) Wave parameter estimation method and device
Rodiana et al. Software and hardware in the loop simulation of navigation system design based on state observer using Kalman filter for autonomous underwater glider
CN103336525A (en) Convenient UKF (Unscented Kalman Filter) filtering method for high weights of stochastic system
Xian et al. Inversion of Full-depth Temperature Profiles Based on Surface Temperature and AUV Measured Data
Xin et al. Sound Velocity Profiles Time Series Prediction Method Based on EMD-NARX Model
Pan et al. AUV Tightly Coupled Terrain Aided Navigation Strategy Based on Isogonal MBES Modeling Method

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